基于BP-S混合模型的大坝安全监测技术研究

大坝安全监控是掌握大坝安全运行性态的重要手段,是指导大坝安全管理的的科学依据,而传统的统计模型在处理复杂的非线性问题存在一些缺陷.本文结合统计模型,介绍了BP神经网络优化算法,建立了神经网络与统计模型的混合模型,通过实例计算表明,采用该模型在模拟及预测监测数据方面精度明显提高,具有推广价值.

基于神经网络的马尔可夫预测模型

首次提出神经网络与马尔可夫链相结合的数学模型,用于随机波动数据序列变形预测,克服随机观测数据固有的随机和波动特征,较好地实现了工程中大量随机波动性数据无法采用灰色GM(1,1)模型精确预测问题。采用神经网络完成观测数据变化动态基准模型的计算,在此基础上应用马尔可夫链确定系统状态转移概率矩阵,通过系统状态的划分、样本值与模型拟合值之间的残差及其中误差等指标的分析,完成观测数据变化值的准确计算。该模型被应用于隧道围岩工程实例,计算证明取得了较好的效果。

Characterization of grain growth behaviors by BP-ANN and Sellars models for nickle-base superalloy and their comparisons

In order to deeply understand the grain growth behaviors of Ni80A superalloy,a series of grain growth experiments were conducted at holding temperatures ranging from 1223 to 1423 K and holding time ranging from 0 to 3600 s.A back-propagation artificial neural network(BP-ANN)model and a Sellars model were solved based on the experimental data.The prediction and generalization capabilities of these two models were evaluated and compared on the basis of four statistical indicators.The results show that the solved BP-ANN model has better performance as it has higher correlation coefficient(r),lower average absolute relative error(AARE),lower absolute values of mean value(μ)and standard deviation(ω).Eventually,a response surface of average grain size to holding temperature and holding time is constructed based on the data expanded by the solved BP-ANN model,and the grain growth behaviors are described.

Mountain ground movement prediction caused by mining based on BP-neural network

Six main influencing factors： slope, aspect, distance, angle, angle of coal seam, and the ratio of depth and thickness, were selected by Grey correlation theory and Grey relational analysis procedure programmed by the MATLAB software package to select the surface movement and deformation parameters. On this basis, the paper built a BP neural network model that takes the six main influencing factors as input data and corresponding value of ground subsidence as output data. Ground subsidence of the 3406 mining face in Haoyu Coal was predicted by the trained BP neural network. By comparing the prediction and the practices, the research shows that it is feasible to use the 13P neural network to predict mountain mining subsidence.

Research on BP-ANN Model of Semi-rigid Connection

The beam-to-column semirigid connection in a steel frame structure is represented by a zero-length rotational spring at the end of the beam element. The beam-to-column semirigid connection behavior is represented by its moment-rotation relationship. Several traditional mathematical models have been proposed to fit the moment-rotation curves from the experimental database,but they may be more reliable within certain ranges. In this paper, the intellectualized analytical model is proposed in the semirigid connections for top and seat angles with double web angles using the feed-forward back-propagation artificial neural network （BP-ANN） technique. the intellectualized analytical model from experimental results based on BP-ANN is more reliable and it is a better choice to the moment-rotation curves for beam-to-column semirigid connection. The results are found to provide effectiveness to the experimental response that is satisfactory for use in steel structural engineering design.

基于改进编码的多聚脯氨酸二型结构预测

目前使用机器学习方法对稀有的PPII二级结构进行预测工作还比较少见。本文在分析数据集的特征基础之上,采用加强局部信息的改进编码方法,并使用生物信息学中常用的BP神经网络来预测PPII二级结构。通过对不同输入窗口长度与不同隐层节点数的神经网络进行训练和测试,得出在输入窗口长度为5个氨基酸残基和隐结点数为8时PPII螺旋结构能够被很好地预测出来,此时的预测敏感度为82.5%。

基于神经网络的多聚脯氨酸二型结构预测

使用“机器学习方法”对稀有的多聚脯氨酸二型(PPII)的二级结构进行预测,在预处理蛋白质序列的基础上,使用生物信息学中常用的 BP神经网络预测 PPII二级结构.通过对不同输入窗口长度与不同隐层节点数的神经网络进行训练和测试,得出在输入窗口长度为 13 个氨基酸残基和隐节点数为15时预测效果最好,此时的预测精度可达73.8%.

Combining the genetic algorithms with artificial neural networks for optimization of board allocating

This paper introduced the Genetic Algorithms (GAs) and Artificial Neural Networks (ANNs), which have been widely used in optimization of allocating. The combination way of the two optimizing algorithms was used in board allocating of furniture production. In the experiment, the rectangular flake board of 3650 mm 1850 mm was used as raw material to allocate 100 sets of Table Bucked. The utilizing rate of the board reached 94.14 % and the calculating time was only 35 s. The experiment result proofed that the method by using the GA for optimizing the weights of the ANN can raise the utilizing rate of the board and can shorten the time of the design. At the same time, this method can simultaneously searched in many directions, thus greatly in-creasing the probability of finding a global optimum.

Application of BP neural network model with fuzzy optimization in retrieval of biomass parameters

The retrieval of the biomass parameters from active/passive microwave remote sensing data （10.2 GHz） is performed based on an iterative inversion of BP neural network model with fuzzy optimization. The BP neural network is trained by a set of the measurements of active and passive remote sensing and the ground truth data versus Day of Year during growth. Once the network training is complete, the model can be used to retrieve the temporal variations of the biomass parameters from another set of observation data. The model was used in weights and microware observation data of wheat growth in 1989 to retrieve biomass parameters change of wheat growth this year. The retrieved biomass parameters correspond well with the real data of the growth, which shows that the BP model is scientific and sound.

TIME SERIES NEURAL NETWORK MODEL FOR HYDROLOGIC FORECASTING

Time series analysis plays an important role in hydrologic forecasting,while the key to this analysis is to establish a proper model.This paper presents a time series neural network model with back propagation procedure for hydrologic forecasting.Free from the disadvantages of previous models,the model can be parallel to operate information flexibly and rapidly.It excels in the ability of nonlinear mapping and can learn and adjust by itself,which gives the model a possibility to describe the complex nonlinear hydrologic process.By using directly a training process based on a set of previous data, the model can forecast the time series of stream flow.Moreover,two practical examples were used to test the performance of the time series neural network model.Results confirm that the model is efficient and feasible.

Neural network based method for compensating model error

Two traditional methods for compensating function model errors, the method of adding systematic parameters and the least-squares collection method, are introduced. A proposed method based on a BP neural network （called the H-BP algorithm） for compensating function model errors is put forward. The function model is assumed as y =f（x1, x2,… ,xn）, and the special structure of the H-BP algorithm is determined as （ n ＋ 1） ×p × 1, where （n ＋ 1） is the element number of the input layer, and the elements are xl, x2,…, xn and y＇ （ y＇ is the value calculated by the function model）; p is the element number of the hidden layer, and it is usually determined after many tests; 1 is the dement number of the output layer, and the element is △y = y0-y＇（y0 is the known value of the sample）. The calculation steps of the H-BP algorithm are introduced in detail. And then, the results of three methods for compensating function model errors from one engineering project are compared with each other. After being compensated, the accuracy of the traditional methods is about ± 19 mm, and the accuracy of the H-BP algorithm is ± 4. 3 mm. It shows that the proposed method based on a neural network is more effective than traditional methods for compensating function model errors.

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.

Pure quasi-P wave equation and numerical solution in 3D TTI media

Based on the pure quasi-P wave equation in transverse isotropic media with a vertical symmetry axis （VTI media）, a quasi-P wave equation is obtained in transverse isotropic media with a tilted symmetry axis （TTI media）. This is achieved using projection transformation, which rotates the direction vector in the coordinate system of observation toward the direction vector for the coordinate system in which the z-component is parallel to the symmetry axis of the TTI media. The equation has a simple form, is easily calculated, is not influenced by the pseudo-shear wave, and can be calculated reliably when δ is greater than ε. The finite difference method is used to solve the equation. In addition, a perfectly matched layer （PML） absorbing boundary condition is obtained for the equation. Theoretical analysis and numerical simulation results with forward modeling prove that the equation can accurately simulate a quasi-P wave in TTI medium.

Coal mine safety production forewarning based on improved BP neural network

Firstly, the early warning index system of coal mine safety production was given from four aspects as per- sonnel, environment, equipment and management. Then, improvement measures which are additional momentum method, adaptive learning rate, particle swarm optimization algorithm, variable weight method and asynchronous learning factor, are used to optimize BP neural network models. Further, the models are applied to a comparative study on coal mine safety warning instance. Results show that the identification precision of MPSO-BP network model is higher than GBP and PSO-BP model, and MPSO- BP model can not only effectively reduce the possibility of the network falling into a local minimum point, but also has fast convergence and high precision, which will provide the scientific basis for the forewarnin~ management of coal mine safetv production.

BP neural network model on the forecast for blasting vibrating parameters in the course of hole-by-hole detonation

According to the neural network theory, combined with the technical characteristicsof the hole-by-hole detonation technology, a BP network model on the forecast forblasting vibration parameters was built.Taking the deep hole stair demolition in a mine asan experimental object and using the raw information and the blasting vibration monitoringdata collected in the process of the hole-by-hole detonation, carried out some training andapplication work on the established BP network model through the Matlab software, andachieved good effect.Also computed the vibration parameter with the empirical formulaand the BP network model separately.After comparing with the actual value, it is discoveredthat the forecasting result by the BP network model is close to the actual value.

Research on the mining roadway displacement forecasting based on support vector machine theory

In view of the difficulty in supporting the surrounding rocks of roadway 3-411 ofFucun Coal Mine of Zaozhuang Mining Group, a deformation forecasting model was putforward based on particle swarm optimization.The kernel function and model parameterswere optimized using particle swarm optimization.It is shown that the forecast result isvery close to the real monitoring data.Furthermore, the PSO-SVM (Particle Swarm Optimization-Support Vector Machine) model is compared with the GM(1,1) model and L-M BPnetwork model.The results show that PSO-SVM method is better in the aspect of predictionaccuracy and the PSO-SVM roadway deformation pre-diction model is feasible for thelarge deformation prediction of coal mine roadway.

Energy-absorption forecast of thin-walled structure by GA-BP hybrid algorithm

In order to analyze the influence rule of experimental parameters on the energy-absorption characteristics and effectively forecast energy-absorption characteristic of thin-walled structure, the forecast model of GA-BP hybrid algorithm was presented by uniting respective applicability of back-propagation artificial neural network （BP-ANN） and genetic algorithm （GA）. The detailed process was as follows. Firstly, the GA trained the best weights and thresholds as the initial values of BP-ANN to initialize the neural network. Then, the BP-ANN after initialization was trained until the errors converged to the required precision. Finally, the network model, which met the requirements after being examined by the test samples, was applied to energy-absorption forecast of thin-walled cylindrical structure impacting. After example analysis, the GA-BP network model was trained until getting the desired network error only by 46 steps, while the single BP-ANN model achieved the same network error by 992 steps, which obviously shows that the GA-BP hybrid algorithm has faster convergence rate. The average relative forecast error （ARE） of the SEA predictive results obtained by GA-BP hybrid algorithm is 1.543%, while the ARE of the SEA predictive results obtained by BP-ANN is 2.950%, which clearly indicates that the forecast precision of the GA-BP hybrid algorithm is higher than that of the BP-ANN.

Kinematics modeling of a 6-PSS parallel mechanism with wide-range flexure hinges

A novel 6-PSS flexible parallel mechanism was presented,which employed wide-range flexure hinges as passive joints.The proposed mechanism features micron level positioning accuracy over cubic centimeter scale workspace.A three-layer back-propagation(BP) neural network was utilized to the kinematics analysis,in which learning samples containing 1 280 groups of data based on stiffness-matrix method were used to train the BP model.The kinematics performance was accurately calculated by using the constructed BP model with 19 hidden nodes.Compared with the stiffness model,the simulation and numerical results validate that BP model can achieve millisecond level computation time and micron level calculation accuracy.The concept and approach outlined can be extended to a variety of applications.

Regression model for daily passenger volume of high-speed railway line under capacity constraint

A non-linear regression model is proposed to forecast the aggregated passenger volume of Beijing-Shanghai high-speed railway(HSR) line in China. Train services and temporal features of passenger volume are studied to have a prior knowledge about this high-speed railway line. Then, based on a theoretical curve that depicts the relationship among passenger demand, transportation capacity and passenger volume, a non-linear regression model is established with consideration of the effect of capacity constraint. Through experiments, it is found that the proposed model can perform better in both forecasting accuracy and stability compared with linear regression models and back-propagation neural networks. In addition to the forecasting ability, with a definite formation, the proposed model can be further used to forecast the effects of train planning policies.

SVM model for estimating the parameters of the probability-integral method of predicting mining subsidence

A new mathematical model to estimate the parameters of the probability-integral method for mining subsidence prediction is proposed.Based on least squares support vector machine(LS-SVM) theory, it is capable of improving the precision and reliability of mining subsidence prediction.Many of the geological and mining factors involved are related in a nonlinear way.The new model is based on statistical theory(SLT) and empirical risk minimization(ERM) principles.Typical data collected from observation stations were used for the learning and training samples.The calculated results from the LS-SVM model were compared with the prediction results of a back propagation neural network(BPNN) model.The results show that the parameters were more precisely predicted by the LS-SVM model than by the BPNN model.The LS-SVM model was faster in computation and had better generalized performance.It provides a highly effective method for calculating the predicting parameters of the probability-integral method.