Search

Accurate prediction of moisture content is significantly crucial for ensuring process stability and product quality in the cylinder drying process. However, the drying process exhibits complex spatio-temporal characteristics and strong interference, which make accurate prediction challenging for the deep learning approach. To address this issue, this article proposes a new spatio-temporal attention-based bidirectional long-short temporal memory network (STA-BiLSTM) model for accurate moisture content prediction. First, Maximum Relevance Minimum Redundancy (mRMR) is adopted to identify optimal features highly related to moisture content. Secondly, bidirectional long-short temporal memory (Bi-LSTM) network is utilized to extract temporal dependencies from the sequential data. Subsequently, spatio-temporal attention mechanisms are designed to adaptively focus on the most relevant features and timesteps, enhancing the model's generalization ability. Finally, due to the harsh industrial environment, eXtreme Gradient Boosting (XGBoost) is adapted to improve generalizability and robustness. Extensive experiments on a real industrial dataset of the drying process demonstrate that the proposed STA-BiLSTM approach significantly outperforms alternative approaches for predicting moisture content, validating its effectiveness and superiority. [ABSTRACT FROM AUTHOR]

To address the challenge of non-destructive detection of sucrose solution application in the tobacco leaf processing stage, a discrimination model for sucrose solution application based on hyperspectral imaging and machine learning had been developed. Hyperspectral data of tobacco leaf samples with varying sucrose solution applications were first acquired using a visible-shortwave infrared hyperspectral imaging system and preprocessed with standard normal variate (SNV). Four discrimination models for sucrose solution application were then constructed and validated using full-spectrum data and principal component analysis (PCA) reduced data, in conjunction with support vector machine (SVM), logistic regression (LR), multilayer perceptron (MLP), and random forest (RF). The results showed that SNV preprocessing significantly enhanced the feature concentration of the hyperspectral data. When modeling with full-spectrum data, the models in the shortwave infrared band demonstrated significantly higher prediction accuracy compared to those in the visible light band, with the LR model in the shortwave infrared band achieving the highest accuracy of 98.23%. Compared to full-spectrum data modeling, the prediction accuracy of models using the top 10 principal components from PCA reduced data showed little change in the shortwave infrared band, while the RF model's accuracy in the visible light band improved significantly to 71.43%. In the visible light band, the highest accuracy for PCA-reduced data models corresponded to 217, 55, 47, and 59 principal components, while in the shortwave infrared band, the numbers were 13, 11, 117, and 46, respectively. Overall, LR and RF models exhibited superior predictive perf ormance, with the LR model based on PCA-reduced data in the shortwave infrared band maintaining high accuracy with fewer principal components, demonstrating the capability for rapid, non-destructive, and precise determination of sucrose solution application on tobacco leaves. [ABSTRACT FROM AUTHOR]

In order to promote the detection accuracy and separation efficiency of CO2 expanded cut tobacco in blended cut tobacco of cigarette, a separator for separating CO2 expanded cut tobacco from blended cut tobacco was designed by using CO2 expanded tobacco as a tracer and combining the chemical specific gravity method with fluid dynamics. The device mainly consists of a separation system, a collection system, a power system, an automatic liquid level control system and an electric control system. Following the introduction of the separating solution, the cut tobacco sample immersed in the solution was stirred and turned into upper, middle and lower layers under the action of floatation, stratification and filtration due to the density difference between expanded tobacco and other components in the cut tobacco sample. The designed separator was tested with the blended cut tobacco of "Septwolves" brand cigarette provided by china Tobacco Fujian Industrial Limited corporation. The results showed that the separator featured high precision, good accuracy and repeatability. The absolute error was less than 0.25%, the standard deviation ranged from 0.17% to 0.38%, and the coefficient of variation was less than 5%. The separation and detection efficiency increased by more than 80% compared with the manual separation method. This technology supports the promotion of tobacco blending accuracy and uniformity in cigarette production. [ABSTRACT FROM AUTHOR]

The Gardner region is a well‐known shield volcanic complex on the Moon. Its magma origin and formation mechanism are of significant interest but still enigmatic. To reveal the subsurface structure of this volcanic complex, we propose a new 3‐D inversion method of the gravity field based on the regularizations of the L1 norm of the model and its gradients. The model test indicates that our proposed method can recover the density structures with high resolution. Subsequently, we apply it to the Bouguer gravity data in the Gardner region. Our result shows a large, dense body with a volume of about 45 × 45 × 13 km3 centered under the topographic bulge of the Gardner Plateau. We infer that this structure is most likely dense basalts trapped at the crust‐mantle boundary as a sill and acted as the magma reservoir that has fed the volcanic complex in the Gardner area. Plain Language Summary: Volcanism is common on the Moon, which provides an important window into its topography and thermal evolution. The surface topography of the Gardner region displays a series of small shield‐like domes, sinuous rilles, and irregular depressions. It has been identified as one of the eight large volcanic complexes on the Moon. However, little is known about its internal structure, and many speculations exist about its formation. Gravity observation is one of the important means to probe the inner structure. In the Gardner area, a distinct quasi‐circular positive Bouguer gravity anomaly is observed, indicating the presence of subsurface mass surplus. Exploring the cause of the gravity anomaly can provide a basis for revealing the origin of volcanism in this area. However, due to the non‐uniqueness of potential field interpretation, available inversion methods of large‐scale gravity data face the problem of low‐depth resolution. In order to reveal the deep structure of the Gardner area, a new gravity inversion method is developed and applied. The results reveal a large, dense, sill‐like structure under the Gardner plateau. It is very likely the magmatic reservoir of the large shield volcanic complex, which provides profound evidence for confirming the volcanic structure of the Gardner region. Key Points: A new inversion method of the gravity field based on the regularizations of the L1 norm of the model and its gradients is proposedThe practical inversion reveals a large dense sill at the 9–22 km depth centered under the lunar Gardner volcanic complexThe estimated dense sill is most likely the magma chamber of the large shield volcanic structure in the Gardner region [ABSTRACT FROM AUTHOR]

Addressing the complexity of cigarette ingredients and the challenges in brand identification, a cigarette brand identification model based on Projection Pursuit based on Fisher's Criterion (PPF) and hyperspectral technology was proposed. Firstly, the PPF projection algorithm was employed to conduct spectral similarity analysis on the original spectra as well as those processed with Multiplicative Scatter Correction (MSC), first-order derivative (1st D), and second-order derivative (2ndD). Secondly, feature wavelength selection was carried out using Successive Projections Algorithm (SPA) and Genetic Algorithm (GA) to enhance the classification performance in subsequent steps. Finally, cigarette brand identification models were established using Support Vector Machine (SVM) and Extreme Gradient Boosting (XGBoost) algorithms, and the personality of the model was evaluated. The results indicated that, in the combination of hyperspectral technology and the PPF projection algorithm, the second-order derivative (2nd D) served as the optimal band processing method, significantly enhancing the spectral separability. The recognition model established based on 2ndD-SPA-SVM emerged as the best model, achieving an overall classification accuracy of 94. 58% and 92. 50% for the training and test sets respectively. Moreover, a greater difference in the proportion of leaf silk and expanded silk among different brand formulations led to a lower similarity value between categories and a higher recognition accuracy. This model can provide a new efficient, fast, accurate, and non-destructive classification and discrimination method for cigarettes of different grades. It offered theoretical support for the application of hyperspectral technology in cigarette brand maintenance, tobacco blend formulation design, and other scenarios. [ABSTRACT FROM AUTHOR]

Aiming at the problems of the large storage, complex composition, low comprehensive utilization rate, and high environmental impact of coal gangue, this paper carried out experimental research on the preparation of iron oxide red from high-iron gangue by calcination activation, acid leaching, extraction, and the hydrothermal synthesis of coal gangue. The experimental results show that when the calcination temperature of coal gangue is 500 °C, the calcination time is 1.5 h, the optimal concentration of iron removal is 6 mol/L, the acid leaching temperature is 80 °C, the acid leaching time is 1 h, and the liquid——solid mass ratio is 4:1; the iron dissolution rate can reach 87.64%. A solvent extraction method (TBP-SK–hydrochloric acid system) was used to extract the leachate, and a solution with iron content up to 99.21% was obtained. By controlling the optimum hydrothermal conditions (pH = 9, temperature 170 °C, reaction time 5 h), high-purity iron oxide red product can be prepared; the yield is 80.07%. The red iron oxide was characterized by XRD, SEM-EDS, particle-size analysis, and ICP-OES. The results show that the red iron oxide peak has a cubic microstructure, an average particle size of 167.16 μm, and a purity of 99.16%. The quality of the prepared iron oxide red product meets the requirement of 98.5% of the "YHT4 Iron oxide Standard for ferrite". It can be used as a raw material to produce high-performance soft magnetic ferrite. In summary, this experimental study on the preparation of iron oxide red from coal gangue is of great significance for the comprehensive utilization of coal gangue to realize the sustainable development of the environment and economy. [ABSTRACT FROM AUTHOR]

Due to traditional dynamic programming approach is only suitable for constant velocity motion target or slow maneuvering target，an algorithm using interacting multiple model(IMM) filter incorporated with track-before-detect approach is proposed for maneuvering target detection and tracking. The algorithm is applied to short-range millimeter rader detection. According to the actual movement characteristics of target, constant acceleration motion model, constant turn motion model and constant velocity motion model are established. The transition and prediction between the motion models are also taken into account based on dynamic programming algorithm, so as to avoid the problem of poor tracking effect caused by model mismatch. Simulation results show that compared with the traditional detection algorithm, the proposed algorithm can detect and track maneuvering targets more effectively, which is beneficial to engineering applications.

In the task of sales forecasting of new clothing products, the lack of historical sales data often necessitates the full utilization of data from other modalities as a supplement. However, multi-modal clothing data are usually redundant and heterogeneous. To solve the problems, a hierarchical multi-modal attention based recurrent neural network ( HMA-RNN) including three main elements is proposed. The hierarchical structure separates high-level semantic information from low-level semantic information to avoid information redundancy. The multi-modal attention ( MMA) is introduced in the fusion stage to mitigate inherent data non-alignment. The shared attention mechanism is utilized to build the dependencies across the multi-modal data. Experimental results on the Visuelle 2.0 dataset show that the proposed approach achieves promising results with 72.07 on the weighted average percentage error (WAPE) and 0. 80 on the mean absolute error ( MAE), outperforming existing works significantly, which indicates the effectiveness of the proposed approach. [ABSTRACT FROM AUTHOR]