Number Of Neurons In Layer Research Articles

Prediction of monthly average and extreme atmospheric temperatures in Zhengzhou based on artificial neural network and deep learning models

IntroductionAtmospheric temperature affects the growth and development of plants and has an important impact on the sustainable development of forest ecological systems. Predicting atmospheric temperature is crucial for forest management planning.MethodsArtificial neural network (ANN) and deep learning models such as gate recurrent unit (GRU), long short-term memory (LSTM), convolutional neural network (CNN), CNN-GRU, and CNN-LSTM, were utilized to predict the change of monthly average and extreme atmospheric temperatures in Zhengzhou City. Average and extreme atmospheric temperature data from 1951 to 2022 were divided into training data sets (1951–2000) and prediction data sets (2001–2022), and 22 months of data were used as the model input to predict the average and extreme temperatures in the next month.Results and DiscussionThe number of neurons in the hidden layer was 14. Six different learning algorithms, along with 13 various learning functions, were trained and compared. The ANN model and deep learning models were evaluated in terms of correlation coefficient (R), root mean square error (RMSE), and mean absolute error (MAE), and good results were obtained. Bayesian regularization (trainbr) in the ANN model was the best performing algorithm in predicting average, minimum and maximum atmospheric temperatures compared to other algorithms in terms of R (0.9952, 0.9899, and 0.9721), and showed the lowest error values for RMSE (0.9432, 1.4034, and 2.0505), and MAE (0.7204, 1.0787, and 1.6224). The CNN-LSTM model showed the best performance. This CNN-LSTM method had good generalization ability and could be used to forecast average and extreme atmospheric temperature in other areas. Future climate changes were projected using the CNN-LSTM model. The average atmospheric temperature, minimum atmospheric temperature, and maximum atmospheric temperature in 2030 were predicted to be 17.23 °C, −5.06 °C, and 42.44 °C, whereas those in 2040 were predicted to be 17.36 °C, −3.74 °C, and 42.68 °C, respectively. These results suggest that the climate is projected to continue warming in the future.

Neuron-specific loss of Ppp6c induces neonatal death and decreases the number of cortical neurons and interneurons

Protein phosphatase 6 (PP6) is a Ser/Thr protein phosphatase with the catalytic subunit Ppp6c. Recent cell-level studies have revealed that Ppp6c knockdown suppresses neurite outgrowth, suggesting that Ppp6c is involved in the development of the nervous system. We found that the function of PP6 in neurons is essential for mouse survival after birth, as all neural-stem-cell-specific KO (Ppp6cNKO) and neuron-specific KO mice died within 2 days of birth. By contrast, approximately 40 % of oligodendrocyte-specific KO mice died within 2 days of birth, whereas others survived until weaning or later, suggesting that the lethality of PP6 loss differs between neurons and oligodendrocytes. Furthermore, the fetal brain of Ppp6cNKO mice exhibited decreased numbers of neurons in layers V-VI and interneurons in layer I of the neocortex. These results suggest for the first time that Ppp6c is essential for neonatal survival and proper development of neurons and interneurons in the neocortex.

Predicting CAR-T Response in Mantle Cell Lymphoma Using Deep Neural Network Models on Single-Cell RNA Sequencing Data

Introduction The approval of CD19 chimeric antigen receptor (CAR) T therapy by the FDA for relapsed mantle cell lymphoma (MCL) represents a significant advancement. However, patients are resistant or experience relapse after receiving CAR T therapy. High degree of transcriptomic heterogeneities among MCL tumor cells makes it challenging to predict response to CAR-T therapy. In this study, we employed deep learning models in conjunction with single-cell RNA sequencing data to predict the response to CAR-T therapy in MCL patients. Methods In this study, we collected and analyzed 39 patient samples treated with CAR-T therapy, of which 30 were sensitive and 9 were resistant to the treatment. Single-cell RNA sequencing was performed on these samples, and downstream analysis was conducted using the Seurat R package (version 4.3.0.1), including dimension reduction and cell type annotation. B cells were isolated in-silico, followed by differential gene expression analysis using the Wixcoxon rank sum test. The identified differentially expressed genes were then used as inputs for deep learning neural networks to predict the response to CAR-T therapy (Figure 1A). We applied a multilayer perceptron (MLP) network as the binary classifier for response prediction. The model consists of four fully connected layers, including one input layer, two intermediate hidden layers, and one output layer. The number of neurons in the input layer corresponds to that of the differentially expressed genes. The two hidden layers have 64 and 16 neurons, respectively, each followed by one activation layer using rectified linear unit (ReLU) and one dropout layer with dropout rate of 0.4. The features from the latter hidden layer were extracted for embedding plots. The output layer has two neurons with activation function of Softmax, each neuron representing the probability of categorical patient response labels. A leave-one-sample-out approach was employed to calculate the prediction accuracy. Results A total of 49,112 cells passed the quality control criteria, with 15,322 identified as tumor B cells (31.19%, Figure 1B). We observed a high level of transcriptomic heterogeneity among tumor cells in patients. Differential expression analysis revealed 1,236 genes with significantly different expression between sensitive and resistant samples (adjusted p-value < 0.05). To evaluate the predictive performance of our model, we employed a leave-one-sample-out approach and achieved an average accuracy of 90.07%. For instance, when sample K0 was used as the test dataset and all other samples were used as the training dataset, 91.49% of cells from sample K0 were accurately predicted to be sensitive (Figure 1C). Further, visualization of the hidden layer using the Uniform Manifold Approximation (UMAP) embedding plot demonstrated that sample K0 fell into the same area as other sensitive samples. Conclusion In conclusion, our study highlights the potential of deep learning models and single-cell RNA sequencing data in predicting the response to CAR-T therapy in MCL. These findings hold promise for clinical applications and may contribute to the development of personalized treatment strategies for MCL patients.

Rail transport electric drive with intelligent control system

Purpose. The study of the conditions for the occurrence of frictional self-oscillations, the synthesis of a neuroregulator eliminating self-oscillation, the development of a system for automatic control of the of railway transport speed depending on the curvature of the track on the basis of computer vision technology. Methodology. Mathematical analysis and modeling. Findings. The paper presents the results of the development and research of an intelligent control system for the electric drive of a DS3 mainline electric locomotive. The developed systems have a single easily implemented motor speed feedback, which does not create difficulties in physical implementation. It is noted that a common feature of the electric drive of rail transport is a nonlinear load characteristic. It is shown that, under certain combinations of parameters, frictional self-oscillations are possible in the traction electric drive. Effective elimination of frictional self-oscillations is done by synthesizing the system with a neuroregulator. The neural network has three input neurons that receive a vector of input signals in the form of a voltage signal, a signal of the motor speed value of the current and previous energy speed values. The number of neurons in the hidden layer of the system is 20 and there is one output neuron.The control actions for the frequency converter are formed on the output neuron. Neural networks of this type are designated NN3-20-1. The genetic algorithm method is used for all optimization of neural network parameters. The simulation model of the electric drive of rail transport has the integration of a computer vision unit. Increasing the level of automation and safety of rail vehicles is possible on the basis of computer vision. A feature of this structure is the presence of an NN neural regulator in it. NN ensures the specified quality of the transient process over the entire load range and when the operating point is located on a falling section. A system for automatic control of the speed of rail vehicles depending on the curvature of the track has been developed to increase the level of automation and traffic safety. Modeling of the system showed its efficiency, which is manifested in a decrease in the speed of rail vehicles when moving along a section of track with curvature. Originality. Effective elimination of frictional self-oscillations due to the use of a neuroregulator. Practical value. A system for automatically adjusting the speed of rail transport depending on the curvature of the track has been developed to increase the level of automation and traffic safety.

A Novel Deep Kernel Incremental Extreme Learning Machine Based on Artificial Transgender Longicorn Algorithm and Multiple Population Gray Wolf Optimization Methods

Redundant nodes in a kernel incremental extreme learning machine (KI-ELM) increase ineffective iterations and reduce learning efficiency. To address this problem, this study established a novel improved hybrid intelligent deep kernel incremental extreme learning machine (HI-DKIELM), which is based on a hybrid intelligent algorithm and a KI-ELM. First, a hybrid intelligent algorithm was established based on the artificial transgender longicorn algorithm and multiple population gray wolf optimization methods to reduce the parameters of hidden layer neurons and then to determine the effective number of hidden layer neurons. The learning efficiency of the algorithm was improved through the reduction of network complexity. Then, to improve the classification accuracy and generalization performance of the algorithm, a deep network structure was introduced to the KI-ELM to gradually extract the original input data layer by layer and realize high-dimensional mapping of data. The experimental results show that the number of network nodes of HI-DKIELM algorithm is obviously reduced, which reduces the network complexity of ELM and greatly improves the learning efficiency of the algorithm. From the regression and classification experiments, its CCPP can be seen that the training error and test error of the HI-DKIELM algorithm proposed in this paper are 0.0417 and 0.0435, which are 0.0103 and 0.0078 lower than the suboptimal algorithm, respectively. On the Boston Housing database, the average and standard deviation of this algorithm are 98.21 and 0.0038, which are 6.2 and 0.0003 higher than the suboptimal algorithm, respectively.

The ANN Architecture Analysis: A Case Study on Daylight, Visual, and Outdoor Thermal Metrics of Residential Buildings in China

Selecting an appropriate ANN model is crucial for speeding up the process of building performance simulation during the design phase of residential building layouts, particularly when evaluating three or more green performance metrics simultaneously. In this study, daylight, visual, and outdoor thermal metrics were selected as main green performance. To find the suitable ANN model, sensitivity analysis was used to obtain a set of proper parameters applied to the ANN structure. To train the ANN model with a higher predicting accuracy, this paper tested four different scenarios of ANN parameter setups to find some general guidelines about how to set up an ANN model to predict DF, sunlight hours, QuVue and UTCI. The results showed that an ANN model with a combined output variable demonstrated better average prediction accuracy than ANN models with a separated output variable. Having two times the number of training samplings compared to the number of input variables can lead to a high accuracy of prediction. The ideal number of neurons in the hidden layer was approximately 1.5 times the number of input variables. These findings of how to improve the ANN model may provide guidance for modeling an ANN for building performance.

Hyperspectral Anomaly Detection with Auto-Encoder and Independent Target

As an unsupervised data representation neural network, auto-encoder (AE) has shown great potential in denoising, dimensionality reduction, and data reconstruction. Many AE-based background (BKG) modeling methods have been developed for hyperspectral anomaly detection (HAD). However, their performance is subject to their unbiased reconstruction of BKG and target pixels. This article presents a rather different low rank and sparse matrix decomposition (LRaSMD) method based on AE, named auto-encoder and independent target (AE-IT), for hyperspectral anomaly detection. First, the encoder weight matrix, obtained by a designed AE network, is utilized to construct a projector for generating a low-rank component in the encoder subspace. By adaptively and reasonably determining the number of neurons in the latent layer, the designed AE-based method can promote the reconstruction of BKG. Second, to ensure independence and representativeness, the component in the encoder orthogonal subspace is made into a sphere and followed by finding of unsupervised targets to construct an anomaly space. In order to mitigate the influence of noise on anomaly detection, sparse cardinality (SC) constraint is enforced on the component in the anomaly space for obtaining the sparse anomaly component. Finally, anomaly detector is constructed by combining Mahalanobi distance and multi-components, which include encoder component and sparse anomaly component, to detect anomalies. The experimental results demonstrate that AE-IT performs competitively compared to the LRaSMD-based models and AE-based approaches.

Oil Formation Volume Factor Prediction Using Artificial Neural Network: A Case Study of Niger Delta Crudes

Artificial intelligence techniques provide an alternative to conventional empirical correlation methods when experimentally determined oil formation volume factors (OFVF) are lacking. A new mathematical model is proposed using an artificial neural network (ANN) for estimating the OFVF for the Niger Delta crude oils. The method consists of two stages: data decorrelation through principal component analysis (PCA) and OFVF estimation through ANN. Data decorrelation was used to reduce redundancy in the data which decreased the number of neurons in the hidden layer needed for an ANN to achieve high accuracy. In the development of the model, 316 data points were obtained from the Niger Delta region of Nigeria. Application of data cleaning, outliers’ elimination and PCA analysis reduced the data to 243 points. 213 data points were used to develop the model of which 75% was used for training, 15% for validation and 10% for testing. The remaining 30 data points were used to test the predictive capability of the proposed model. The results obtained were compared with widely accepted empirical correlations of Standing, Glaso, Vazquez, Ikiensikimama & Ajienka, and Al-Marhoun. The proposed new model performed better than all of them in terms of coefficient of correlation, AAPE and RMSE. Hence the ANN model will reduce cost, save time, and also predict the OFVF of Niger Delta crudes with higher precision.

Prediction of pipeline fatigue crack propagation under rockfall impact based on multilayer perceptron

Recently, the artificial intelligence technologies have been widely used in the field of pipeline integrity management. When crossing mountains, pipelines would inevitably encounter rockfall impact, which will potentially affect the growth of crack. However, previous research barely investigated the effect of sudden rockfall impact on health management of pipelines with fatigue cracks. To overcome this limitation, a novel crack propagation prediction algorithm based is proposed for pipelines subjected to rockfall impact. The stress intensity factor (SIF) rockfall impact ratio is introduced to describe the interaction effect of rockfall on the fatigue crack of pipelines. And the dynamic SIF values are acquired by finite element modeling (FEM) where 354 models with different parameters are analyzed. To more accurately forecast the crack growth under the rockfall impact, a method integrates multilayer perceptron (MLP) with Paris’ law is proposed based on the above reliable database. Two parameters impacting the performance of the network including the number of neurons in the hidden layer and the hidden layer's activation function are evaluated and network with the most precise prediction results is selected. Quantitative analyses are performed for key factors including rockfall mass, impact velocity, impact position and crack size. The prediction results using dynamic SIF values are compared with the static ones to indicate the effect of rockfall on the crack propagation. The proposed method is valuable to support decision-making in pipeline reliability assessment and integrity management.

Prediction and evaluation of polymer turbulent drag reduction efficiency based on BP neural network

In the process of oil exploitation and transportation, in order to effectively predict and control energy consumption for drag reduction of oil flow, in this paper a BP neural network was proposed based method for predicting and evaluating the turbulent drag reduction efficiency of polymers, which can greatly improve the current situation of relying on empirical formulas and low generality in polymer turbulent drag reduction efficiency prediction. Based on the experimental data sets of four commercial polymer drag-reducing agents FLOXL, М-Flowtreat, Necadd-447, and FLO MXA, obtained at different polymer concentrations, viscosity, density, and Reynolds number, a BP neural network has been established and the optimal number of neurons in the hidden layer was selected using the root mean square error (RMSE) value to obtain the optimal BP neural network prediction model. The BP neural network prediction models for the four polymer drag-reducing agents all have a good fit of 0.98 or above, and the R2 of the trained BP neural network for the Necadd-447 drag-reducing agents is 0.9949, which is the best among the four polymer drag-reducing agents. The BP neural network established in this paper can be applied to the turbulent drag reduction transport of long-distance pipelines for oil products to achieve the prediction of the drag reduction efficiency of polymer additives.

Particle swarm optimization of Elman neural network applied to battery state of charge and state of health estimation

Lithium-ion batteries have emerged as an energy storage solution for electrified vehicles. A Battery Management System (BMS) is critical for efficient and reliable system operation, in which State of Charge (SoC) estimation and State of Health (SoH) monitoring are of major importance to ensure optimal energy management in battery vehicles for increased autonomy and battery life. This paper presents a neural network with Elman architecture trained for a lithium-ion cell, aiming at SoC and SoH estimation. The multi-objective optimization approach based on the particle swarm algorithm is used for the training in order to lower the root mean square error in calculating the SoC and SoH. For such purposes, the neural network characteristics are optimized, such as the number of hidden layers, the number of neurons in each hidden layer, the activation functions, the bias value, and the weights of the inputs and outputs. The best trade-off solution has an error of 2.56% in the average SoC estimate and 0.003% in the average SoH estimate.

Efficient inverse design method of AWG based on BPNN-PSO algorithm

This paper presents an efficient inverse design method of arrayed waveguide grating (AWG) based on BPNN-PSO. Five important structural parameters of AWG are selected as inputs, and the spectral responses of three output channels are used as outputs. A forward network is trained, and the prediction accuracy of the network is improved by optimizing the number of hidden layer neurons. By combining the forward network with the traditional particle swarm optimization (PSO) algorithm, a series of structure designs targeting the desired performance can be generated. This method significantly improves the efficiency and accuracy of the PSO algorithm by introducing neural networks, making it suitable for efficient inverse design and multi-structure analysis of AWG.

Analysis of car–following behaviors based on data–driven and theory–driven car–following models: Heterogeneity and asymmetry

The data–driven car–following model can realistically depict driving behavior, and is widely used in simulating the longitudinal movements of vehicles. However, it has some unexpected results such as uncontrollable output and potential security risks, making it challenging to apply practically. To address this issue, this paper proposes a combination model that integrates the theory–driven model (IDM) with the data–driven model (PSO–Bi–LSTM). This novel integration is referred to as the IDM–Bi–LSTM combination car–following model. This combination model offers two key features. First, to accurately predict vehicle trajectories, rather than manually configuring the number of neurons in the hidden layer and iterations, we employ the Particle Swarm Optimization (PSO) algorithm to optimize these parameters in the data–driven model. Consequently, we establish a PSO–Bi–LSTM data–driven model framework. Second, our combination model integrates the prediction outcomes of the IDM model with those of the PSO–Bi–LSTM neural network. This amalgamation retains the controllability and security offered by the theory–driven model, while also leveraging the prediction accuracy of the data–driven model. To evaluate the effectiveness of our model, we conduct numerical simulations of heterogeneous driving behaviors. The results reveal a significant reduction in error for the combination car–following model, exhibiting reductions of 60.2% and 54.4% compared to the individual IDM theory–driven model and PSO–Bi–LSTM data–driven model, respectively. Additionally, the simulation outcomes for asymmetric driving behavior further confirm the superior performance of the combination model.

The investigation of transferability related to the ImageNet-based transfer learning model in galaxy classification

The phenomenon of transfer learning, specifically the transferability of ImageNet, in the context of galaxy datasets has been relatively under-researched. This study seeks to address this gap by employing the Galaxy10 DECals dataset, which is a 10-classification dataset. Three classic models, namely MobileNet, VGG, and ResNet, were developed and customized to the current dataset by modifying the number of neurons in the last layer. The experimental phase is divided into three parts, including the comparison between the use of ImageNet and non-ImageNet, model performance comparison, and confusion matrix analysis. The results demonstrate that the utilization of ImageNet produces better outcomes, with the MobileNet model exhibiting the highest performance. Further analysis revealed that the inclusion of ImageNet weights can enhance the classification accuracy of some data. Although the present study was successful in achieving its objectives, future research should focus on exploring and elucidating the underlying mechanisms driving these findings.

Face-emotion classification guided by deep convolutional neural network

With the rapid development of computer vision and convolutional neural networks, the task of automatic face emotion classification has become a reality. The aim of this study is to improve the underlying neural network model to achieve effective face emotion classification. By presenting a simplified network to generate the recognition model, the author enhances the underlying neural network architecture. The model, in particular, augments the underlying neural network with a convolutional layer, a maximum pooling layer, and a discard layer, and increases the number of neurons in the dense layer from 25 to 128. The convolutional layer allows for the automatic extraction of sentiment features. To decrease the parameters in the feature maps, the maximum pooling layer is applied. The experiments are constructed on the Facial Emotion Recognition 2013 dataset (FER-2013). The streamlined network model improves performance by 6% to 56.32% as compared with the basic neural network model. Numerous experiments show that the proposed streamlined network model can effectively recognize facial emotions. In addition, the author analysis the confusion matrix and finds that the model has weak feedback for aversive emotions. Future research will focus on improving the representation of unclear features such as aversive emotions to enhance model generalization.

Investigating and Modeling of the Scour Downstream of a Tree Trunk Deflector in a Straight Channel

Scouring depends on several factors, including the water flow of artificial obstacles, sections, piers, and foundations, the disturbance of bed materials, and soil permeability. The other factors are the non-parallelism between piers and the water flow, the type of river activity (static or dynamic), and the existence of a waterfall or an obstacle that forms a waterfall in natural bed materials, causing the underlying bed materials to be washed away. This study fully investigated how the movement of a tree trunk affects a river’s flow by considering different flow conditions using the artificial neural network (ANN) model. A feedforward optimal network with the error back-propagation training algorithm and sigmoid transfer functions was used for four models. To determine the number of neurons in the hidden layer, one and ten neurons were selected in the hidden layer according to verification indicators. In addition, a physical model was utilized to measure data. To verify and test the models, our data were gathered in a laboratory using the physical model. Considering the network structure of one neuron in the hidden layer, a comparison was made between dimensional and dimensionless parameter models that are effective in terms of the dimensions of the scour hole. The comparison between the results of the ANN and the measured data using nonlinear regression models demonstrated that the ANN was more accurate and capable of simulating phenomena. Additionally, R and RMSE values were between 0.93 and 0.98, as well as 0.18 and 0.013, respectively. Finally, the results related to the width, height, length, and depth of the scour revealed that the modified DOT model had the best agreement with Mahdavizadeh’s measured data.

SRTM DEM Correction Based on PSO-DBN Model in Vegetated Mountain Areas

The Shuttle Radar Topography Mission (SRTM) digital elevation model (DEM) is extensively utilized in various fields, such as forestry, oceanography, geology, and hydrology. However, due to limitations in radar side-view imaging, the SRTM DEM still contains gaps and anomalies, particularly in areas with an intricate topography, like forests. To enhance the accuracy of the SRTM DEM in forested regions, commonly employed approaches include regression analysis and artificial neural networks (ANN). Nevertheless, existing regression methods struggle to accurately capture the intricate nonlinear relationship between the error and influencing factors. Additionally, traditional ANN models are susceptible to overfitting, resulting in subpar accuracy. Deep Belief Network (DBN) is a highly precise algorithm in deep learning. However, the intricate combination of hyperparameters often leads to limited generalization ability and model robustness when correcting DEM. The present study proposes an error prediction model based on the DBN optimized by Particle Swarm Optimization (PSO) for SRTM DEM correction. By utilizing the PSO algorithm, we aim to identify the optimal combination of hyperparameters of DBN, including the number of neurons in the hidden layer and the learning rates. The experiment focuses on two regions in Hunan Province, China, characterized by abundant vegetation cover. The reference data utilized for comparison is ICESat/GLAS data. The experimental results demonstrate that the mean error (ME) and root mean square error (RMSE) of the SRTM DEM corrected by the proposed algorithm in these two regions are significantly reduced by 93.5%–96.0% and 21.5%–23.5%, respectively. Moreover, there is an improvement of over 26.1% in accuracy within complex terrain areas. Specifically, in broadleaf forest, the PSO-DBN method exhibits a remarkable accuracy improvement of 26.2%, while the DBN-corrected SRTM DEM shows an improvement of 15.3%. In coniferous forest, the PSO-DBN method achieves an accuracy improvement of 14.8%, whereas the DBN-corrected SRTM DEM demonstrates a gain of 5.8%. The approach provides a more effective and robust tool for correcting SRTM DEM or other similar DEMs over vegetated mountain areas.

Application of Neural Network Modeling in Problems of Predicting the Level of River Floods

The purpose of this article is to consider the theoretical and practical issues of developing neural network models for river flood forecasting (in case of the Belaya River near Ufa), as well as to implement the corresponding neural network in Python. To build a training sample, archival data from meteorological services and meteorological observation sites for the flood periods of the Belaya (Agidel) River in 2018–2022 were used. The following indicators were collected and analyzed: water level, water temperature, day and night air temperature, precipitation, snow depth, including information about the pre-flood condition of the snow cover. The software implementation of the neural network was performed using the PyTorch deep learning library; in addition, modules from the Matplotlib and Pandas libraries were used. The stability of the operation of this neural network was studied when the following parameters were changed: the optimizers used (Adam, Adamax and Rprop); learning rate coefficient; the number of neurons in the hidden layer; number of learning epochs. It is concluded that the developed neural network can be used to model the flood level when creating short-term forecasts. In order to move to longer-term forecasts in the future, it is planned to further expand the size of the factors in the training sample.

Multi-objective optimization of breakthrough times for hydrogen purification through layered bed pressure swing adsorption based on genetic algorithm and artificial neural network model

Hydrogen purification from steam methane reforming (SMR) by pressure swing adsorption (PSA) technology is a common method to obtain high-purity hydrogen. The breakthrough time can well reflect the adsorption dynamics and help PSA cycle design. In order to avoid time-consuming and labor-intensive breakthrough curve experiments, it is very necessary to develop a fast and accurate surrogate model. In this study, a genetic algorithm (GA) and artificial neural network (ANN) are combined to predict and optimize breakthrough times of adsorbates in activated carbon/zeolite layered beds. Using the Latin hypercube sampling strategy, training data sets of GA-optimized ANN (GA-ANN) obtains from the physical model of adsorption, heat and mass transfer model. The genetic algorithm (GA) optimizes the weights and biases of ANN for better performance. The ANN topology has 4 input variables (superficial velocity, activated carbon height, adsorption pressure and feed temperature) and 3 output variables (breakthrough times of CH4, CO and CO2). The number of neurons in the hidden layer for the GA-ANN model was optimized as 7 to predict and optimize the maximum breakthrough time of SMR with a four-component (H2/CH4/CO/CO2) system. The sensitivity analysis displays that relative importance to the breakthrough time is in the order of superficial velocity (40.88%) > adsorption pressure (24.55%) > activated carbon height (21.04%) > feed temperature (13.53%). To obtain high-purity hydrogen, the GA-ANN model combined with multi-objective GA optimization is used to maximize the breakthrough times of CH4 and CO. The GA-ANN surrogate model proposed in this paper can not only accurately and quickly achieve the purpose of predicting the system breakthrough time with a high correlation coefficient (R = 0.9937) but also obtains the optimal operating conditions of PSA hydrogen purification.

Extreme learning machine model with honey badger algorithm based state-of-charge estimation of lithium-ion battery

Accurate state-of-charge (SOC) detection was still a challenging task to complete due to complex battery dynamics and constantly changing external conditions. The formula for SOC was difficult to determine since external parameters including voltage, current, temperature, and battery arrangement were complex. Also the methods for estimating SOC that were already in use were not always appropriate for the same car operating in various road and climatic conditions. In all situations, the conventional methodologies did not deliver an accurate estimation performance. Here, a unique optimization-based Extreme Learning Machine (ELM) was created to accurately determine a battery's SOC and enhance the operation and safety of battery systems. A lithium ion battery was first created, and data on its current, voltage, SOC, capacity, duration, and discharge rate were gathered to produce a real-time dataset at several temperatures, including 00,250 and 450. The dataset underwent additional pre-processing to standardize the values and enhance the accuracy of the data. To determine the precise state of the battery, these pre-data were loaded into the ELM model. However, the performance of ELM was significantly influenced by the length of training and the number of neurons in a hidden layer. An advanced Honey Badger Optimization Algorithm (HBA) was used to choose the appropriate hidden neurons and increase the estimation accuracy in order to overcome this problem. The proposed SOC estimation model provides 97% accuracy in the FUDS drive cycle and 99% accuracy in the US06 drive cycle. The proposed model provides a well performance for estimating SOC in lithium-ion battery at various temperature, also the proposed model was fit for real time implementation.