Search

Motor oil pollutants have a non-negligible impact on crop growth and soil matrix, causing phenomena such as crop yield reduction and even crop failure. In order to solve the problem of predicting the concentration of motor oil pollutants in the soil surface layer, the fluorescence induction technique was used to obtain the spectral curves of motor oil, and the method of predicting the concentration of pollutant oils in the soil surface layer using the wavelet kurtosis as a quantitative parameter was proposed, and a comparative analysis was carried out by combining the three different kinds of motor oils on the market with the random forest regression algorithm. The experimental results show that the concentration prediction results of random forest with the selected wavelet kurtosis parameter for the three kinds of motor oils are evaluated using the correlation coefficient RP and the root mean square deviation (RMSD), and the prediction of gear oil, engine oil, and motorcycle oil are improved by 1. 2%, 2. 2%, and 1. 9%, and 14. 9%, 32. 4%, and 16. 8%, respectively. Among the experimentally prepared samples of three kinds of engine oils, from which 30 groups of samples with concentrations of 0. 01 ~ 0. 3 mL/ g each are selected for model prediction validation, the recognition accuracy of which is improved by 6. 67%, 6. 66%, 9. 96%, respectively. It is also verified that the prediction accuracy of the wavelet kurtosis parameter is improved in several regression models with high prediction performance. The research results provide a certain reference for the regression model for the prediction of the concentration of other hydrocarbon pollutants in the soil surface layer, and provides an effective detection means for the sustainable development of agricultural production and soil environment. [ABSTRACT FROM AUTHOR]

In alumina production, the evaporation as the key process uses recyclable resources and reduces environmental pollution. In fact, the quality of export product with offline and delayed, results in low precision of process control and high energy consumption. To ensure green and efficient production, in this paper, a new prediction method integrating process knowledge and data-driven spatial-temporal adaptive model is put forward. First, to preprocessed production data for ensuring modeling accuracy, data reconciliation technology is adopted. Then, based on material and heat transfer mechanism, for equipment and industrial process, the mechanism models are established. Furthermore, with time difference and moving window model, an error compensation method is utilized in terms of double locally weighted kernel PLS for estimation error in hypothesis-based mechanism modeling. Finally, the data-driven spatial-temporal adaptive model and the process knowledge-based mechanism model are integrated. To illustrate the model feasibility, an industrial sodium aluminate solution evaporation is used. It demonstrates that, for the developed model, the prediction accuracy can reach more than 90% within the ± 2% error range, and effectively estimate the actual product quality and ensure the prediction effect. • A new concentration prediction method integrating knowledge and data-driven spatial-temporal adaptive model is proposed. • Mechanism modelling based on data reconciliation and process analysis is achieved to estimate concentration. • A TD-MW-DLWKPLS model considering both sample similarity and variable correlation is adopted for error compensation. • The performances of the developed method are validated on evaporation industrial application. [ABSTRACT FROM AUTHOR]

The new energy structure needs to balance energy security and dual carbon goals, which has brought major challenges to coal-fired power plants. The pollution reduction and carbon emissions reduction in coal-fired power plants will be a key task in the future. In this paper, an optimization technique for the operation of an electrostatic precipitator is proposed. Firstly, the voltage-current model is constructed based on the modified dust charging mechanism; the modified parameters are trained through the gradient descent method. Then, the outlet dust concentration prediction model is constructed by coupling the mechanism model with the data model; the data model adopts the long short-term memory network and the attention mechanism. Finally, the particle swarm optimization algorithm is used to achieve the optimal energy consumption while ensuring stable outlet dust concentration. By training with historical data collected on site, accurate predictions of the secondary current and outlet dust concentration of the electrostatic precipitator have been achieved. The mean absolute percentage error of the voltage-current characteristic model is 1.43%, and the relative root mean-squared error is 2%. The mean absolute percentage error of the outlet dust concentration prediction model on the testing set is 5.2%, and the relative root mean-squared error is 6.9%. The optimization experiment is carried out in a 330 MW coal-fired power plant. The results show that the fluctuation of the outlet dust concentration is more stable, and the energy saving is about 43% after optimization; according to the annual operation of 300 days, the annual average carbon reduction is approximately 2621.34 tons. This method is effective and can be applied widely. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Chongqing University of Technology (Natural Science) is the property of Chongqing University of Technology and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

With the rapid growth of urbanization and the proliferation of mobile phone usage, smart city initiatives have gained momentum in leveraging data-driven insights to enhance urban planning and resource allocation. This paper proposes a novel approach for predicting people's concentration and movements within a smart city environment using mobile phone data provided by telecommunication operators. By harnessing the vast amount of anonymized and aggregated mobile phone data, we present a predictive framework that offers valuable insights into urban dynamics. The methodology involves collecting and processing location-based data obtained from telecommunication operators. Using machine learning techniques, including clustering and spatiotemporal analysis, we developed models to identify patterns in people's movements and concentration across various city regions. Our proposed approach considers factors such as time of day, day of the week, and special events to capture the intricate dynamics of urban activities. The predictive models presented in this paper demonstrate the ability to predict areas of high concentration of people, such as commercial districts during peak hours, as well as the people flow during the time. These insights have significant implications for urban planning, traffic management, and resource allocation. Our approach respects user privacy by working with aggregated and anonymized data, ensuring compliance with privacy regulations and ethical considerations. The proposed models were evaluated using real-world mobile phone data collected from a smart city environment in Lisbon, Portugal. The experimental results demonstrate the accuracy and effectiveness of our approach in predicting people's movements and concentration. This paper contributes to the growing field of smart city research by providing a data-driven solution for enhancing urban planning and resource allocation strategies. As cities continue to evolve, leveraging mobile phone data from telecommunication operators can lead to more efficient and sustainable urban environments. [ABSTRACT FROM AUTHOR]

Abstract Background A Raman spectroscopy method can quickly and accurately measure the concentration of ofloxacin in solution. This method has the advantages of accuracy and rapidity over traditional detection methods. However, the manual analysis methods for the collected Raman spectral data often ignore the nonlinear characteristics of the data and cannot accurately predict the concentration of the target sample. Methods To address this drawback, this paper proposes a novel kernel-Huber loss function that combines the Huber loss function with the Gaussian kernel function. This function is used with an improved genetic algorithm-convolutional neural network (GA-CNN) to model and predict the Raman spectral data of different concentrations of ofloxacin in solution. In addition, the paper introduces recurrent neural networks (RNN), long short-term memory (LSTM), bidirectional long short-term memory (BiLSTM) and gated recurrent units (GRU) models to conduct multiple experiments and use root mean square error (RMSE) and residual predictive deviation (RPD) as evaluation metrics. Results The proposed method achieved an $$R^2$$ R 2 of 0.9989 on the test set data and improved by 3% over the traditional CNN. Multiple experiments were also conducted using RNN, LSTM, BiLSTM, and GRU models and evaluated their performance using RMSE, RPD, and other metrics. The results showed that the proposed method consistently outperformed these models. Conclusions This paper demonstrates the effectiveness of the proposed method for predicting the concentration of ofloxacin in solution based on Raman spectral data, in addition to discussing the advantages and limitations of the proposed method, and the study proposes a solution to the problem of deep learning methods for Raman spectral concentration prediction.

Background: A Raman spectroscopy method can quickly and accurately measure the concentration of ofloxacin in solution. This method has the advantages of accuracy and rapidity over traditional detection methods. However, the manual analysis methods for the collected Raman spectral data often ignore the nonlinear characteristics of the data and cannot accurately predict the concentration of the target sample. Methods: To address this drawback, this paper proposes a novel kernel-Huber loss function that combines the Huber loss function with the Gaussian kernel function. This function is used with an improved genetic algorithm-convolutional neural network (GA-CNN) to model and predict the Raman spectral data of different concentrations of ofloxacin in solution. In addition, the paper introduces recurrent neural networks (RNN), long short-term memory (LSTM), bidirectional long short-term memory (BiLSTM) and gated recurrent units (GRU) models to conduct multiple experiments and use root mean square error (RMSE) and residual predictive deviation (RPD) as evaluation metrics. Results: The proposed method achieved an R 2 of 0.9989 on the test set data and improved by 3% over the traditional CNN. Multiple experiments were also conducted using RNN, LSTM, BiLSTM, and GRU models and evaluated their performance using RMSE, RPD, and other metrics. The results showed that the proposed method consistently outperformed these models. Conclusions: This paper demonstrates the effectiveness of the proposed method for predicting the concentration of ofloxacin in solution based on Raman spectral data, in addition to discussing the advantages and limitations of the proposed method, and the study proposes a solution to the problem of deep learning methods for Raman spectral concentration prediction. [ABSTRACT FROM AUTHOR]

Analyzing the influencing factors of PM2.5 concentration, scenario simulations, and countermeasure research to address the problem of PM2.5 pollution in Guangdong Province is of great significance for governments at all levels for formulating relevant policies. In this study, the ChinaHighPM2.5 dataset and economic and social statistics for Guangdong Province from 2010 to 2019 were selected, and a PM2.5 pollution management compliance path formulation method based on the multi-scenario simulation was proposed by combining the differences in city types and PM2.5 concentration prediction. Based on the prediction model of PM2.5 concentration constructed by the Ridge and SVM models and facing the PM2.5 pollution control target in 2025, the urban PM2.5 pollution control scenario considering the characteristics of urban development was constructed. According to the scenario simulation results of the PM2.5 prediction model, the PM2.5 pollution control path suitable for Guangdong Province during the 14th Five-Year Plan period was explored. The coupling coordination model was used to explore the spatial and temporal pattern evolution of PM2.5 pollution collaborative governance in various prefecture-level cities under the standard path, and the policy recommendations for PM2.5 pollution control during the 14th Five-Year Plan period are proposed. The results showed the following: ① in the case of small samples, the model can provide effective simulation predictions for the study of urban pollutant management compliance pathways. ② Under the scenario of PM2.5 management meeting the standard, in 2025, the annual average mass concentration of PM2.5 in all prefecture-level cities in Guangdong Province will be lower than 22 μg/m3, and the annual average concentration of PM2.5 in the whole province will drop from 25.91 μg/m3 to 21.04 μg/m3, which will fulfil the goal of reducing the annual average concentration of PM2.5 in the whole province to below 22 μg/m3, as set out in the 14th Five-Year Plan for the Ecological Environmental Protection of Guangdong Province. ③ Under the path of PM2.5 control and attainment, the regional coordination relationship among prefecture-level cities in Guangdong Province is gradually optimized, the number of intermediate-level coordinated cities will increase, and the overall spatial distribution pattern will be low in the middle and high in the surrounding area. Based on the characteristics of the four city types, it is recommended that a staggered development strategy be implemented to achieve synergy between economic development and environmental quality. Urban type I should focus on restructuring freight transportation to reduce urban pollutant emissions. City type II should focus on urban transportation and greening. For city type III, the focus should be on optimizing the industrial structure, adjusting the freight structure, and increasing the greening rate of the city. For city type IV, industrial upgrading, energy efficiency, freight structure, and management of industrial pollutant emissions should be strengthened. [ABSTRACT FROM AUTHOR]

Laser-induced breakdown spectroscopy (LIBS) technique is employed for quantitative analysis of aluminum samples by different classical machine learning approaches. A Q-switch Nd:YAG laser at a fundamental harmonic of 1064 nm is utilized for the creation of LIBS plasma in order to predict constituent concentrations of the aluminum standard alloys. In the current research, concentration prediction is performed by linear approaches of support vector regression (SVR), multiple linear regression (MLR), principal component analysis (PCA) integrated with MLR (PCA–MLR), and SVR (PCA–SVR), as well as nonlinear algorithms of artificial neural network (ANN), kernelized support vector regression (KSVR), and the integration of traditional principal component analysis with KSVR (PCA–KSVR), and ANN (PCA–ANN). Furthermore, dimension reduction is applied to various methodologies by the PCA algorithm in order to improve the quantitative analysis. The results indicated that the combination of PCA with the KSVR algorithm model had the best efficiency in predicting most of the elements among other classical machine learning algorithms. [ABSTRACT FROM AUTHOR]

PM 2.5 is a significant environmental pollutant that damages the environment and endangers human health. Precise forecast of PM 2.5 concentrations is very important to control air pollution and improve people's life quality. In the subway indoor air quality (IAQ) system, the data collected by telemonitoring systems is frequently lost due to many reasons. A deep learning model called RF-CNN-GRU, which combines random forest (RF), convolutional neural network (CNN) and gated recurrent unit (GRU), is proposed to predict atmospheric PM 2.5 concentrations with incomplete original data. The RF-CNN-GRU model employs the RF to fill in missing values in the data and subsequently applies the CNN to extract features from the imputed data. The data is finally sent to the GRU network to train and predict PM 2.5 concentrations. Comparing with single CNN, GRU and long short-term memory (LSTM) models, the predictive accuracy of the RF-CNN-GRU model is significantly improved. The RF-CNN-GRU model shows a slight improvement in prediction results when compared to models such as CNN-GRU, RF-CNN, RF-GRU, and RF-LSTM. The findings demonstrate that the RF-CNN-GRU model has excellent accuracy in PM 2.5 concentration prediction when the original data is incomplete. [ABSTRACT FROM AUTHOR]

China's sour gas reservoir is very rich in reserves, taking the largest whole offshore natural gas field in China-Puguang gas field as an example, its hydrogen sulfide content reaches 14.1%. The use of renewable energy, such as solar energy through photocatalytic technology, can decompose hydrogen sulfide into hydrogen and monomeric sulfur, thus realizing the conversion and resourceization of hydrogen sulfide gas, which has important research value. In this study, a concentration sample database of a hydrogen sulfide leakage scenario in a chemical park is constructed by Fluent software simulation, and then a leakage concentration prediction model is constructed based on the data samples to predict the hydrogen sulfide leakage diffusion concentration in real-time. Several machine learning algorithms, such as neural networks, support vector machines, and deep confidence networks, are implemented and compared to find the model algorithm with the best prediction performance. The prediction performance of the support vector machine model optimized by the sparrow search algorithm is found to be the best. The prediction model ensures the accuracy of the prediction results while greatly reducing the computational time cost, and the accuracy meets the requirements of practical engineering applications. [ABSTRACT FROM AUTHOR]

Pseudomonas syringae pv. Lachrymans-8 and Fusarium graminearum-ACCC37687 were used to quickly identify the pathogenic microorganisms of plant fungal diseases and bacterial diseases. To explore the feasibility of rapid identification of plant fungal and bacterial diseases by three-dimensional fluorescence spectroscopy. By collecting the three-dimensional fluorescence spectrum data of gradient mixed bacterial solution samples, the data were analyzed by second-order correction algorithm alternating trilinear decomposition (ATLD), parallel factor analysis (PARAFAC), self weighted trilinear decomposition (SWATLD), alternating penalty trilinear decomposition (APTLD) and first-order algorithm partial least squares regression coefficient method (PLS), and the characteristic excitation and emission wavelengths were extracted. The concentration prediction model was established by multiple linear regression between the fluorescence intensity data of characteristic wavelength and the absorbance (OD600) of bacterial solution at 600 nm wavelength. The prediction performance of the model was measured by the left one out cross validation (LOOCV), so as to realize the qualitative and quantitative analysis of single component in complex bacterial solution mixed system. The fluorescence characteristic peaks of Pseudomonas syringae were excitation/emission=285 nm/340 nm, 290 nm/340 nm, 285 nm/332.4 nm, 280 nm/361.6 nm, 295 nm/361.6 nm. The fluorescence characteristic peaks of Fusarium graminearum were excitation/emission=380 nm/468 nm, 390 nm/512 nm, 340 nm/511.2 nm, 415 nm/511.2 nm. The results showed that the concentration prediction model of Pseudomonas syringae-8 (R2cv=0.92441191, RMSEP=0.005163633, R=0.961463421) was better than that of Fusarium graminearum-ACCC37687 (R2cv=0.583953931, RMSEP=0.027653679, R=0.764168784). The results of this study provide an available method for rapid identification of fungal and bacterial hazards.

Methane (CH 4) and nitrous oxide (N 2 O) as two typical greenhouse gases, have important effects on global climate change. In this paper, an external cavity quantum cascade laser (ECQCL) based gas sensor was developed for simultaneous CH 4 and N 2 O detection by employing calibration-free direct absorption spectroscopy. In view of the important influence of spectral noise and background normalization process on gas concentration inversion, dual-convolutional neural networks (D-CNN) and baseline normalization algorithms were developed for spectral signal de-noising and concentration inversion, respectively. Compared to traditional methods, the results indicate that the proposed D-CNN de-noising algorithm can successfully improve the signal-to-noise ratio (SNR) by 2.37 times, and the correlation coefficients of the retrieved concentrations were improved from 0.9965 to 0.9991 for CH 4 and 0.9983 to 0.9994 for N 2 O, respectively, which shows a great potential for analyzing unresolved mixture absorption spectra with high-precision and accuracy in simultaneous detection of multiple gas components. • A versatile ECQCL based gas sensor was developed for simultaneous CH 4 and N 2 O detection. • Calibration-free direct absorption spectroscopy was utilized for acquiring molecular fingerprint spectra. • Dual-convolutional neural networks (D-CNN) algorithms were proposed for signal de-noising and concentration inversion. • The D-CNN algorithms assisted sensor shows a great potential for simultaneous detection of multiple components with unresolved spectra. [ABSTRACT FROM AUTHOR]

Accurate prediction of PM 2.5 concentrations in ports is crucial for authorities to combat ambient air pollution effectively and protect the health of port staff. However, in port clusters formed by multiple neighboring ports, we encountered several challenges owing to the impact of unique meteorological conditions, potential correlation between PM 2.5 levels in neighboring ports, and coupling influence of background pollutants in city zones. Therefore, considering the spatiotemporal correlation among the factors influencing PM 2.5 concentration variations within the harbor cluster, we developed a novel blending ensemble deep learning model. The proposed model combined the strengths of four deep learning architectures: graph convolutional networks (GCN), long short-term memory networks (LSTM), residual neural networks (ResNet), and convolutional neural networks (CNN). GCN, LSTM, and ResNet served as the base models aimed at capturing the spatial correlation of PM 2.5 concentrations in neighboring ports, the potential long-term dependence of meteorological factors and PM 2.5 concentrations, and the effects of urban ambient air pollutants, respectively. Following the blending ensemble technique, the prediction outcomes of three base models were used as the input data for the meta-model CNN, which employs the blending ensemble technique to produce the final prediction results. Based on actual data obtained from 18 ports in Nanjing, the proposed model was compared and analyzed for its prediction performance against six state-of-the-art models. The findings revealed that the proposed model provided more accurate predictions. It reduced mean absolute error (MAE) by 10.59 %–20.00 %, reduced root mean square error (RMSE) by 13.22 %–17.11 %, improved coefficient of determination (R2) by 10 %–35.38 %, and improved accuracy (ACC) by 3.48 %–7.08 %. Additionally, the contribution of each component to the prediction performance of the proposed model was measured using a systematic ablation study. The results demonstrated that the GCN model exerted the most substantial influence on the prediction performance of the GCN-LSTM-ResNet model, followed by the LSTM model. The influence of urban background pollutants can significantly enhance the generalizability of the complete model. Moreover, a comparison with three blended ensemble models incorporating any two base models demonstrated that the GCN-LSTM-ResNet model exhibited superior prediction performance and was particularly excellent in predicting the occurrence of high-concentration events. Specifically, the GCN-LSTM-ResNet model improved MAE and RMSE by at least 12.3% and 9.2%, respectively, but reduced R2 and ACC by 26.1% and 6.8%, respectively. The proposed model provided reliable PM 2.5 concentration prediction outcomes and decision support for air quality management strategies in dry bulk port clusters. [Display omitted] • A novel deep learning model was developed to capture spatiotemporal correlations affecting PM 2.5 variations within port clusters and urban background pollutants. • Comparison with six state-of-the-art models, the GCN-LSTM-ResNet model significantly improved prediction performance. • A systematic ablation study revealed that the GCN module significantly impacts the GCN-LSTM-ResNet model's prediction performance at port cluster zones. [ABSTRACT FROM AUTHOR]

There are two primary methods to address metal oxide semiconductor sensor's poor selectivity issues, dynamic testing and constructing sensor arrays. The integration of sensor array with convolutional neural network (CNN) has exhibited tremendous potential in the direction of gas mixture identification. However, the utilization of sensor arrays often leads to increased costs, complexity, and elevated failure rates. Hence, in this work, we amalgamated CNN and single sensor's dynamic testing methodologies to conduct qualitative and quantitative identification of methanol-ethanol mixtures. The acquisition of dynamic response data was achieved through temperature modulation. Prudent adjustments to the heating waveform and associated parameters extended the one-dimensional information inherent in gas sensitive response, effectively addressing the challenges of poor selectivity encountered in metal oxide semiconductor gas sensors. Subsequently, two CNNs were harnessed for the categorization and concentration prognosis of mixed gases. The utilization of a two-dimensional CNN for recognizing data transformed by the gramian angular field resulted in the highest classification accuracy of 99.33 %. Furthermore, one-dimensional CNN exhibited minimal errors when estimating the concentrations of both components within the mixture. Experimental results indicate that the proposed approach can achieve the classification and concentration prediction of binary mixed gases with high precision, potentially emerging as a novel strategy to resolve gas mixture identification challenges. [Display omitted] • Binary gas mixture identification using a single, simple sensor. • Temperature modulation voltage is optimized for gas mixture data acquisition. • Test set is unknown concentration data from other sensors. • Deep learning models studied for mixture classification and regression. [ABSTRACT FROM AUTHOR]

Micro/nanoplastics (MNPs) and heavy metals (HMs) coexist worldwide. Existing studies have reported different or even contradictory toxic effects of co-exposure to MNPs and HMs on plants, which may be related to various influencing factors. In this study, existing publications were searched and analyzed using CiteSpace, meta-analysis, and machine learning. CiteSpace analysis showed that this research field was still in the nascent stage, and hotspots in this field included accumulation, cadmium (Cd), growth, and combined toxicity. Meta-analysis revealed the differential association of seven influencing factors (MNP size, pollutant treatment duration, cultivation media, plant species, MNP type, HM concentration, and MNP concentration) and 8 physiological parameters receiving the most attention. Co-exposure of the two contaminants had stronger toxic effects than HM treatment alone, and phytotoxicity was generally enhanced with increasing concentrations and longer exposure durations, especially when using nanoparticles, hydroponic medium, dicotyledons producing stronger toxic effects than microplastics, soil-based medium, and monocotyledons. Dry and fresh weight analysis showed that co-exposure to MNPs and Cd resulted in significant phytotoxicity in all classifications. Concerning the MNP types, polyolefins partially attenuated plant toxicity, but both modified polystyrene (PS) and biodegradable polymers exacerbated joint phytotoxicity. Finally, machine learning was used to fit and predict plant HM concentrations, showing five classifications with an accuracy over 80 %, implying that the polynomial regression model could be used to predict HM content in plants under complex pollution conditions. Overall, this study identifies current knowledge gaps and provides guidance for future research. [Display omitted] • The research about the effects of microplastic/nanoplastic and heavy metal co-exposure on plants is still in the nascent stage. • The toxic effects of microplastic/nanoplastics and heavy metal co-exposure on plant physiological parameters were influenced by various factors. • Plant heavy metal content prediction in the five classifications was demonstrated with an accuracy over 80 %. [ABSTRACT FROM AUTHOR]

Porosity is one of the core parameters for Eq.10 which can be measured by complex instruments but can also be calculated as follows: (12) Graph HT ht When the mass of the air can be neglected, Graph HT ht , the fiber mass Graph HT ht approximately equal to fabric mass Graph HT ht . [Extracted from the article]

The Fourier transform infrared (FTIR) spectra combined with one-dimensional convolutional neural network (1DCNN) based on multi-level feature fusion, that is, MLF-1DCNN, were used to determine the concentration of imidacloprid in water. The FTIR spectra of imidacloprid water solutions with different concentrations (0-0.41 g/L) in 700-4,000 cm-1 were measured and the corresponding dataset was constructed, and the concentrations were predicted by the MLF-1DCNN. The effect of the spectral data preprocessing by multivariate scattering correction (MSC) and standard normal variate (SNV) transformation on improving the concentration prediction accuracy was studied. The result shows that the SNV preprocessing has the better prediction effect. The comparison of our model with partial least squares (PLS), support vector regression (SVR) and multiple linear regression (MLR) shows that our model can effectively predict the imidacloprid concentrations with a higher prediction accuracy than the other comparative models. The results obtained in this study demonstrate the analytical potential of applying this method to rapidly predict imidacloprid concentration in water. [ABSTRACT FROM AUTHOR]

The green and pollution-free mining of resources has always been a research field that people have focused on. In the process of mining resources, the production of CO, SO2 and other pollutants directly affects the health of miners and the atmospheric environment in the mining area. Therefore, it is particularly important to deal with and control the polluting gases generated by mining. Taking the underground roadway of a coal mine in Hengdong City, Hunan Province, as the research object, we studied the migration law of pollutant gas there. Comsol software was used to determine the changing state of pollutant gas migration in the roadway, and a simulation model of the wind field and the pollutant concentration field in the roadway under turbulent conditions was established. The results showed that, when the air flow moved to the front face of the roadway, it generated backflow to form a counterclockwise-rotating air flow vortex. Here, the air flow stagnated, hindering the diffusion of pollutants. The gas moved with the air flow in the roadway, and the flow's velocity decreased in the middle of the roadway, causing pollutants to accumulate. The whole wind field tended to be stable at a plane 25 m from the roadway's outlet. This indicates that the middle part of the roadway is the place where the polluted gas accumulates, and it is of representative significance to study the concentration of the polluted gas in the roadway in this section. [ABSTRACT FROM AUTHOR]

Accurate prediction of PM2.5 concentration is one of the key tasks of air pollution assessment, early warning, and treatment. In this paper, four monitoring sites were arranged in Jiangbei New District of Nanjing City, China. The environmental parameters such as PM2.5/PM10 concentration, temperature, and humidity were monitored from January to February 2020. A gated recurrent unit (GRU) network based on the PM2.5 concentration prediction model was established to predict PM2.5 concentration. The mean relative error (MRE), root mean square error (RMSE), and Pearson correlation coefficient were selected as the evaluation criteria for the accuracy of the GRU model. The data set was divided into a training set, a test set and a validation set at a ratio of 7:2:1, and the GRU model was used to predict the hourly value of PM2.5 concentration in the next week. The prediction results show that the Pearson correlation coefficients between the predicted values and the monitored values of the four monitoring sites have reached more than 0.9, reflecting a strong correlation. The relative average errors are around 10%. The GRU model prediction of NJAU (Nanjing Agricultural University)-Pukou Campus Site is the most accurate, and the correlation coefficient, MRE, and RMSE are 0.970, 7.85%, and 9.6049, respectively, reflecting the good prediction performance of the model. Therefore, this research supports the prediction of air quality in different cities and regions, so people can take protective measures in advance and reduce the damage caused by air pollution to human bodies. [ABSTRACT FROM AUTHOR]

Heavy metal pollution poses a huge challenge to the soil environment. With the increasing pollution level, the traditional monitoring methods cannot quickly obtain information on large-area pollution. Therefore, a large-scale mapping method with high precision is urgently needed to effectively control heavy metal pollution. This study explored a method for mapping soil heavy metal concentrations through hyperspectral images. On this basis, a new Stacked AdaBoost ensemble learning algorithm was constructed to construct the inversion model of soil heavy metal contents. The characteristic spectral bands of heavy metals were extracted as model input variables using Pearson’s correlation coefficient and successive projections algorithm. With three sets of heavy metal content data, the prediction accuracy and mapping outcomes of various machine learning methods were compared. Furthermore, the potential sources of heavy metal pollution in the study area were analyzed based on the Moran’s index. The results showed that the Stacked AdaBoost model was relatively stable with higher accuracy than traditional machine learning models. For Cr, Cu, and As, the determination coefficients (R2) of the verification set were 0.66, 0.61, and 0.74, respectively. Afterward, the results of this model were used to map the heavy metal concentration over the study area. The mapping results suggested that the heavy metal conditions of soils in the Ganhetan area were caused by nature and human activities. The As pollution in agricultural soils was the most serious, with an exceedance rate of 38.66%. Industrial areas were potential sources of soil heavy metal pollution in the study area. In summary, the Stacked AdaBoost ensemble learning model provides detailed and reliable data for agricultural ecological protection and industrial pollution control, allowing the effective management of heavy metal pollution sources.

Analyzing the influencing factors of PM2.5 concentration, scenario simulations, and countermeasure research to address the problem of PM2.5 pollution in Guangdong Province is of great significance for governments at all levels for formulating relevant policies. In this study, the ChinaHighPM2.5 dataset and economic and social statistics for Guangdong Province from 2010 to 2019 were selected, and a PM2.5 pollution management compliance path formulation method based on the multi-scenario simulation was proposed by combining the differences in city types and PM2.5 concentration prediction. Based on the prediction model of PM2.5 concentration constructed by the Ridge and SVM models and facing the PM2.5 pollution control target in 2025, the urban PM2.5 pollution control scenario considering the characteristics of urban development was constructed. According to the scenario simulation results of the PM2.5 prediction model, the PM2.5 pollution control path suitable for Guangdong Province during the 14th Five-Year Plan period was explored. The coupling coordination model was used to explore the spatial and temporal pattern evolution of PM2.5 pollution collaborative governance in various prefecture-level cities under the standard path, and the policy recommendations for PM2.5 pollution control during the 14th Five-Year Plan period are proposed. The results showed the following: ① in the case of small samples, the model can provide effective simulation predictions for the study of urban pollutant management compliance pathways. ② Under the scenario of PM2.5 management meeting the standard, in 2025, the annual average mass concentration of PM2.5 in all prefecture-level cities in Guangdong Province will be lower than 22 μg/m3, and the annual average concentration of PM2.5 in the whole province will drop from 25.91 μg/m3 to 21.04 μg/m3, which will fulfil the goal of reducing the annual average concentration of PM2.5 in the whole province to below 22 μg/m3, as set out in the 14th Five-Year Plan for the Ecological Environmental Protection of Guangdong Province. ③ Under the path of PM2.5 control and attainment, the regional coordination relationship among prefecture-level cities in Guangdong Province is gradually optimized, the number of intermediate-level coordinated cities will increase, and the overall spatial distribution pattern will be low in the middle and high in the surrounding area. Based on the characteristics of the four city types, it is recommended that a staggered development strategy be implemented to achieve synergy between economic development and environmental quality. Urban type I should focus on restructuring freight transportation to reduce urban pollutant emissions. City type II should focus on urban transportation and greening. For city type III, the focus should be on optimizing the industrial structure, adjusting the freight structure, and increasing the greening rate of the city. For city type IV, industrial upgrading, energy efficiency, freight structure, and management of industrial pollutant emissions should be strengthened.

An electronic nose (E-nose) based on three metal oxide semiconductor (MOS) gas sensors is designed to quantitatively analyse six types of volatile organic compounds (VOCs). Support vector machine (SVM), extreme learning machine (ELM), and back-propagation (BP) neural network, are used to design different classifiers and regressors for integrating a suitable pattern recognition model of the E-nose system. By using the output of the classifier as one of the input features of the regressor, the models can predict the concentration of different types of VOCs at the same time. The 5-fold cross-validation is applied to search optimal parameters of each model and the independent test is conducted to evaluate the generalization performance. A pipeline is used to connect the best classifier and the best regressor, constructing an integrated model for pattern recognition. The integrated model based on ELM-ELM structure exhibits the best performance. Classification accuracy can be as high as 99% and the R2 score of regression as high as 0.97 in the 5-fold cross-validation. Classification accuracy is up to 93% and the R2 score of regression is up to 0.94 in the independent tests. Furthermore, the integrated model has prominent training and test efficiency, of which time consumptions are both shorter than 0.1 s. This work provides an approach to design an efficient integrated model with high performance for gas type identification and concentration prediction in the E-nose system. [ABSTRACT FROM AUTHOR]

The quick leakage alarm and the accurate concentration prediction are two important aspects of natural gas safety monitoring. In this paper, a rapid monitoring method of sensor data sharing, rapid leakage alarm and simultaneous output of concentrations prediction is proposed to accelerate the alarm speed and predict the possible impact of leakage. In this method, the Dempster-Shafer evidence theory is used to fuse the trend judgment and the CUSUM (cumulative sum) and the Gauss-Newton iteration is used to predict the concentration. The experiment system based on the TGS2611 natural gas sensor was built. The results show that the fusion method is significantly better than the single monitoring method. The alarm time of fusion method was more advanced than that of the CUSUM method and the trend method (being averagely, 10.4% and 7.6% in advance in the CUSUM method and the trend method respectively). The relative deviations of the predicted concentration were the maximum (13.3%) at 2000 ppm (parts per million) and the minimum (0.8%) at 6000 ppm, respectively.

Electronic nose (e-nose) technology has emerged as a pivotal tool in various domains, which has been widely utilized for odor identification, concentration evaluation, and prediction tasks. This review provides a comprehensive survey on the most recent advances in the development of e-nose systems and their algorithmic applications, emphasizing the roles of various methodologies and deep learning technologies in odor classification and concentration forecasting. Additionally, we delve into model evaluation methods, including multidimensional performance assessment and cross-validation. Future trends encompass broader application domains, advanced drift correction techniques, comprehensive multifactorial analysis, and enhanced capabilities for dealing with unknown interferents. These trends are set to propel significant breakthroughs in e-nose technology within scientific research and practical applications, solidifying the e-nose system as a crucial tool in many areas such as environmental monitoring, biomedicine, and public safety. [Display omitted] • The most recent advances in e-noses and their algorithmic applications are outlined. • Various methodologies and deep learning technologies in odor classification and concentration forecasting are summarized. • Future trends of e-noses as crucial tools in many areas are proposed and discussed. [ABSTRACT FROM AUTHOR]

Gases typically exist as mixed states, which normally contain more information and exhibit more intricate features than single gas. Hence, the task of predicting mixed gas concentrations poses a challenging endeavor in the field of gas detection. In this study, a Variational Mode Decomposition method was combined with Long and Short-term Time-series Network-Attention to build a VMD-LSTNet-Attention model, based on which the response signals of electronic nose (E-nose) is processed, achieving a high-precision concentrations prediction of carbon monoxide (CO) and ethylene mixed gases. Firstly, a dataset containing the raw response signals of CO and ethylene was collected using an gas data collection system. Then, VMD algorithm was used to extract and decompose the essential features from the E-nose signals into multiple components. Further, in pursuit of optimal decomposition results, Particle Swarm Optimization (PSO) algorithm was employed to optimize the penalty factor α and determine the most effective decomposition level K. Subsequently, the LSTNet-Attention model was employed to analyze the concentrations of mixed gases, and the predictive R-square values for CO and ethylene reach 0.993 and 0.975, respectively. Finally, the comparison with the traditional deep learning models and the popular models demonstrates the effectiveness of the proposed VMD-LSTNet-Attention model for concentration prediction of mixed gases. • A VMD algorithm was proposed to address the challenges in mixed gases analysis. • A mixed gases dataset was generated to validated the effectiveness of the model. • High-precision concentration prediction has been achieved by signal decomposition. [ABSTRACT FROM AUTHOR]

The demand for gas sensors is experiencing rapid growth, driven by the increasing need for air quality detection in the face of environmental pollution and industrial emissions. However, the widely used metal oxide semiconductor (MOS) gas sensors suffer from drawbacks such as low ability to predict gas concentration and poor selectivity. In this work, a sensor array was fabricated consisting of 10 various gas sensors and combined with an advanced deep learning algorithm to achieve precise predictions of nitrogen dioxide (NO 2) at low concentrations. First, graphitic carbon nitride (g-C 3 N 4) and In 2 O 3 nanoparticles were composited with different mass ratios, resulting in the formation of distinct g-C 3 N 4 /In 2 O 3 composites as sensing materials. Afterward, SEM, TEM, and XRD were employed to characterize the morphology, structure, and elemental composition of the samples. Furthermore, sensing properties including response, selectivity, and repeatability, are studied for 10 gas sensors based on g-C 3 N 4 /In 2 O 3 composites. Finally, the convolutional neural network-efficient channel attention-gate recursive unit (CNN-AGRU) model was proposed to analyze the patterns of the signals collected from the sensor array when exposed to various NO 2 concentrations from 1 to 9 ppm. The results demonstrated the integration of high-selectivity sensing materials to NO 2 and an advanced deep learning algorithm CNN-AGRU enables to realize a high concentration prediction accuracy of about 97.04% and a low prediction error of 0.20527 ppm for NO 2 gas. • This work combines sensing material and algorithm for NO 2 concentration prediction. • Highly responsive and selective g-C 3 N 4 /In 2 O 3 composites were prepared. • Collected low NO 2 concentration datasets in the range of 1–9 ppm. • Proposed a CNN-AGRU pattern recognition algorithm that achieved high accuracy. [ABSTRACT FROM AUTHOR]

Recurrent Neural Networks are classes of Artificial Neural Networks that establish connections between various nodes in a directed or undirected graph for investigation of the temporal dynamical. In this study, different Recurrent Neural Network (RNN) architectures are utilized for quantitative analysis of aluminum alloys by the laser induced breakdown spectroscopy (LIBS) technique. The fundamental harmonic (1064 nm) of a nanosecond Nd:YAG laser pulse is employed to generate the LIBS plasma for the prediction of constituent concentrations of the aluminum standard samples. For the purpose of predicting concentration, Recurrent Neural Networks based on different networks, such as Long Short Term Memory (LSTM), Gated Recurrent Unit (GRU), Simple Recurrent Neural Network (Simple RNN), and as well as Recurrent Convolutional Networks comprising of Conv-SimpleRNN, Conv-LSTM and Conv-GRU are employed. Then, a comparison is made among prediction by classical machine learning methods of support vector regressor (SVR), the Multilayer Perceptron (MLP), Decision Tree algorithm, Gradient Boosting Regression (GBR), Random Forest Regression (RFR), Linear Regression, and k-Nearest Neighbor (KNN) algorithm. Results demonstrated that the machine learning tools based on Convolutional Recurrent Networks had the best efficiencies in prediction of the most of the elements among other multivariate methods. [ABSTRACT FROM AUTHOR]

[Display omitted] • A new ensemble learning-based UV absorption spectroscopy analysis network, SFA, was proposed. • Local feature information extraction module and global feature information extraction module were established. • A dataset of ultraviolet spectra for mixed copper and zinc heavy metal ions in seawater mixed solutions at different concentrations was established. Heavy metal pollution poses a significant threat to marine ecological environments and serves as a crucial indicator for seawater quality assessment. The enrichment of heavy metals severely impacts marine ecosystem health and poses threats to human food safety, underscoring the critical importance of monitoring heavy metal levels in marine waters. In this paper, UV absorption spectroscopy was used to collect the spectra of seawater containing different concentrations of Cu and Zn ions, which overcomes the problems of traditional methods that require extensive pre-treatment and lossy samples. To address the characteristics of the data, the Savitzky-Golay (SG) filtering algorithm is used for preprocessing. The acquired spectra are then subjected to further data analysis using the proposed multidimensional information fusion method called SpectraNet Fusion Algorithm (SFA) for UV absorption spectroscopy. The local feature information extraction module and the whole domain information feature extraction module were respectively used to predict the regression of the data in parallel two-channel, and the output results were integrated to form a new feature map, which was further processed by the feature information fusion module, and finally output the prediction results corresponding to the spectra to realize the accurate detection of Cu and Zn heavy metal ions in seawater. The coefficient of determination (R2) of the model training set established by this method reaches 98.00 %, the root mean square error (RMSE) reaches 0.0389, and the mean absolute error (MAE) reaches 0.0243, combined with the experimental design, the detection range was 5 μg·L−1 to 90 μg·L−1,which realizes simultaneous and high-precision prediction of Cu and Zn heavy metal ions in seawater. [ABSTRACT FROM AUTHOR]

The predictive model of aging indicator based on intelligent algorithms has become an auxiliary method for the aging condition of transformer polymer insulation. However, most of the current research on the concentration prediction of aging products focuses on dissolved gases in oil, and the concentration prediction of alcohols in oil is ignored. As new types of aging indicators, alcohols (methanol, ethanol) are becoming prevalent in the aging evaluation of transformer polymer insulation. To address this, this study proposes a prediction model for the concentration of alcohols based on a genetic-algorithm-optimized support vector machine (GA-SVM). Firstly, accelerated thermal aging experiments on oil-paper insulation are conducted, and the concentration of alcohols is measured. Then, the data of the past 4 days of aging are used as the input feature of SVM, and the GA algorithm is utilized to optimize the kernel function parameter and penalty factor of SVM. Moreover, the concentrations of methanol and ethanol are predicted, after which the prediction accuracy of other algorithms and GA-SVM are compared. Finally, an industrial software program for predicting the concentration of methanol and ethanol is established. The results show that the mean square errors (MSE) of methanol and ethanol concentration predictions of the model proposed in this paper are 0.008 and 0.003, respectively. The prediction model proposed in this paper can track changes in methanol and ethanol concentrations well, providing a theoretical basis for the field of alcohol concentration prediction in transformer oil. [ABSTRACT FROM AUTHOR]

As VOCs pose a threat to human health, it is important to accurately capture changes in VOCs concentrations and sense VOCs concentrations in relevant areas. Therefore, it is necessary to improve the accuracy of VOCs concentration prediction and realise the VOCs aggregation situation sensing. Firstly, on the basis of regional grid division, the inverse distance spatial interpolation method is used for spatial interpolation to collect regional VOCs data information. Secondly, extreme gradient boosting (XGBoost) is used for spatio-temporal feature selection, combined with graph convolutional neural network (GCN) to construct regional spatial relationships of VOCs, and multiple linear regression (MLR) to process VOCs time series data and predict the VOCs concentration in the grid. Finally, the aggregation potential values of VOCs are calculated based on the prediction results, and the potential perception results are visualised. A VOCs aggregation perception method based on concentration prediction is proposed, using the XGBoost-GCN-MLR method with a scenario-aware approach for VOCs to perceive the VOCs aggregation in the relevant region. VOCs concentration prediction and VOCs aggregation trend perception were carried out in Xi'an, Baoji, Tongchuan, Weinan and Xianyang. The results show that compared with the GCN model, XGBoost model, MLR model and GCN-MLR model, the XGBoost-GCN-MLR model reduces the input variables, achieves the optimisation of the input parameters of the VOCs concentration prediction model, reduces the complexity of the prediction model and improves the prediction accuracy. Intelligent sensing of VOCs aggregation can visualise the regional VOCs. The intelligent sensing of VOCs aggregation can visualise the development trend and status of regional VOCs aggregation and convey more information, which has practical value. [ABSTRACT FROM AUTHOR]

Field or laboratory measurements are commonly conducted to determine phthalates concentrations in spaces. This study investigated the association between various influencing factors and indoor phthalates concentrations. Back-propagation (BP) Neural Network was employed to verify a prediction model of indoor phthalates concentration with 80% of experimental data and 20% remaining data. The validation of remaining data shows a reasonable accuracy for model application, where the ratios of standard deviations were all greater than 0.45, most ERMS were close to 0 and all the EMR were less than 15.5%. In addition, we used relevant data from the China, Children, Homes, Health (CCHH) study conducted in Tianjin for further inspection. The prediction on di (2-ethylhexyl) phthalate (DEHP) concentration was performed, which indicated a high accuracy. Furthermore, the Monte-Carlo simulation was applied to quantify the effect of temperature on phthalates concentration combining with the prediction model. When the temperature increment value was 4°C, the average relative decrease ratio of dibutyl phthalate (DBP), DEHP, diisobutyl phthalate (DIBP) concentration was about 7.8%, 12.8% and 9.3%, respectively. The findings have established the validity of the prediction model and provided a quantification of the influence of temperature on the concentration of phthalates in the indoor dust phase. [ABSTRACT FROM AUTHOR]

Accurate prediction of PM 10 concentration is important for effectively managing PM 10 exposure and mitigating health and economic risks posed to humans in dry bulk ports. However, accurately capturing the time-series nonlinear variation characteristics of PM 10 concentration is challenging owing to the specific intensity of port operation activities and the influence of meteorological factors. To address such challenges, a novel combined deep learning model （CLAF） was proposed, merging cascaded convolutional neural networks （CNN）, long short-term memory （LSTM）, and an attention mechanism （AM）. This integrated model aimed to forecast hourly PM 10 concentration in dry bulk ports. First, using the random forest characteristic importance algorithm, the distinct meteorological factors were identified among the selected five meteorological factors. These selected factors were incorporated into the prediction model along with the PM 10 concentration. Subsequently, the CNN layer was employed to extract high-dimensional time-varying features from the input variables, while the LSTM layer captured sequential features and long-term dependencies. In the AM layer, different weights were assigned to the output components of the LSTM layer to amplify the effects of important information. Finally, three evaluation metrics were applied to compare the performance of the CLAF model with three basic models and three commonly used prediction models. Real-case data was collected and used in this study. Comparison results demonstrated that considering the meteorological factors could improve the prediction accuracy and fitting performance of PM 10 concentration in ports. The CLAF model reduced the mean absolute error statistic by 13.92%-56.9%, decreased the mean square error statistic by 45.99%-81.02%, and improved the goodness-of-fit statistic by 3.2%-15.5%.

[Objectives] Based on terahertz time domain spectroscopy, three heavy metals of mercury (Hg), cadmium (Cd) and copper (Cu) in water were detected, aiming to find out the characteristic frequency points of three heavy metals within the terahertz spectrum, and provide a method reference for the construction of the classification and concentration identification of three heavy metals in water and the prediction model of the concentration content. [Methods] Hg, Cd and Cu heavy metal solutions of different concentrations were respectively configured, and time domain data of the samples were collected using terahertz spectral attenuation reflection mode, and denoising was performed by discrete cosine transform (DCT), standard normal transformation (SNV) and second derivative (SD). Dimensionality reduction was performed by principal component analysis (PCA), multiple dimension scaling (MDS) and linear discriminant analysis (LDA). Then the detection modeling of heavy metal categories and concentrations were carried out by random forest (RF), k-nearest neighbor (KNN) and probabilistic neural network (PNN). The least squares support vector machines (LSSVM) and back propagation neural network (BPNN) were adopted for concentration prediction modeling. [Results] The results showed that at 1.7 THz and 1.2 THz, the absorption coefficient spectra of Hg and Cd in a certain heavy metal concentration range had obvious peak changes, respectively, while the absorption coefficient spectra of Cu solution in the tested terahertz range with the concentration change rule was not found. The PNN and KNN models could accurately detect and identify three heavy metals in water. The PCA-PNN model could identify the concentrations of Hg, Cd and Cu solutions with the accuracy of 99.45%, 95.93% and 99.25%, respectively. The DCT-LDA-BPNN model could be used to predict the contents of Hg, Cd and Cu in solution with the determination coefficients of 0.996, 0.986 and 0.999, respectively, and the mean square errors of 0.008, 0.026 and 2.164, respectively. [Conclusions] This experiment proved that terahertz spectrum had good qualitative and quantitative analysis ability for Hg, Cd and Cu solutions with different concentrations, which could provide important reference for the detection of heavy metals in water. [ABSTRACT FROM AUTHOR]

[Display omitted] • Predicting brown tide microalgae concentrations using EEM fluorescence spectroscopy. • Reconstruction of EEM into three-dimensional data. • Augmentation of reconstructed data using the channel & spatial attention mechanism. • Using CNN as a predictive model for microalgae concentration. • High improvement in prediction accuracy of reconstructed data compared to original EEM. The non-destructive and sensitive three-dimensional excitation-emission matrix (3D-EEM) fluorescence spectroscopy shows great potential for real-time monitoring of brown tide, while current methods of 3D-EEM spectra combined with convolutional neural network (CNN) do not take full advantage of its spatial characteristics. To address this issue, we propose a data reconstruction method based on 3D-EEM spectra. By stratifying the 3D-EEM spectra according to fluorescence intensity, we can reconstruct the original two-dimensional data of the spectra into more informative three-dimensional data, then the reconstructed data are enhanced according to the characteristics of the reconstructed data by using the attention mechanism, so as to better utilize the spatial characteristics of the 3D-EEM spectra and to improve the performance of the prediction of microalgae concentration. Experimental result shows a significant improvement in the prediction of microalgae concentration using reconstructed data, the mean squared error (MSE) of the brown tide causal algae Aureococcus anophagefferens is reduced by 85.76%, and the mean squared errors of Chlorella and Synechococcus elongatus is also reduced by 96.52% and 63.76% respectively. This work demonstrates the role of convolutional neural network in 3D-EEM spectral analysis and highlights the feasibility of using 3D-EEM spectral reconstruction data to improve prediction accuracy. [ABSTRACT FROM AUTHOR]

Accurate prediction of PM2.5 concentration is one of the key tasks of air pollution assessment, early warning, and treatment. In this paper, four monitoring sites were arranged in Jiangbei New District of Nanjing City, China. The environmental parameters such as PM2.5/PM10 concentration, temperature, and humidity were monitored from January to February 2020. A gated recurrent unit (GRU) network based on the PM2.5 concentration prediction model was established to predict PM2.5 concentration. The mean relative error (MRE), root mean square error (RMSE), and Pearson correlation coefficient were selected as the evaluation criteria for the accuracy of the GRU model. The data set was divided into a training set, a test set and a validation set at a ratio of 7:2:1, and the GRU model was used to predict the hourly value of PM2.5 concentration in the next week. The prediction results show that the Pearson correlation coefficients between the predicted values and the monitored values of the four monitoring sites have reached more than 0.9, reflecting a strong correlation. The relative average errors are around 10%. The GRU model prediction of NJAU (Nanjing Agricultural University)-Pukou Campus Site is the most accurate, and the correlation coefficient, MRE, and RMSE are 0.970, 7.85%, and 9.6049, respectively, reflecting the good prediction performance of the model. Therefore, this research supports the prediction of air quality in different cities and regions, so people can take protective measures in advance and reduce the damage caused by air pollution to human bodies.

With the aggravation of air pollution, it is very important to realize the prediction of PM2.5 concentration. However, the existing algorithms have the problem of low prediction accuracy. Therefore, a PM2.5 concentration prediction algorithm based on echo state network (ESN) in big data environment is proposed. First, the PM2.5 data of the monitoring base station in Deyang City, Sichuan Province were collected, and the correlation analysis was carried out with the nearest neighboring site. Then, the sequence to sequence (seq2seq) network is used to fuse PM2.5 concentration, meteorological station observation and meteorological element grid data to generate spatiotemporal series data for training and testing, and use ESN to obtain the spatiotemporal characteristics. Finally, the seq2seq model is used to predict the hourly and long-term concentrations of PM2.5 in the future, and the root mean square error (RMSE), mean absolute error (MAE) and mean absolute percentage error (MAPE) are used to evaluate its performance. The experimental results show that, compared with other algorithms, the proposed algorithm can effectively improve the prediction accuracy of PM2.5 concentration in different time periods, and has a high generalization ability by using the prediction method of ESN and Seq2Seq network. [ABSTRACT FROM AUTHOR]

As VOCs pose a threat to human health, it is important to accurately capture changes in VOCs concentrations and sense VOCs concentrations in relevant areas. Therefore, it is necessary to improve the accuracy of VOCs concentration prediction and realise the VOCs aggregation situation sensing. Firstly, on the basis of regional grid division, the inverse distance spatial interpolation method is used for spatial interpolation to collect regional VOCs data information. Secondly, extreme gradient boosting (XGBoost) is used for spatio-temporal feature selection, combined with graph convolutional neural network (GCN) to construct regional spatial relationships of VOCs, and multiple linear regression (MLR) to process VOCs time series data and predict the VOCs concentration in the grid. Finally, the aggregation potential values of VOCs are calculated based on the prediction results, and the potential perception results are visualised. A VOCs aggregation perception method based on concentration prediction is proposed, using the XGBoost-GCN-MLR method with a scenario-aware approach for VOCs to perceive the VOCs aggregation in the relevant region. VOCs concentration prediction and VOCs aggregation trend perception were carried out in Xi’an, Baoji, Tongchuan, Weinan and Xianyang. The results show that compared with the GCN model, XGBoost model, MLR model and GCN-MLR model, the XGBoost-GCN-MLR model reduces the input variables, achieves the optimisation of the input parameters of the VOCs concentration prediction model, reduces the complexity of the prediction model and improves the prediction accuracy. Intelligent sensing of VOCs aggregation can visualise the regional VOCs. The intelligent sensing of VOCs aggregation can visualise the development trend and status of regional VOCs aggregation and convey more information, which has practical value.

• A collaborative and efficient algorithm for detecting mixed gases is constructed. • LSTM-Attention is designed as a backbone network to extract potential features. • Training the model using double loss function, simultaneous completion of two tasks. As an advanced sensor system, electronic nose (E-nose) has been widely used in the field of gas analysis. A novel algorithm that leverages Long Short-Term Memory Attention as a shared framework and integrates it with multi-task learning (MTL-LSTMA) is proposed to enable concurrent prediction of gas category and concentration. Numerous experiments have demonstrated that the MTL-LSTMA model effectively integrates these tasks, fast and simultaneous gas detection for CO, ethylene, and methane gas was achieved (response time of 30 s). All of the classification accuracies exceed 0.98, and the concentration prediction task also exhibits a high degree to match actually. Additionally, we compared results at a variety of response times. It is revealed that MTL-LSTMA model is the best for type identification and concentration prediction of gas mixtures and achieves good results using only the first 30 s of response data. [ABSTRACT FROM AUTHOR]

This study developed a prediction method to determine the distribution of phthalate esters (PAEs) in indoor dust. A gradient boosting decision tree model (GBRT) was trained by using 267 samples in Shanghai, including PAEs concentrations in indoor dust and data obtained from continuous monitoring, as well as the survey of indoor environment. Environmental exposure, residents' lifestyle, and building characteristics data were collected from 8 cities in China. Based on this, the well-trained GBRT model accurately predicted PAEs concentrations, with goodness of fit (R 2 ) > 0.94, mean absolute error (MAE) approaching 0, and mean squared error (MSE) approaching 0. This study identified key relationships between input parameters and PAEs concentrations. The average increment of PAEs concentration was greater than 50 % when using more than 2 electronic devices in bedroom. Diisobutyl phthalate (DiBP) concentration increased by approximately 200 % when cleaning frequency was less than once every fortnight. Bis (2-ethylhexyl) phthalate (DEHP) concentration increased by over 43 % when dampness-related exposure indicators exceeding 3, and by up to 74 % with extensive usage of polyvinyl chloride (PVC) floorings. Furthermore, the study found higher PAEs concentrations in southern China compared to northern cities. • Apply multidisciplinary method to study the factors affecting the differences in indoor phthalate distribution. • Set up a gradient boosted regression trees model to predict the concentration of phthalate and evaluate the result. • Predict and analyze the distribution of phthalate in eight typical cities in China, providing a data support for impacting assessment of human health. • Provide a theoretical support and scientific method for predicting concentration of phthalate in dust in children's bedroom, with a high popularization value. [ABSTRACT FROM AUTHOR]

Plastic waste is a pernicious environmental pollutant that threatens ecosystems and human health by releasing contaminants including di(2-ethylhexyl) phthalate (DEHP) and bisphenol A (BPA). Therefore, a machine-learning (ML)-powered electrochemical aptasensor was developed in this study for simultaneously detecting DEHP and BPA in river waters, particularly to minimize the electrochemical signal errors caused by varying pH levels. The aptasensor leverages a straightforward and effective surface modification strategy featuring gold nanoflowers to achieve low detection limits for DEHP and BPA (0.58 and 0.59 pg/mL, respectively), excellent specificity, and stability. The least-squares boosting (LSBoost) algorithm was introduced to reliably monitor the targets regardless of pH; it employs a layer that adjusts the number of multi-indexes and the parallel learning structure of an ensemble model to accurately predict concentrations by preventing overfitting and enhancing the learning effect. The ML-powered aptasensor successfully detected targets in 12 river sites with diverse pH values, exhibiting higher accuracy and reliability. To our knowledge, the platform proposed in this study is the first attempt to utilize ML for the simultaneous assessment of DEHP and BPA. This breakthrough allows for comprehensive investigations into the effects of contamination originating from diverse plastics by eliminating external interferent-caused influences. [Display omitted] • ML-powered aptasensor monitors DEHP/BPA in rivers minimizing pH-induced errors. • AuNF-modified electrochemical aptasensor can simultaneously detect DEHP/BPA. • LSBoost uses parallel learning and multi-index adjustments for precise predictions. • ML-powered aptasensor assesses trace DHEP/BPA in 12 rivers with varying pH. [ABSTRACT FROM AUTHOR]

Sensor drift is an urgent issue in the machine olfaction community. To date, most studies have focused on gas classification tasks based on an offline method, while neglecting concentration prediction and labeling cost. To permit multitasking including sensor drift, gas classification, concentration prediction, and labeling cost, this paper presents a novel online drift compensation framework based on active learning. Specifically, a Query Strategy for Gas Classification (QSGC) and a Query Strategy for Concentration Prediction (QSCP) are designed respectively, and an Online Domain-adaptive Extreme Learning Machine (ODELM) is proposed. First, the QSGC/QSCP is employed to select the most valuable samples for labeling in the gas classification task/concentration prediction task. Second, the ODELM utilizes only one labeled sample to update the prediction model, and thus adapts to evolving sensor drift. The proposed framework is compared with several state-of-the-art methods. Experimental results demonstrate that the proposed method achieves the best generalization ability with the minimum labeling cost. • An online drift compensation framework for gas sensors is proposed. • Two query strategies are designed to capture drift information. • An online domain-adaptive extreme learning machine is designed to continuously suppress the evolving drift by self-updating. • The framework can effectively handle gas classification and concentration prediction with drift at the lowest labeling cost. [ABSTRACT FROM AUTHOR]

In this paper, an algal identification and concentration determination method based on discrete excitation fluorescence spectra is proposed for online algae identification and concentration prediction. The discrete excitation fluorescence spectra of eight species of harmful algae from four algal categories were assessed. After determining typical excitation wavelengths according to the distribution of photosynthetic pigments and eliminating strongly correlated wavelengths by applying the hierarchical clustering, seven characteristic excitation wavelengths (405, 435, 470, 490, 535, 555, and 590 nm) were selected. By adding the ratios between feature points (435 and 470 nm, 470 and 490 nm, as well as 535 and 555 nm), standard feature spectra were established for classification. The classification accuracy in pure samples exceeded 95%, and that of dominant algae species in a mixed sample was 77.4%. Prediction of algae concentration was achieved by establishing linear regression models between fluorescence intensity at seven characteristic excitation wavelengths and concentrations. All models performed better at low concentrations, not exceeding the threshold concentration of red tide algae outbreak, which provides a proximate cell density of dominant algal species.