Use Of Back-propagation Neural Network Research Articles

An Improved Self-Adaptive Flower Pollination-Backpropagation Neural Network for the State of Charge Estimation of Lithium-Ion Batteries with Output Sliding Average Window Strategy

With the rapid development of electric vehicles and green energy sources, the use of backpropagation neural network (BPNN) to precisely estimate the state of charge (SOC) in lithium-ion batteries has become a popular research topic. However, traditionally BPNN has low prediction accuracy and large output fluctuations. To address the shortcomings of BPNN, self-adaptive flower pollination algorithm (SFPA) was proposed to optimize the initial weights and thresholds of BPNN, and an output sliding average window (OSAW) strategy is proposed to smooth SOC outputs in this research, which SOC estimation method is named SFPA-BP-OSAW. In addition, the performance of the newly proposed method is compared with other common related algorithms under different working conditions to verify the effectiveness of SFPA-BP-OSAW. The experimental results show that the mean absolute error of SFPA-BP-OSAW is 0.771% and 0.897%, and the root mean square error is 0.236% and 0.37%, respectively, under HPPC and BBDST working conditions. Experimental data and error analysis show that the method proposed in this paper has fast convergence, high prediction accuracy, and curve smoothness.

Methods for controlling heavy metals in environmental soils based on artificial neural networks

The problem of heavy metal pollution in soil has become a global environmental problem, and it is very important to predict and manage the heavy metals in the environmental soil in a timely manner. The changes in heavy metal content in soil have characteristics such as nonlinearity and large delay, making it difficult to predict heavy metals in soil using traditional methods. Traditional prediction methods are complex and cumbersome, which can lead to longer treatment time and easy secondary pollution. This article analyzed the Back Propagation neural network (BPNN) in artificial neural networks (ANN) and applied it to the prediction of heavy metals in environmental soils. BPNN has good nonlinear function approximation ability, so it can be well applied to complex problems such as soil heavy metal prediction. The methods of treating soil heavy metals include physical repair method, chemical repair method, microbial repair method, plant repair method, plant microbial combined repair method and so on. The use of BPNN can predict heavy metals in environmental soils through adaptive dynamic learning. However, the training time of the BPNN is relatively long and the convergence speed is relatively slow. Therefore, additional momentum terms were added to adjust the weights and thresholds of the network to improve the BPNN. In the experiment, the prediction performance of the improved BPNN was compared before and after the improvement. This article took 50 monitoring data of heavy metals in the same soil in a certain region in 2021 as sample data and predicted the content of heavy metals in the soil using improved and improved BPNN. Due to time constraints, this article only conducted experimental analysis on heavy metals such as lead and cadmium. In the first experiment, when the soil sample data was 50, the prediction accuracy of the BPNN for cadmium before and after improvement was 75.95% and 89.56%, respectively. In the second experiment, when the soil sample data was 50, the prediction accuracy of the BPNN for cadmium before and after improvement was 77.99% and 89.85%, respectively. The improved BPNN has good prediction accuracy and can effectively predict the status of heavy metals in soil. The analysis in this article can provide scientific basis for the comprehensive prevention and control of heavy metals in regional soil, and also provide reference for the development of pollution-free agriculture and ensuring food safety.

KLASIFIKASI MOOD MUSIK BERDASARKAN MEL FREQUENCY CEPSTRAL COEFFICIENTS DENGAN BACKPROPAGATION NEURAL NETWORK

In music industry, each music is grouped by type, including music genre, artist identification, instrument introduction, and mood. Then came a field of research called Music Information Retrieval (MIR) which is a field of science that retrieves and processes the metadata of music files to perform the grouping. This research is based on the uniqueness of music that has its own mood implied in it. By creating a Machine Learning model using Backpropagation Neural Network (BPNN) based on the Mel Frequency Cepstral Coefficients (MFCC) input feature, it will be able to classify types of music based on mood. Grouping is carried out on four mood classes based on Thayer's model. Based on several previous studies, the use of MFCC in voice processing produces very good accuracy as well as the use of BPNN for classification, which is expected to result in better machine learning model performance. The data used in this study were obtained from the Internet with a total dataset of 200. The results obtained from this study are the classification of music mood using BPNN based on the MFCC feature capable of producing 87.67%. accuracy.

High-accuracy Automated Diagnosis of Parkinson's Disease

Parkinson's disease (PD), which is the second most common neurodegenerative disease following Alzheimer's disease, can be diagnosed clinically when about 70% of the dopaminergic neurons are lost and symptoms are noticed. Neuroimaging methods such as single photon emission computed tomography have become useful tools in vivo to assess dopamine transporters (DATs) in the striatal region. However, inter- and intra-reader variability of construing the images might result in misdiagnosis. To overcome the challenges posed by classification of the disease, image preparation techniques and a back propagation neural network (BPNN) have been proposed. The aim of this study is to show that the proposed method can be used for the classification of PD with high accuracy. In this study, we used basic image preparation techniques and a BPNN on DAT imaging datasets from the Parkinson's Progression Markers Initiative. 1,334 PD and 212 normal control (NC) subjects were included. In the image preparation phase, adaptive histogram equalization was applied to the cropped images, followed by image binarization. Then, the mass-difference method was applied to separate the regions of interest with similar values. Finally, the binarized images were subtracted from the original images, and the average pixel per node approach was applied to the images to minimize the inputs. In the BPNN phase, 400 input neurons and 2 output neurons were used. The dataset was divided into three sets: training, validation, and test. The BPNN was trained several times in order to obtain the optimum values. The use of 40 hidden neurons, a learning rate of 0.00079, and a momentum factor of 0.90 produced superior results and were applied in the final BPNN architecture. The tolerance value used was 0.80. Uniquely, we found the sensitivity, specificity, and accuracy for PD vs. NC classification to be 99.7%, 99.2%, 99.6%, respectively. To the best of our knowledge, this is the highest accuracy value achieved in the existing literature. Our method increases computational speed together with improved performance. We have shown that effective image processing methods and the use of BPNN can successfully be applied to PD datasets to accurately determine any abnormalities in DATs. Using the shallow neural network, this procedure requires less processing time compared to other methods, and its accuracy, sensitivity, and specificity are reliable. However, further studies are needed to establish a prediction method for the preclinical and prodromal stages of the disease.

Analysing wear behaviour of Al–CaCO3 composites using ANN and Sugeno-type fuzzy inference systems

Design of experiment for the development of stir cast calcium carbonate-reinforced aluminium composite is a search for optimum combination of material and process control parameters for best physical and mechanical properties. A soft-computing model can accurately learn the complex interactions between process parameters to provide great insights in the development of this composite. This paper demonstrates and analyses the potential of artificial neural network (ANN) and Sugeno-type fuzzy inference systems (FIS) for wear behaviour prediction of calcium carbonate-reinforced aluminium composites. The models were trained with data collected from the experiment. The data consist of filler particle size of 150 μm with weights fractions varied from 0 to 25 wt%, in step of 5. Wear test data at different time of contacts (30, 60, 90, 120 and 150 s) and variable loads of 2.27 N, 4.54 N and 6.80 N were collected, resulting to 120 length vectors. Comparing the experimental results of wear test with those predicted using the ANN and Sugeno-type FIS, the integration of calcium carbonate particulate enhanced the wear characteristics of Al matrix up to 200%. On the use of back-propagation neural network with 4–3–1 architecture for wear prediction, the Levenberg–Marquardt training algorithm performs better. For Sugeno-type FIS, the Gaussian membership function resulted to the best prediction of wear rate. When ANN and Sugeno-type FIS performance on the test set were analysed based on some statistical parameters, the later returned an R2 value of 0.9775 as against ANN’s value of 0.3684. The predicted wear rate using ANFIS with Gaussian membership functions was in good agreement with the experimental values.

Eczema Disease Detection and Recognition in Cloud Computing

ED (eczema disease) is the most common form of skin inflammation in humans. Most of the skin disease is curable at initial stages with the advancement of technology. So, an early detection of skin disease can save the patient's life and prevent the progression of the disease. It is proposed to have a study on the diagnosis of ED using BpNN (backpropagation neural network) in cloud computing approach due to that BpNN is currently widespread research area in medicine and plays an important role in a decision support system. In this paper, an attempt has been made to make use of BpNN in the medical field along cloud computing to detect ED.

ECG signal diagnoses using intelligent systems

Accurate diagnosis of arrhythmias plays a crucial role in saving the lives of many heart patients. The aim of this research is to find the more efficient method to diagnosis electrocardiogram (ECG) diseases. This work presents the use of Backpropagation neural network (BPNN) and fuzzy logic for automatic detection of cardiac arrhythmias based on analysis of the ECG. These a more valuable tool used to classify ECG signals in cardiac patients. Data collected from physioBank ATM. The analysis of the ECG signal is performed in MATLAB environment. In BPNN the results appear that the only two misclassifications happened to result in an accuracy of 90.4%. while in fuzzy inference system the results appear that the classification accuracy is 100%.

Identification of pummelo cultivars by using Vis/NIR spectra and pattern recognition methods

Vis/NIR spectroscopy was used in combination with pattern recognition methods to identify cultivars of pummelo (Citrus grandis (L.) Osbeck). A total of 240 leaf samples, 60 for each of the four cultivars were analyzed by Vis/NIR spectroscopy. Soft independent modeling of class analogy (SIMCA), partial least square discriminant analysis (PLS-DA), back propagation neural network (BPNN) and least squares support vector machine (LS-SVM) were applied to the spectral data. The first 8 principal components extracted by principal component analysis were used as inputs in building the BPNN and the LS-SVM models. The results showed that a 97.92 % of discrimination accuracy was achieved for both the BPNN and the LS-SVM models when used to identify samples of the validation set, indicating that the performance of the two models was acceptable. Comparatively, the results of the PLS-DA and the SIMCA models were unacceptable because they had lower discrimination accuracy. The overall results demonstrated that use of Vis/NIR spectroscopy coupled with the use of BPNN and LS-SVM could achieve an accurate identification of pummelo cultivars.

Forecasting Models for Hydropower Unit Stability Using LS-SVM

This paper discusses a least square support vector machine (LS-SVM) approach for forecasting stability parameters of Francis turbine unit. To achieve training and testing data for the models, four field tests were presented, especially for the vibration inY-direction of lower generator bearing (LGB) and pressure in draft tube (DT). A heuristic method such as a neural network using Backpropagation (NNBP) is introduced as a comparison model to examine the feasibility of forecasting performance. In the experimental results, LS-SVM showed superior forecasting accuracies and performances to the NNBP, which is of significant importance to better monitor the unit safety and potential faults diagnosis.

Exploiting a Modified Gray Model in Back Propagation Neural Networks for Enhanced Forecasting

Back propagation (BP) neural network has been widely used for various data predictions in application. One of the challenging issues in various applications of BP neural network is how to improve its reliability as well as its stability. In this paper, by using a prototype for the forecast of atmospheric radiation and atmospheric ozone concentration in the state of Ohio, USA, we show that a direct use of the BP neural network may lead to the loss of forecast reliability during its evolutionary process. Under the framework of Verhulst biological model with embedding cognitive computation, we show that one can effectively reduce the dispersion of pure randomness of data sets in BP neural network and improve the prediction of future data by incorporating accumulation generation operation (AGO) into the system. We demonstrate that the integration of a modified gray model and the AGO can offer a more desirable BP neural network algorithm, and both stability and reliability become much improved compared to the direct use of BP neural network reported in current literature.

Biotech firm valuation in an emerging market – evidence from Taiwan

PurposeThis paper aims to explore the relationship of various financial and non‐financial factors to corporate value and how these factors can be used for the purpose of firm valuation. The focus is placed on a developing high‐tech industry.Design/methodology/approachThe authors collect and compare data from companies within the time window of 1997 through 2010. The techniques of stepwise regression and back‐propagation neural network (BPNN) are applied to analyze this data, where the variables of operating profit margin, ROE, ROA, net income ratio, Tobin's Q and stock price are chosen to indicate firm value.FindingsEach firm value variable appears to have a different set of estimator variables consisting of financial and non‐financial factors. The estimator variable in the set that has a high influence relative to the others tends to be financial factor. However, certain non‐financial factors appear to be considered as an estimator variable for different firm value variables more often than financial factors such as employee productivity, wealth created per employee, revenue growth rate, management expense per employee, R&D expense to management expense ratio, and R&D expenditure to total assets ratio. Further, the incorporation of BPNN shows an improvement of the result of the regression method in terms of overall estimation error, especially for operating profit margin.Originality/valueThe authors' investigation highlights the importance of the use of non‐financial factors for firm valuation in developing biotech industries. The result can be helpful for investors who seek to examine information variables and indicators for the opportunity presented by the above industries. In addition, the significant estimation improvement by incorporating the BNPP method into the commonly used regression method suggests the beneficial use of BPNN in refining the traditional methods in the field.

Forecasting performance of LS-SVM for nonlinear hydrological time series

This paper presents a Least-Square Support Vector Machine (LS-SVM) approach for forecasting nonlinear hydrological time series. LS-SVM is a machine-learning algorithm firmly based on the statistical learning theory. The objective of this paper is to examine the feasibility of using LS-SVM in the forecasting of nonlinear hydrological time series by comparing it with a statistical method such as Multiple Linear Regression (MLR) and a heuristic method such as a Neural Network using Back-Propagation (NNBP). And the performance of prediction model is also dependent on the degrees of linearity (or persistency) of data, not only on the performance of model itself. Thus, we would clearly verify that prediction performance of three models according to linear extent using daily water demand and daily inflow of dam data. In the experimental results, LS-SVM showed superior forecasting accuracies and performances to those of MLR and NNBP and LS-SVM demonstrated better forecasting efficiency in nonlinear hydrological time series using Relative Correlation Coefficient (RCC) which is a relative measure of forecasting efficiency with different persistency.

Humic substance coagulation: Artificial neural network simulation

This paper investigates the use of backpropagation neural network (BPNN) to predict humic substance (HS) UV absorbance experimental results. The studied experimental sets include HS and heavy metal agglomeration, HS coagulation using polyelectrolytes and HS and heavy metal coagulation using polyelectrolytes. BPNN simulation showed high prediction accuracy where regression coefficient ( R) was > 0.95 for all simulations. Lower and higher than optimum training data input reduces BPNN reliability due to under training or over-fitting. The number of neurons study showed that a lower number of neurons led to under training, while a higher number of neurons resulted in the network memorizing the input dataset.

Artificial Neural Network Model as a Data Analysis Tool in Precision Farming

Spatial variation in landscape and soil properties combined with temporal variations in weather can result in yield patterns that change annually within a field. The complexity of interactions between a number of yield-limiting factors makes it difficult to accurately attribute yield losses to conditions that occur within a field. In this research, a back-propagation neural network (BPNN) model was developed to predict the spatial distribution of soybean yields and to understand the causes of yield variability. First, we developed a BPNN model by relating soybean yield to topography, soil, weather, and site factors and evaluated model predictions for the same field for independent years. We also explored the potential use of BPNN for predicting yields in independent fields. Finally, we evaluated the ability of the BPNN to attribute yield losses due to soybean cyst nematodes (SCN), soil pH, and weeds. A total of 14 input datasets with combinations of four controlling factors (topographic, soil fertility, weather, and site) were used. For each objective, data from fields in Iowa were used for training the BPNN, while a portion of the data was withheld to verify the accuracy of yield predictions. All BPNN models had fully connected feed-forward architecture with a back-propagation weight adjustment algorithm. When tested for a particular field, the BPNN captured the major patterns of yield variability in independent years; the root mean square error of prediction (RMSEP) was 14.2% of actual yield. When the BPNN was trained with inputs from five fields, the RMSEP at test sites was 11.2% of actual yield. When the BPNN was used to attribute yield losses to soil pH, SCN, and weed populations, standard errors were 92, 262, and 171 kg ha-1, respectively. The technique showed that the BPNN could predict spatial yield variability with an RMSEP of about 14%.

Use of BP Neural Network to Predict Hydrogen Content in Coal

Neural network is a popular method for predicting unknown process variables from measured process data. Many learning algorithms have been proposed in the literature to improve model prediction. In this paper, we introduce the concept of sample study risk in neural network (NN) to improve the prediction of hydrogen content in coal using Back Propagation (BP) NN. Targeting the problem of training convergence quality impaired by the interfering infonnation of some samples in BP NN, the validity of the concept of sample study in NN and the feasibility of analysing the chemical element in coal using NN are discussed.

Use of backpropagation neural network for landslide monitoring: a case study in the higher Himalaya

This paper presents a case study of landslide monitoring and evaluation at Okharpauwa, 19 km Chainage along Kathmandu–Trishuli highway in Nepal. An attempt has been made to predict slope movements using backpropagation neural network (BPNN). A Matlab-based BPNN model is developed, and the data from the case study are used to train and test the developed model to enable prediction of the magnitude of the ground movements with the help of input variables that have direct physical significance. An infiltration coefficient is introduced in the network architecture apart from antecedent rainfall, slope profile, groundwater level and shear strength of soil. A four-layered backpropagation neural network with an input layer, two hidden layers and one output layer is found optimal. The developed BPNN model demonstrates a promising result and fairly accurately predicts the slope movement.

Use of backpropagation neural network for landslide monitoring: a case study in the higher Himalaya

Use of backpropagation neural network for landslide monitoring: a case study in the higher Himalaya