Optimizing Weight and Threshold of BP Neural Network Using SFLA: Applications to Nonlinear Function Fitting

Among all improved BP neural network algorithms, the one improved by heuristic approach is studied in this paper. Firstly, three types of improved heuristic algorithms of BP neural network are programmed in the environment of MATLAB7.0. Then network training and simulation test are conducted taking a nonlinear function as an example. The approximation performances of BP neural networks improved by different numerical optimization approaches are compared to aid the selection of proper numerical optimization approach.

Download Full-text

Enterprise Network Marketing Prediction Using the Optimized GA-BP Neural Network

Complexity ◽

10.1155/2020/6682296 ◽

2020 ◽

Vol 2020 ◽

pp. 1-9

Author(s):

Ruyi Yang

Keyword(s):

Neural Network ◽

Bp Neural Network ◽

Time Series Prediction ◽

Nonlinear Function ◽

Development Trend ◽

Production Volume ◽

Network Marketing ◽

Bp Network ◽

Enterprise Network ◽

The Neural Network

As a brand-new marketing method, network marketing has gradually become one of the main ways and means for enterprises to improve profitability and competitiveness with its unique advantages. Using these marketing data to build a model can dig out useful information that the business is concerned about, and the company can then formulate marketing strategies based on this information. Sales forecasting is to speculate on the future based on historical sales. It is a tool for companies to determine production volume and ensure the balance of product supply and sales. It can help companies make correct business decisions to maximize profits. The neural network can approximate the nonlinear function with arbitrary precision, and the time series prediction model based on the neural network can well reflect the nonlinear development trend of information. Based on the analysis of the shortcomings of the traditional BP network, this paper uses a genetic algorithm with good global search capabilities to improve the neural network. The thought and theory of optimizing the initial weight and threshold of the neural network of the GA algorithm are discussed in detail. While expounding the forecasting method, it uses specific examples to analyze the performance and characteristics of the GA-BP network in the enterprise network marketing forecasting. The results show that the GA-BP neural network is higher than the traditional BP neural network in terms of prediction accuracy and adaptability.

Download Full-text

A Novel Nonlinear Function Fitting Model Based on FOA and GRNN

Mathematical Problems in Engineering ◽

10.1155/2019/2697317 ◽

2019 ◽

Vol 2019 ◽

pp. 1-10 ◽

Cited By ~ 2

Author(s):

Yongli Zhang ◽

Jianguang Niu ◽

Sanggyun Na

Keyword(s):

Neural Network ◽

Statistical Methods ◽

Back Propagation ◽

Nonlinear Function ◽

Step Length ◽

Small Samples ◽

Main Function ◽

Function Fitting ◽

Griewank Function ◽

Searching Method

The nonlinear function fitting is an essential research issue. At present, the main function fitting methods are statistical methods and artificial neural network, but statistical methods have many inherent strict limits in application, and the back propagation (BP) neural network used widely has too many optimized parameters. For the gaps and lacks of existing researches, the FOA-GRNN was proposed and compared with the GRNN, GA-BP, PSO-BP, and BP through three nonlinear functions from simplicity to complexity for verifying the accuracy and robustness of the FOA-GRNN. The experiment results showed that the FOA-GRNN had the best fitting precision and fastest convergence speed; meanwhile the predictions were stable and reliable in the Mexican Hat function and Rastrgrin function. In the most complex Griewank function, the prediction of FOA-GRNN was becoming unstable and the model did not show better than GRNN model adopting equal step length searching method, but the performance of FOA-GRNN is superior to that of GA-BP, PSO-BP, and BP. The paper presents a new approach to optimize the parameter of GRNN and also provides a new nonlinear function fitting method, which has better fitting precision, faster calculation speed, more few adjusted parameters, and more powerful processing ability for small samples. The processing capacity of FOA for treating high complex nonlinear function needs to be further improved and developed in the future study.

Download Full-text

A Neural Network Model of Restrained Recovery for Shape Memory Alloys

Advanced Materials Research ◽

10.4028/www.scientific.net/amr.650.172 ◽

2013 ◽

Vol 650 ◽

pp. 172-177

Author(s):

Shuang Wu ◽

Shou Gen Zhao ◽

Da Fang Wu ◽

Xue Mei Yu

Keyword(s):

Neural Network ◽

Shape Memory Alloys ◽

Shape Memory ◽

Network Model ◽

Neural Network Model ◽

Bp Neural Network ◽

Constitutive Models ◽

Nonlinear Function ◽

Bp Neural Network Model ◽

Bp Model

The methods of constitutive modeling of restrained recovery for Shape memory alloys (SMAs) were described in this paper and experiments were carried out to provide the essential data for the methods. The present mathematical constitutive models are inconvenient for engineering applications. Then a back propagation (BP) neural network model was developed for restrained recovery of SMAs. This BP neural network model can learn the hysteresis of SMAs in the process of heating and cooling based on its properties of nonlinear function mapping and adaptation, and it can predict the complete restrained recovery stress of SMAs with different initial strains. The predicted results obtained from the proposed BP model agree well with the experimental data. Moreover, the proposed BP model is more simple, convenient and low cost compared with the present mathematical constitutive models.

Download Full-text

Application of Neural Network in Displacement Prediction of Dongping Tunnel

Advanced Materials Research ◽

10.4028/www.scientific.net/amr.163-167.2666 ◽

2010 ◽

Vol 163-167 ◽

pp. 2666-2669

Author(s):

Zhi Xiang Yin ◽

Zhe Gao ◽

Yao Feng

Keyword(s):

Neural Network ◽

Bp Neural Network ◽

Surface Deformation ◽

Control Method ◽

Nonlinear Function ◽

Dynamic Matrix Control ◽

Network Computing ◽

Dynamic Matrix ◽

Hidden Layer ◽

Matrix Control

It has been proved that Multi-layer network computing can solve the nonlinear separable problem, but the problem which the hidden layer makes the learning more difficult limits the development of multi-layer network.Back-propagation (BP) algorithm solve this problem and promote multi-network research to regain attention. In this paper, A new method which dynamic matrix control method based on neural network is found. Its essence is that the resulting prediction signal is produced by the manner which regards neural network model as prediction model, and the predictive control of nonlinear systems would be realized by the control law which using the receding optimization algorithm. Neural network Selects the BP neural network which possess a good nonlinear function approximation capability.Aiming at dongping tunnel surface deformation prediction, the article adopt BP Neural Network to train the system basing on the given data. It shows the hiding neural node is close to precision; predictive value in good agreement with measured values, and to some extent be able to guide the construction.

Download Full-text

Risk assessment of logistics finance enterprises based on BP neural network and fuzzy mathematical model

Journal of Intelligent & Fuzzy Systems ◽

10.3233/jifs-189066 ◽

2020 ◽

Vol 39 (4) ◽

pp. 5915-5925

Author(s):

Wang Aiqun ◽

He Zicong ◽

Wang Yilin

Keyword(s):

Neural Network ◽

Mathematical Model ◽

Risk Assessment ◽

Bp Neural Network ◽

Evaluation Method ◽

Nonlinear Function ◽

Financial Risks ◽

Nonlinear Relationship ◽

Fuzzy Evaluation ◽

Good Ability

Neural network is used to deal with the nonlinear relationship, usually there is a strong nonlinear relationship between input and output. Through the self-learning of neural network, the weight of data samples is determined after training, and the optimal solution is obtained according to the process steps. In this paper, thea authors analyze the risk assessment of logistics finance enterprises based on BP neural network and fuzzy mathematical model. For logistics companies, it is necessary to determine the ability of logistics companies to engage in logistics finance business, and then to make detailed and accurate grasp of relevant information. The difference between the actual output and the expected output of the training sample is small, so the fitting is completed well, and the parameters of the neural network are further adjusted. The results show that the model has a good ability of learning nonlinear function relations. To sum up, in order to reduce logistics financial risks, we must fully understand the factors that affect logistics financial risks, determine the proportion of risk factors, and then use the fuzzy evaluation method to analyze the financial business risks.

Download Full-text

Design of PID Temperature Control System Based on BP Neural Network

Applied Mechanics and Materials ◽

10.4028/www.scientific.net/amm.602-605.1244 ◽

2014 ◽

Vol 602-605 ◽

pp. 1244-1247

Author(s):

Zhi Yong Meng ◽

Guo Qing Yu ◽

Rui Jin

Keyword(s):

Neural Network ◽

Control System ◽

Temperature Control ◽

Bp Neural Network ◽

Pid Controller ◽

Simulation Analysis ◽

Nonlinear Function ◽

Temperature Control System ◽

Bp Network ◽

Pid Algorithm

Based on BP neural network PID controller has the ability to approximate any nonlinear function, can achieve real-time online tuning PID controller parameter . Through the system simulation analysis, simulation results show that the BP neural network tuning PID control than traditional PID algorithm and BP network algorithm has a greater degree of improvement, the system has better robustness and adaptability, its output can also achieve the desired control accuracy through online adjustments. Suitable for temperature control system.

Download Full-text