The Application of BP Networks to Land Suitability Evaluation

The back propagation (BP) model of artificial neural networks (ANN) has many good qualities comparing with ordinary methods in land suitability evaluation.Through analyzing ordinary methods’ limitations,some sticking points of BP model used in land evaluation,such as network structure,learning algorithm,etc.,are discussed in detail,The land evaluation of Qionghai city is used as a case study.Fuzzy comprehensive assessment method was also employed in this evaluation for validating and comparing.

Prediction of Hot Ductility of Low-Carbon Steels Based on BP Network

The purpose of the research is to obtain an effective method to predict the hot ductility of low-carbon steels, which will be a reference to evaluate the crack sensitivity of steels. Several sub-networks modeled from BP network were constructed for different temperature use, and the measured reduction of area (A(R)) of 12 kinds of low-carbon steels under the temperature of 600 to 1000 degreesC were processed as training samples. The result of software simulation shows that the model established is relatively effective for predicting the hot ductility of steels.

Classification of Infrared Monitor Images of Coal Using an Feature Texture Statistics and Improved BP Network

It is very important to accurately recognize and locate pulverized and block coal seen in a coal mine's infrared image monitoring system. Infrared monitor images of pulverized and block coal were sampled in the roadway of a coal mine. Texture statistics from the grey level dependence matrix were selected as the criterion for classification. The distributions of the texture statistics were calculated and analysed. A normalizing function was added to the front end of the BP network with one hidden layer. An additional classification layer is joined behind the linear layer. The recognition of pulverized from block coal images was tested using the improved BP network. The results of the experiment show that texture variables from the grey level dependence matrix can act as recognizable features of the image. The innovative improved BP network can then recognize the pulverized and block coal images.

Performance of Feedback BP Networks

Through adding feedbacks in multi-layer BP networks, the network performance is improvedconsiderably compared with general BP network and Hopfield network, particularly the associative memorizing ability. In this paper, we analyze the two networks: feedback BP network and Hopfiled network andcompare the property between them. The conclusion shows that feedback BP network has more powerfulassociation memorizing ability than Hopfiled network.

Real-time multi-step prediction control for BP network with delay

Real time multi step prediction of BP network based on dynamical compensation of system characteristics is suggested by introducing the first and second derivatives of the system and network outputs into the network input layer, and real time multi step prediction control is proposed for the BP network with delay on the basis of the results of real time multi step prediction, to achieve the simulation of real time fuzzy control of the delayed time system.

The applying of BP network in forecasting the demand and its growth rate for coal

Based on the statistical data from 1975 to 1997, we forecast the growth rate of coal consuming and the quantity in coming decade with the BP neuron network in the article.

Optimization of Injection Molding Process of Bearing Stand Based on BP Network Method

The quality of injection plastic molded parts relates to precise geometry,smooth surface,strength,durability,and other indicators that are associated with the mold,materials,injection process,and service environment.The warpage is one of main defects of injection products,which cost much time and materials.In order to minimize warpage to ensure the precise shape of molded parts,it needs to combine design,service conditions,process parameters,material properties,and other factors in the design and manufacturing.Finite element tools and material database are used to analyze the occurrence of warpage,and analysis results contribute to the improvement and optimization of injection molding process of typical parts.To find the optimal process parameters in the solution space,experimental data are used to establish backpropagation(BP)network for predicting warpage of a bearing stand based on analysis with Moldflow.With a proper transfer function and the BP network architecture,results from the BP network method satisfiy the criteria of accuracy.The optimal solutions are searched in the BP network by the genetic algorithm with the finding that the optimization method based on the BP network is efficient.

MENDED GENETIC BP NETWORK AND APPLICATION TO ROLLING FORCE PREDICTION OF 4-STAND TANDEM COLD STRIP MILL

In order to make good use of the ability to approach any function of BP (backpropagation) network and overcome its local astringency, and also make good use of the overallsearch ability of GA (genetic algorithms), a proposal to regulate the network's weights using bothGA and BP algorithms is suggested. An integrated network system of MGA (mended genetic algorithms)and BP algorithms has been established. The MGA-BP network's functions consist of optimizing GAperformance parameters, the network's structural parameters, performance parameters, and regulatingthe network's weights using both GA and BP algorithms. Rolling forces of 4-stand tandem cold stripmill are predicted by the MGA-BP network, and good results are obtained.

Forecasting Loss of Ecosystem Service Value Using a BP Network: A Case Study of the Impact of the South-to-north Water Transfer Project on the Ecological Environmental in Xiangfan, Hubei Province, China

Objective To recognize and assess the impact of the South-to-north Water Transfer Project (SNWTP) on the ecological environment of Xiangfan, Hubei Province, situated in the water-out area, and develop sound scientific countermeasures. Methods A three-layer BP network was built to simulate topology and process of the eco-economy system of Xiangfan. Historical data of ecological environmental factors and socio-economic factors as inputs, and corresponding historical data of ecosystem service value (ESV) and GDP as target outputs, were presented to train and test the network. When predicted input data after 2001 were presented to trained network as generalization sets, ESVs and GDPs of 2002, 2003, 2004... till 2050 were simulated as output in succession. Results Up to 2050, the area would have suffered an accumulative total ESV loss of RMB 104.9 billion, which accounted for 37.36% of the present ESV. The coinstantaneous GDP would change asynchronously with ESV, it would go through an up-to-down process and finally lose RMB89.3 billion, which accounted for 18.71% of 2001. Conclusions The simulation indicates that ESV loss means damage to the capability of socio-economic sustainable development, and suggests that artificial neural networks (ANNs) provide a feasible and effective method and have an important potential in ESV modeling.

Application of genetic BP network to discriminating earthquakes and explosions

We developed a GA-BP algorithm by combining the genetic algorithm (GA) with the back propagation (BP) algorithm and established a genetic BP neural network. We also applied the BP neural network based on the BP algorithm and the genetic BP neural network based on the GA-BP algorithm to discriminate earthquakes and explosions. The obtained result shows that the discriminating performance of the genetic BP network is slightly better than that of the BP network.

基于图像处理和BP神经网络的森林防火无人机系统

对无人机设计方案、图像处理和火焰分割算法的技术原理进行了介绍,并利用BP神经网络对图像中的火焰面积变化率和火焰尖角等特征进行识别,实现了对森林火灾的快速监测。实验结果表明:系统的准确率为98.5%,比普通神经网络的84.5%更高;耗时仅22 s,比普通神经网络159 s缩短很多。这表明,BP神经网络是更可靠且更有效率的火灾识别方案。

基于改进WOA-BP神经网络的电气火灾预警算法

电气火灾是一种严重危害人员安全和财产损失的事件,因此增强对电气火灾的早期预测和预警至关重要。基于提高电气火灾预测准确性的目的,采用了改进鲸鱼算法优化BP神经网络的方法,构建了电气火灾预警模型。使用剩余电流、工作电流电压和线缆温度作为神经网络的输入特征,结合上述改进方法对权值和阈值进行优化。优化后的参数作为初始参数进行模型训练,用于输出电气火灾的概率。采用电气柜中回路数据进行试验,将预测概率与剩余电流异常持续时间进行模糊化处理,得出火灾决策。研究结果表明,所提模型相关系数达到0.97,相较于传统方法提高了0.08,具有更高的准确性和可靠性。

基于PSO-BP模糊PID的变距取苗机构控制系统设计

为满足番茄、辣椒等蔬菜作物的移栽需求,基于向下取苗原理设计了一种适用72穴和128穴两种主要番茄钵苗穴盘规格的变距取苗机构,通过建立数学模型获得了取苗机械手参数的目标函数,并利用粒子群和模拟退火混合算法对其结构参数进行优化。同时,为实现变距取苗机构的精确控制,提出了一种基于PSO-BP的模糊PID算法以提高控制精度,介绍了系统的结构与工作原理,并通过选型计算与分析建模建立了控制系统的数学模型。针对传统PID控制器稳定性差、响应速度慢等不足之处,利用PSO-BP模糊PID对控制器的参数进行在线调整,以满足控制过程中对参数的不同需求。仿真结果与试验数据的分析表明:在参数相同条件下,基于PSO-BP模糊PID控制系统系统稳定性更好、响应速度更快,具有良好的鲁棒性,提升取苗成功率的同时降低了基质损伤率,能够满足变距取苗机构高精度快速稳定控制的需求。

Analysis of Factors Related to Vasovagal Response in Apheresis Blood Donors and the Establishment of Prediction Model Based on BP Neural Network Algorithm

Objective:To analyze the factors related to vessel vasovagal reaction(VVR)in apheresis donors,establish a mathematical model for predicting the correlation factors and occurrence risk,and use the prediction model to intervene in high-risk VVR blood donors,improve the blood donation experience,and retain blood donors.Methods:A total of 316 blood donors from the Xi'an Central Blood Bank from June to September 2022 were selected to statistically analyze VVR-related factors.A BP neural network prediction model is established with relevant factors as input and DRVR risk as output.Results:First-time blood donors had a high risk of VVR,female risk was high,and sex difference was significant(P value<0.05).The blood pressure before donation and intergroup differences were also significant(P value<0.05).After training,the established BP neural network model has a minimum RMS error of o.116,a correlation coefficient R=0.75,and a test model accuracy of 66.7%.Conclusion:First-time blood donors,women,and relatively low blood pressure are all high-risk groups for VVR.The BP neural network prediction model established in this paper has certain prediction accuracy and can be used as a means to evaluate the risk degree of clinical blood donors.

Study on Remote Sensing of Water Depths Based on BP Artificial Neural Network

A momentum BP neural network model （MBPNNM） was constructed to retrieve the water depth information for the South Channel of the Yangtze River Estuary using the relationship between the reflectance derived from Landsat 7 satellite data and the water depth information. Results showed that MBPNNM, which exhibited a strong capability of nonlinear mapping, allowed the water depth information in the study area to be retrieved at a relatively high level of accuracy. Affected by the sediment concentration of water in the estuary, MBPNNM enabled the retrieval of water depth of less than 5 meters accurately. However, the accuracy was not ideal for the water depths of more than 10 meters.

Seabed Classification Using BP Neural Network Based on GA

Side scan sonar imaging is one of the advanced methods for seabed study. In order to be utilized in other projects, such as ocean engineering, the image needs to be classified according to the distributions of different classes of seabed materials. In this paper, seabed image is classified according to BP neural network, and. Genetic Algorithm is adopted in train network in this paper. The feature vectors are average intensity, six statistics of texture and two dimensions of fractal. It considers not only the spatial correlation between different pixels, but also the terrain coarseness. The texture is denoted by the statistics of the co-occurrence matrix. Double Blanket algorithm is used to calculate dimension. Because a uniform fractal may not be sufficient to describe a seafloor, two dimensions are calculated respectively by the upper blanket and the lower blanket. However, in sonar image, fractal has directivity, i. e. there are different dimensions in different direction. Dimensions are different in acrosstrack and alongtrack, so the average of four directions is used to solve this problem. Finally, the real data verify the algorithm. In this paper, one hidden layer including six nodes is adopted. The BP network is rapidly and accurately convergent through GA. Correct classification rate is 92.5 % in the result.

An exploration of the uncertainty relation satisfied by BP network learning ability and generalization ability

This paper analyses the intrinsic relationship between the BP network learning ability and generalization ability and other influencing factors when the overfit occurs, and introduces the multiple correlation coefficient to describe the complexity of samples; it follows the calculation uncertainty principle and the minimum principle of neural network structural design, provides an analogy of the general uncertainty relation in the information transfer process, and ascertains the uncertainty relation between the training relative error of the training sample set, which reflects the network learning ability, and the test relative error of the test sample set, which represents the network generalization ability; through the simulation of BP network overfit numerical modeling test with different types of functions, it is ascertained that the overfit parameter q in the relation generally has a span of 7×10-3 to 7×10-2; the uncertainty relation then helps to obtain the formula for calculating the number of hidden nodes of a network with good generalization ability under the condition that multiple correlation coefficient is used to describe sample complexity and the given approximation error requirement is satisfied; the rationality of this formula is verified; this paper also points out that applying the BP network to the training process of the given sample set is the best method for stopping training that improves the generalization ability.

The Applicative Investigation of Adaptive BP Networks for Multi-user Detection in Asynchronous DS-CDMA Mobile Communications

Three-layer Adaptive Back-Propagation Neural Networks(TABPNN) are employed for the demodulation of spread spectrum signals in a multiple-access environment. A configuration employing three-layer adaptive Back-propagation neural networks is put forward for the demodulation of spread-spectrum signals in asynchronous Gaussian channels. The theoretical arguments and practical performance based on the neural networks are analyzed. The results show that whether the resistance to the multiple access interference or the robust to near-far effects, the proposed detector significantly outperforms not only the conventional detector but also the BP neural networks detector and is comparable to the optimum detector.

Quantitative Detection Model of Pernicious Gases in Pig House Based on BP Neural Network

To find a neural network model suitable to identify the concentration of mixed pernicious gases in pig house, the quantitative detection model of pernicious gases in pig house was set up based on BP （ Back propagation） neural network. The BP neural network was trained separately by the three functions, trainbr, traingdm and trainlm, in order to identify the concentration of mixed pernicious gases composed of ammonia gas and hepatic gas. The neural network toolbox in MATLAB software was used to simulate the detection. The results showed that the neural network trained by trainbr function has high average identification accuracy and faster detection speed, and it is also insensitive to noise; therefore, it is suitable to identify the concentration of pemidous gases in pig house. These data provide a reference for intelligent monitoring of pemicious gases in pigsty.