Search

Wind energy is attracting increasing attention with its sustainability and cleanliness. However, owing to the volatility and intermittency of wind speed, it is challenging to establish a scientific and reliable forecasting system. Most research has mainly been based on simple data preprocessing, single objective optimization, and point prediction, which may lead to poor forecasting performance. Hence, in this study, an innovative wind speed forecasting system is developed that incorporates effective data preprocessing and a novel algorithm. In order to alleviate the complexity and chaos of a wind speed series, a fuzzy data preprocessing scheme is designed based on “decomposition and ensemble” and a fuzzy time series. Following this, a multi-objective imperialist competitive algorithm (MOICA) is proposed and applied for optimizing an extreme learning machine (ELM), and a corresponding hybrid predictor MOICA-ELM is conducted for wind speed forecasting. For further investigation the uncertainty of wind speed, both point and interval forecasting are employed in this system. Simulation results on four wind speed datasets collected from two wind farms in China are in good accordance with the empirical data with multiple criterion and scientific evaluation; these results and show a good performance of the proposed system in terms of accuracy and stability.

Short-term electrical load forecasting is an important part in the management of electrical power because electrical load is an extreme, complex non-linear system. To obtain parameter values that provide better performances with high precision, this paper proposes a new hybrid electrical load forecasting model, which combines ensemble empirical mode decomposition, extreme learning machine, and grasshopper optimization algorithm for short-term load forecasting. The most important difference that distinguishes this electrical load forecasting model from other models is that grasshopper optimization can search suitable parameters (weight values and threshold values) of extreme learning machine, while traditional parameters are selected randomly. It is applied in Australia electrical load prediction to show its superiority and applicability. The simulation studies are carried out using a data set collected from five main states (New South Wales, Queensland, Tasmania, South Australia and Victoria) in Australia from February 1 to February 27, 2018. Compared with all considered basic models, the proposed hybrid model has the best performance in predicting electrical load.

Nowadays, petrochemical industries with many integrated units and equipment have characteristics of high uncertainty and nonlinearity. Therefore, it becomes more and more difficult to make reliable and accurate point measurement of energy modeling. To tackle this problem, a novel prediction intervals (PIs) method integrating error & self-feedback extreme learning machine (ESF-ELM) with particle swarm optimization (PSO) is proposed. For improving the energy modeling accuracy of extreme learning machine (ELM), the input weights are initialized using cosine similarity coefficients but not randomly initialized. In addition, an error-feedback layer and a self-feedback layer are added to the input layer and the hidden layer for enhancing generalization performance, respectively. Finally, PSO with a comprehensive measure is developed to evaluate the mean coverage probability and the mean width percentage of PIs. The proposed ESF-ELM with PSO is applied to constructing PIs of energy consumption for a Purified Terephthalic Acid production process. Simulation results show the proposed model can generate high-quality PIs with large coverage probability, narrow width, and superiority in adaptability and reliability, which provides guidance for decision makers to maximize benefits and give reasonable future plans.

Accurate and reliable multi-step wind speed forecasting is extremely crucial for the economic and safe operation of power systems. A novel dynamic hybrid model, which combines an adaptive secondary decomposition (ASD), a leave-one-out cross-validation-based regularized extreme learning machine (LRELM) and the backtracking search algorithm (BSA), is proposed to mitigate the practical difficulties of the traditional decomposition-ensemble forecasting models (DEFMs) through adaptive dynamic decomposing and modeling when new data is added. The new ASD method, which fuses ensemble empirical mode decomposition (EEMD), adaptive variational mode decomposition (AVMD) with sample entropy (SE), is developed for smoothing the raw series to reduce computational time as well as enhance generalization and stability of forecasting models. BSA is employed to optimize LRELM to overcome the drawback of instability. To validate its efficacy, the proposed model and thirteen benchmark models are compared by diverse lead-time forecasting of several real cases. Comprehensive comparisons with a coherent set of indices suggest that the proposed model is an effective and powerful tool for short-term wind speed forecasting not only from the perspective of reliability and sharpness but also from the view of overall skills.

With the increasing complexity of energy modeling data, it becomes more and more demanding to build a robust and accurate energy analysis model using a single neural network. To deal with this problem, a novel robust ensemble model integrated extreme learning machine with multi-activation functions is proposed to develop robust and accurate energy analysis models. There are two salient features in the proposed model: one is that different effective nonlinear activation functions are adopted in extreme learning machine to enhance the ability in dealing with the high nonlinearity of energy modeling data, i.e. multi-activation functions are utilized; the other salient feature is that several single models with different effective nonlinear activation functions are combined to build an ensemble model for enhancing the performance in terms of accuracy and stability, i.e. the generalization and robustness capability of the proposed model is much improved through aggregating multiple activation functions based extreme learning machine models. To verify the performance of the proposed model, two case studies of developing energy analysis models for complex chemical processes are carried out. Simulation results demonstrate that the proposed model achieves high accuracy and good stability.

Battery health monitoring and management is critically important for electric vehicle performance and economy. This paper presents a multiple health indicators-based and machine learning-enabled state-of-health estimator for prognostics and health management. The multiple online health indicators without the influence of different loading profiles are used as effective signatures of the health estimator for effective quantification of capacity degradation. An extreme learning machine is introduced to capture the underlying correlation between the extracted health indicators and capacity degradation to improve the speed and accuracy of machine learning for online estimation. The proposed estimator is also compared to the traditional BP neural network. The associated results indicate that the maximum estimation error of the proposed health management strategy is less than 2.5%, and it has better performance and faster speed than the BP neural network.

One of the key issues of the sustainable development in all countries is industrial productivity improvement, especially the production capacity improvement and energy-saving of complex chemical processes. Therefore, this paper proposes a production prediction and energy-saving model based on Extreme Learning Machine (ELM) integrated Interpretative Structural Modeling (ISM) and Analytic Hierarchy Process (AHP). The factors that affect the productivity are divided into different levers by the ISM. And then the attributes of each layer are fused by the AHP based on the entropy weight, which greatly reduces the complexity of the input attributes. Moreover, the production prediction and energy-saving model is established based on the ELM. Compared with the traditional ELM, the validity and the practicability of the proposed method are verified by University of California Irvine (UCI) datasets. Finally, the proposed method is applied in the production capacity prediction and energy-saving of ethylene production systems and Purified Terephthalic Acid (PTA) production systems. The experimental results show the proposed method could reduce the number of hidden layer nodes and improve the training time of the ELM. Furthermore, the prediction accuracy of the ethylene production and the PTA production reaches about 99% to improve the energy efficiency of complex chemical processes.

Air conditioning system is extensively used in large commercial buildings. The fast and accurate building cooling load forecasting is the basis for improving the operation efficiency of the air conditioning system, which is conducive to implement the effective management of the air conditioning system. Therefore, a hybrid prediction model based on random forest-improvement parallel whale optimizing-extreme learning machine neural network (RF-IPWOA-ELM) is proposed to predict the cooling load of large commercial buildings. First, the influence of different parameters on the cooling load is analyzed, and the random forest (RF) method is used to extract the parameters with high degree of influence as the input variables of prediction model. Then, the extreme learning machine (ELM) optimized by the improved parallel whale optimization algorithm (IPWOA) is established to predict. Finally, a simulation experiment is carried out using measured data of two large commercial buildings in north of China. The experimental results show that the root mean square error (RMSE) and mean average percentage error (MAPE) of RF-IPWOA-ELM predicting the cooling load for these two buildings are 2.8735, 0.2% and 4.7721, 0.45%, respectively. Compared with the other prediction model, the RMSE and MAPE of this model are reduced by 66.17%–90.62%, 81.48%–95.79% and 71.91%–84.40%, 74.14%–86.15%, respectively, which has higher prediction accuracy. Simultaneously, for different prediction models, RF-IPWOA-ELM has a shorter prediction time, which presents superiority in time complexity. And when there are few training samples, RF-IPWOA-ELM can still effectively predict the cooling load in different months, indicating that it possesses strong generalization ability. Therefore, the proposed hybrid model can be used as a reliable tool for cooling load prediction in the energy conservation of air conditioning system and energy management.

Current electricity price forecasting models rely on only simple hybridizations of data preprocessing and optimization methods while ignoring the significance of adaptive data preprocessing and effective optimization and selection strategies to obtain optimal models that improve the forecasting performance. To solve these problems, this study develops an improved electricity price forecasting model that offers the advantages of adaptive data preprocessing, advanced optimization method, kernel-based model, and optimal model selection strategy. Specifically, the adaptive parameter-based variational mode decomposition technology is proposed to provide desirable data preprocessing results, and a leave-one-out optimization strategy based on the chaotic sine cosine algorithm is proposed and applied to develop optimal kernel-based extreme learning machine models. In addition, a newly proposed optimal model selection strategy is applied to determine the developed model that provides the most desirable forecasting result. Numerical results show that the developed model's performance metrics were best, and the average values of mean absolute error, root mean square error, mean absolute percentage error, index of agreement, and Theil's inequality coefficient in four datasets are 0.5121, 0.7607, 0.5722%, 0.9997 and 0.0041, respectively, which imply that the developed model is a promising, applicable and effective electricity price forecasting technique in the real electricity market.

Wind energy has attracted much attention because it is sustainable and renewable energy with a much smaller impact on the environment than fossil fuels. Accurate wind speed forecasting can pave the way to the goal of sustainable development. However, many papers only use wind speed series but ignore weather factors, which lead to poor forecasting results. In this paper, we propose an ensemble system that considers historical wind speed series and other weather factors to make wind speed forecasting. The proposed system integrates noise reduction, clustering, and multi-objective optimization. Firstly, the historical wind speed series are decomposed by complementary ensemble empirical mode decomposition (CEEMD). Secondly, the samples are clustered by self-organizing map (SOM) according to the weather factors. Finally, a regularized extreme learning machine (RELM) model is trained to forecast the wind speed for each cluster, and multi-objective grey wolf optimizer (MOGWO) is employed to optimize the parameters in the model. An empirical study using three datasets from the M2 tower of the national wind power technology center (NWTC) of the national renewable energy laboratory (NREL) illustrates that the proposed system performs better than other comparative models in terms of four performance indicators.

The state of charge (SOC) plays a crucial role in battery management system, which directly reflects the usage of the battery. Recently the extreme learning machine (ELM) model as a data-driven method has been utilized to estimate SOC due to its simple single hidden layer structure and fast learning performance. The battery management system, however, may usually work in complex working conditions, which means that the non-Gaussian complex noise (or outliers) interference problem may exist in some measurement data for model training. So the performance of the classical ELM with mean square error (MSE) criterion may be degraded under this case. This work considers Non-Gaussian noise interference issue, the MSE in ELM is substituted by mixture generalized maximum correntropy criterion (MGMCC), and a novel robust ELM model is developed to improve the SOC estimation capability which mainly relies on the stable and robust nonlinear similarity characteristics of the MGMCC. A data set from a Panasonic 18,650 battery cell is used to verify the robustness of the proposed model, the experiment results demonstrate that it can achieve better estimation performance in terms of different evaluation metrics compared with the traditional methods.

The excessive emission of carbon is one of the important factors causing environmental pollution, and the prediction of carbon trading market price is an important mean of emission reduction. In order to accurately predict the carbon price, a new carbon price prediction model is proposed in this paper. Firstly, the data is decomposed into multiple intrinsic mode functions (IMFs) by optimized variational mode decomposition (OVMD). Secondly, the complexity of IMFs is analyzed by spatial-dependence recurrence sample entropy (SdrSampEn). Thirdly, the IMFs with higher complexity are integrated and decomposed by complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) to get high complexity IMFs. Then, particle swarm optimized extreme learning machine (PSOELM) is used to predict the high complexity IMFs, and extreme learning machine (ELM) is used to predict other. Finally, the predicted value is reconstructed to complete the prediction. In this paper, OVMD is proposed to solve the selection of decomposition layers K by variational mode decomposition (VMD) from the perspective of variance contribution rate. Through the experimental results, the effectiveness of the proposed model is verified, and it can be used to predict the supply and demand of carbon market and evaluate the effectiveness of current carbon trading policies.

The safe and stable operation of wind power systems requires the support of wind speed prediction. To ensure the controllability and stability of smart grid dispatching, a novel hybrid model consisting of data-adaptive decomposition, reinforcement learning ensemble, and improved error correction is established for short-term wind speed forecasting. In decomposition module, empirical wavelet transform algorithm is used to adaptively disassemble and reconstruct the wind speed series. In ensemble module, Q-learning is utilized to integrate gated recurrent unit, bidirectional long short-term memory, and deep belief network. In error correction module, wavelet packet decomposition and outlier-robust extreme learning machine are combined to developing predictable components. An appropriate correction shrinkage rate is used to obtain the best correction effect. Ljung-Box Q-Test is utilized to judge the termination of the error correction iteration. Four real data are utilized to validate model performance in the case study. Experimental results show that: (a) The proposed hybrid model can accurately capture the changes of wind data. Taking 1-step prediction results as an example, the mean absolute errors for site #1, #2, #3, and #4 are 0.0829 m/s, 0.0661 m/s, 0.0906 m/s, and 0.0803 m/s, respectively; (b) Compared with several state-of-the-art models, the proposed model has the best prediction performance.

Advances in energy-saving technology is main way to achieve carbon neutrality. With the development of digital twin, building the physical-virtual data space for improving energy management capacity of enterprises has received tremendous attention. The energy behaviour model implementing accurate simulation and prediction of energy state is the core meta-model of energy-efficient manufacturing digital twin (EMDT). The widely used state-based energy modelling assumes constant power in operation state and approximately fits the energy behaviour without considering uncertain operation environment, resulting in energy behaviour distortion. A data-driven hybrid petri-net (DDHPN) inspired by both the state-based energy modelling and machine learning was developed for establishing the energy behaviour meta-model. Gaussian kernel extreme learning machine is proposed to fit the instantaneous firing speed of energy consumption continuous transitions in DDHPN. DDHPN-based energy behaviour model is driven by physical data under real-time working conditions, operating parameters, and production load for generating a virtual data space of energy management. Finally, DDHPN was integrated into the EMDT model using unified modelling language. The application in extrusion process and die casting process show that the presented model has higher accuracy in energy behaviour prediction. Furthermore, a digital-twin-based energy management prototype system for extrusion workshop demonstrates its potential.

Inspired by the interesting idea of randomization, some powerful but time-consuming decomposition-ensemble learning paradigms can be extended into extremely efficient and fast variants by using randomized algorithms as individual forecasting tools. In the proposed methodology, Three major steps, (1) data decomposition via ensemble empirical mode decomposition, (2) individual prediction via a randomized algorithm (using randomization to mitigate training time and parameter sensitivity), and (3) results ensemble to produce final prediction, are included. Different from other existing decomposition-ensemble models using traditional econometric approaches or computational intelligence methods in individual prediction, this study employs some emerging randomized algorithms—extreme learning machine, random vector functional link network (using randomly fixed weights and bias in neural networks), and random kitchen sinks (using randomly mapping features to approximate kernels)—to dramatically save computational time and enhance prediction accuracy. With the Brent oil prices and the Henry Hub natural gas prices as studying samples, the empirical study statistically confirms that the proposed randomized-algorithm-based decomposition-ensemble learning models are proved to be excellently efficient and fast, relative to popular single techniques (including computational intelligence methods and randomized algorithms) and similar decomposition-ensemble counterparts (using the aforementioned single techniques as individual forecasting tools).

For district heating systems, prediction of the heat load is a very important topic for energy storage and optimized operation. For large and complex heating systems, most prediction models in previous publications only considered the influence of outdoor temperature, whereas the indoor temperature and thermal inertia of buildings were not included. For an energy-efficient residential building in Shijiazhuang (China), the heat load prediction is investigated using various prediction models, including a wavelet neural network (WNN), extreme learning machine (ELM), support vector machine (SVM) and back propagation neural network optimized by a genetic algorithm (GA-BP). In these models, the indoor temperature and historical loads are considered as influencing factors. It is found that the prediction accuracies of the ELM and GA-BP are slightly higher than that of WNN, so the ELM and GA-BP models provide feasible methods for the heat load prediction. The SVM shows smaller relative errors in the model prediction compared with three neural network algorithms.

An accurate energy prediction and optimization model plays a very important role in the petrochemical industries. Due to the imbalanced and uncompleted characteristics of complex petrochemical small data, it is a big challenge to build accurate prediction and optimization models for energy analysis. In order to solve this problem, a nonlinear interpolation virtual sample generation method integrated with extreme learning machine is proposed. Well virtual input and output variables can be generated through interpolation of the hidden layer outputs of extreme learning machine. The generated virtual samples are put together with the original samples to train models for enhancing accuracy performance. To validate the effectiveness of the proposed nonlinear interpolation virtual sample generation method, a standard function is firstly selected, and then the proposed nonlinear interpolation virtual sample generation method is applied to developing a model of energy analysis for ethylene production systems. Simulation results showed that the prediction accuracy could be significantly improved, which provided helpful guidance for production departments and government to achieve the goal of energy management of petrochemical industries.

The prediction of crude oil prices has important research significance. The paper contributes to the literature of hybrid models for forecasting crude oil prices. We apply ensemble empirical mode decomposition (EEMD) to decompose the residual term (RES), which contains complex information after variational mode decomposition (VMD), further combining with a kernel extreme learning machine (KELM) optimized by particle swarm optimization (PSO) to construct the VMD-RES.-EEMD-PSO-KELM model. In order to verify the validity of the model, this paper conducts empirical analyses of Brent crude oil and West Texas Intermediate (WTI) crude oil. The empirical results show that the prediction model proposed in this paper improves the prediction accuracy of crude oil prices.

Accurate photovoltaic power prediction is important to ensure the safety, stability, and economic operation of the power system after high photovoltaic demand. However, due to the intermittent and stochastic volatility characteristics of photovoltaic power, it is difficult to build satisfactory photovoltaic power prediction models. In this study, a similar day-based ultrashort-term multi-step ahead prediction model of photovoltaic power is proposed, which combines enhanced colliding bodies optimization (ECBO), variational mode decomposition (VMD), and a deep extreme learning machine (DELM). First, a new training data selection method based on grey correlation analysis and a Pearson correlation coefficient is proposed to find days that are similar to the predicted day. Second, the variational mode decomposition, based on enhanced colliding bodies optimization, is used to decompose the original photovoltaic power data into an ensemble of components with different amplitudes and frequencies, in which a new fitness function is designed based on a weighted-permutation entropy algorithm. Finally, the deep extreme learning machine network is employed to complete the prediction of each component, and the enhanced colliding bodies optimization algorithm is utilized again to obtain the optimal neuron numbers in hidden layers of the deep extreme learning machine; the ultimate prediction results of the photovoltaic power is obtained by reconstructing the prediction value of each component. The proposed model is tested with data from a real-world photovoltaic power station located in Xinjiang, China. The experimental results show that: (a) among all the competing models, the proposed model can achieve the highest multi-step prediction accuracy; (b) a qualitative breakthrough on the prediction accuracy improvement has been made by the proposed model in the 1-step and 2-step ahead prediction; (c) only a few parameters need to be tuned manually in the whole prediction process; the proposed model has good intelligence capabilities and can be easily applied and popularized in the practical projects.

Energy-saving and emission reduction are crucial to modern society, but the previous resource optimization methods based on traditional neural networks are complex and low accuracy. Therefore, this paper presented a novel extreme learning machine (ELM) method based on t-distributed stochastic neighbor embedding (t-SNE) to optimize the energy and reduce the carbon emission. In terms of mapping high-dimensional production data to low-dimensional space, the t-SNE can deal with the major factors affecting the energy efficiency, which are taken as inputs of the ELM. Then the resource optimization model for energy-saving is obtained based on the ELM to predict the output and achieve the optimal configuration. Finally, the proposed method is used to establish the resource optimization model for the ethylene and purified terephthalic acid (PTA) production processes in complex industrial processes. The experimental results demonstrate that the proposed model can improve the prediction accuracy of resource optimization models of complex industrial processes, and realize energy-saving and carbon emissions reduction.

In this study, kernel-based extreme learning machine (K-ELM) and artificial neural network (ANN) models were developed in order to predict the conditions of an alkaline-catalysed transesterification process. The reliability of these models was assessed and compared based on the coefficient of determination (R2), root mean squared error (RSME), mean average percent error (MAPE) and relative percent deviation (RPD). The K-ELM model had higher R2 (0.991) and lower RSME, MAPE and RPD (0.688, 0.388 and 0.380) compared to the ANN model (0.984, 0.913, 0.640 and 0.634). Based on these results, the K-ELM model is a more reliable prediction model and it was integrated with ant colony optimization (ACO) in order to achieve the highest Ceiba pentandra methyl ester yield. The optimum molar ratio of methanol to oil, KOH catalyst weight, reaction temperature, reaction time and agitation speed predicted by the K-ELM model integrated with ACO was 10:1, 1 %wt, 60 °C, 108 min and 1100 rpm, respectively. The Ceiba pentandra methyl ester yield attained under these optimum conditions was 99.80%. This novel integrated model provides insight on the effect of parameters investigated on the methyl ester yield, which may be useful for industries involved in biodiesel production.

Extreme learning machine (ELM), which is a simple single-hidden-layer feed-forward neural network with fast implementation, has been widely applied in many engineering fields. However, it is difficult to enhance the modeling ability of extreme learning in disposing the high-dimensional noisy data. And the predictive modeling method based on the ELM integrated fuzzy C-Means integrating analytic hierarchy process (FAHP) (FAHP-ELM) is proposed. The fuzzy C-Means algorithm is used to cluster the input attributes of the high-dimensional data. The Analytic Hierarchy Process (AHP) based on the entropy weights is proposed to filter the redundant information and extracts characteristic components. Then, the fusion data is used as the input of the ELM. Compared with the back-propagation (BP) neural network and the ELM, the proposed model has better performance in terms of the speed of convergence, generalization and modeling accuracy based on University of California Irvine (UCI) benchmark datasets. Finally, the proposed method was applied to build the energy saving and predictive model of the purified terephthalic acid (PTA) solvent system and the ethylene production system. The experimental results demonstrated the validity of the proposed method. Meanwhile, it could enhance the efficiency of energy utilization and achieve energy conservation and emission reduction.

Energy optimization and analysis of complex chemical processes play a significant role in the sustainable development procedure. In order to deal with the high-dimensional and noise data in complex chemical processes, we present an energy optimization and analysis method based on extreme learning machine integrating the index decomposition analysis. First, index decomposition analysis has been used to decompose the high-dimensional data to three energy performance indexes of the activity effect, the structure effect and the intensity. And then, those indexes and the production/conductivity of the chemical process are defined as inputs and outputs of the extreme learning machine respectively to build energy optimization and analysis model. Finally, the proposed method has been applied to optimizing and analyzing energy status of the ethylene system and the purified terephthalic acid solvent system in complex chemical processes. The experiment results show that the proposed method has the characteristics of fast learning, stable network outputs and high model accuracy in handling with the high-dimensional data. Moreover, it can optimize energy of chemical processes and guide the production operation. In our experiment, the production of ethylene plants can be increased by 5.33%, and the conductivity of purified terephthalic acid plants can be reduced by 0.046%.

© 2016 Elsevier Ltd The ultimate issue in electricity loads modelling is to improve forecasting accuracy as well as guarantee a robust prediction result, which will save considerable manual labor material resources and economic consumption. For addressing this challenge, many researchers are committed to investigating highly accurate forecasting models, and feature selection (FS) technologies are considered as a powerful tool to improve performance of models in many literature. However, common FS technologies applied for Short-term load forecasting (STLF) ignore to select date information of the observed series as feature candidates and pay less attention to reduction rates of feature candidates, which will result in loss of date information and redundancy of features. Both drawbacks provide a significant roadblock for improving forecasting accuracy. Aiming to overcome both drawbacks and develop an effective model for STLF, this paper successfully investigates the date-framework strategy (DFS) to construct the pool of features and develops an FS technology, genetic algorithm binary improved cuckoo search (GABICS), to search a solution with the lowest reduction rate. Assigning the extreme learning machine (ELM) to be the forecast, GABICS-DFS-ELM not only obtains a minimum and effective subset of features but also has a satisfactory forecasting result with high accuracy and robustness.

Accurate health status estimation and capacity prediction of lithium-ion batteries are important means to prevent a series of problems such as capacity loss, driving range and safety accidents caused by the aging of batteries. Research on battery capacity prediction based on constant discharge rate has become increasingly mature. However, as the main power source for electric vehicles, discharge current of lithium-ion battery is constantly changed by the influence of time-varying vehicle speed. Considering the effect of random variable current (RVC) discharge on battery capacity degradation, a novel predicting method for circulating capacity of lithium-ion battery is proposed. Firstly, features are extracted from the battery charging and discharging process. Secondly, the correlation between features and battery capacity is analyzed by using the grey relational analysis, and features with the higher correlation coefficient are selected as final health features. Thirdly, the online sequential extreme learning machine optimized by beetle antenna search is proposed and used to predict capacity of lithium-ion battery. Experimental results show that the minimum battery capacity RMSE predicted is 1.0294, and the cycle capacity error is mostly within the range of -3mAh∼3mAh, which proves that the method can more accurately estimate the capacity of lithium-ion batteries under RVC conditions.

The tight coupling of information and communication technology and traditional energy system has given birth to a cyber physical energy system (CPES). CPES indeed improves the economic operation and control efficiency of the energy system, but it also brings new cyber risk issues, threatening the secure operation of the energy system. Consequently, this paper proposes a new multivariate ensemble classification (MEC) method to detect intrusions in CPES, thereby enhancing the baseline cybersecurity of CPES. MEC simultaneously takes into account the detection accuracy, stability and computing efficiency. In MEC, extreme gradient boosting, light gradient boosting machine and extreme learning machine are separately designed as individual detectors for intrusion identification. Then, ensemble learning based decision-making is developed to strategically aggregate the results of all individual detectors. Finally, the effectiveness of the proposed MEC is validated on IEEE standard 14-, 57- and 118-bus systems. The obtained results demonstrate that the MEC method has an attractive potential in real applications.

Nowadays with the growing requirements of sustainable development, energy efficiency enhancement plays a crucial role in the petrochemical industries. In fact, the energy efficiency level depends on the coordination management of energy and materials. Therefore, a two-stage optimization scheme with respect to coordination of energy and materials from the entire process to the key sub-process is proposed for improving energy efficiency level of ethylene production. Firstly, a data-driven model for feedstock and plant-level economic indicators and is established by extreme learning machine (ELM) integrated domain adaptive manifold regularization (DAMR) and particle swarm optimization (PSO), called DAMR-PSO-ELM. Secondly, a multi-objective optimization model for feedstock proportions is established and a cascading-priority weight strategy is made for obtaining the optimal feedstock proportions under the different market demands. Finally, fuel–feedstock ratio optimization models for cracking furnaces are built with respect to the obtained optimal proportions, and an improved elitist teaching–learning-based optimization algorithm is proposed to acquire the optimal fuel. The effectiveness and feasibility of the proposed coordinated optimization scheme are validated from a practical ethylene plant, results show that energy consumption of cracking production is reduced by 4.89% and energy efficiency of the entire ethylene process is increased by 6.82% on average.

Accurate estimation of battery state-of-health (SOH) is of great importance for ensuring the safety and reliability of battery energy storage system. Due to the complicated degradation mechanism of batteries, the transfer application of SOH estimation for different types of the batteries with unknown usage levels is challenging. To solve this issue, a novel metabolic extreme learning machine (MELM) framework for SOH estimation is proposed in this study. A degradation state model based on the extreme learning machine (ELM) is developed to describe the complex battery degradation mechanism, and the established model can map the relationship between the degradation features and the degradation dynamics for different batteries. To realize SOH estimation at different usage levels with a few data, the metabolic mechanism is introduced to update the input of the degradation state model and reflect the latest trend of degradation. To reduce the errors caused by the metabolism, the grey model is adopted to extrapolate the trend of error accumulation and correct the estimation results. The prominent performances of the MELM framework are synthetically verified from different aspects, the results indicate the MELM framework can effectively realize the SOH estimation for different types of batteries with unknown usage levels.