基于GA-PSO-BP 的大坝变形监测模型

提出了一种基于GA-PSO 混合优化BP 神经网络的大坝变形监测模型, 将遗传算法(GA)和粒子群算法(PSO)的寻优过程进行融合, 利用GA 算法的全局性和PSO 算法收敛速度快的特点,通过迭代选取最优的粒子作为BP神经网络的连接权值和阈值,以减小网络输出误差, 提高其收敛速度和加强网络泛化能力。运用GA-PSO-BP 模型对大坝自动监测数据进行预测分析, 实验结果表明GA-PSO-BP 模型优化了BP 神经网络的连接权值和阈值, 能有效提高网络训练精度与收敛速度, 有效避免早熟收敛, 使模型的整体预测效果得到提高。     

粒子群与遗传算法优化支持向量机的应用

为了提高参数优化精度,研究将粒子群算法与支持向量机相结合,建立基于粒子群算法的支持向量机复杂过程系统稳态模型。在此基础上,为解决粒子群算法容易出现早熟收敛、搜索精度不高、在迭代的后期效率低、容易陷入局部极优点等问题,提出了引入遗传算法的改进粒子群算法。通过利用改进后的粒子群算法对支持向量机参数进行优化,并应用到青霉素发酵这一复杂工业系统。仿真结果表明,改进算法提高了工业产量,实现了对系统结果的优化。     

灰关联与混合模型相结合在大坝变形中的应用

在大坝变形监测分析中采用灰关联分析法与混合模型相结合的方法,利用有限元法与实测值进行优化拟合而建立模型,从混合模型的分析结果可以看出,大坝时效位移虽有增加,但增加量在逐年减小.因此,水平位移的变化规律总体上正常.     

参数优化支持向量机的人参价格预测模型

为了对人参价格进行预测,分析了影响人参价格因素,通过K-fold交叉验证方法,利用粒子群算法对支持向量机的惩罚参数c和ggamma值进行寻优,建立起2010年1月～2011年12月林下参的价格预测模型。利用粒子群算法优化惩罚参数c为3.6974,利用radial basis function核函数的SVM（Support Vector Machine）对预测集1的预测相关系数为97.316%。     

支持向量机在水库大坝变形中的预测分析

简单介绍支持向量机(Support Vector Machine,SVM)的基本原理,并用该原理建立用于大坝变形监测的SVM模型.通过实例的验算和对比分析,验证SVM模型在大坝监测中比BP神经网络模型、多项式回归模型在处理非线性、小样本、高维线性化等问题上具有更明显的优势.     

基于支持向量机的进化神经网络集成股市模型

利用粒子群优化算法的全局搜索功能,进化设计神经网络的网络结构与连接权,得到一组独立的神经网络集成个体.利用主成份分析法提取其综合信息,再用支持向量机回归方法对其处理,生成神经网络的输出结果,以此建立股市预测模型.通过实例验证,该方法能有效提高神经网络集成的泛化能力,模型的预测精度高、稳定性好、具有应用推广前景.     

深基坑变形预测的进化支持向量机模型

提出了深基坑变形预测的进化支持向量机方法。利用遗传算法来搜索支持向量机与核函数的参数,避免了人为选择参数的盲目性,同时提高了支持向量机的推广预测能力。利用优化后的模型对基坑实例进行了变形预测,并将预测结果与监测结果进行了对比。研究结果表明,该模型与神经网络模型相比,具有预测精度高、泛化能力强等优点,对基坑安全监控具有实用价值。     

大坝监测空间位移统计模型研究

突破大坝监测资料分析中各方向位移单独建模分析的局限,综合测点三方向位移测值信息建立了的空间位移拟合模型,对正确描述,预测预测点的总体变形提供了可行途径。     

基于支持向量机的城市电网空间负荷预测方法

提出了一种基于支持向量机(Support Vector Machine,SVM)的城市电网空间负荷预测(Spatial Load forecasting,SLF)方法。该方法首先以等大小网格划分的规则生成元胞,并获取元胞历年负荷;然后将各元胞历年负荷最大值及其对应的年份输入支持向量机预测模型进行训练,其中采用粒子群优化(Particle Swarm Optimization,PSO)算法寻求预测模型的最优参数,预测各元胞目标年负荷最大值,从而实现空间负荷预测;最后对吉林市城市电网进行实例分析,结果验证了该方法的实用性和有效性。     

Game model of safety monitoring for arch dam deformation

Arch dam deformation is comprehensively affected by water pressure,temperature,dam's structural behavior and material properties as well as other factors.Among them the water pressure and temperature are external factors(source factors) that cause dam deformation,and dam's structural behavior and material properties are the internal factors of deformation(resistance factors).The dam deformation is the result of the mutual game playing between source factors and resistance factors.Therefore,resistance factor...     

Game model of safety monitoring for arch dam deformation

Arch dam deformation is comprehensively affected by water pressure, temperature, dam’s structural behavior and material properties as well as other factors. Among them the water pressure and temperature are external factors (source factors) that cause dam deformation, and dam’s structural behavior and material properties are the internal factors of deformation (resistance factors). The dam deformation is the result of the mutual game playing between source factors and resistance factors. Therefore, resistance factors of structure and materials that reflect resistance character of arch dam structure are introduced into the traditional model, where structure factor is embodied by the flexibility coefficient of dam body and the maximum dam height, and material property is embodied by the elastic modulus of dam. On the basis of analyzing the correlation between dam deformation and resistance factors, the game model of safety monitoring for arch dam deformation is put forward.     

Game model of safety monitoring for arch dam deformation

Arch dam deformation is comprehensively affected by water pressure, temperature, dam’s structural behavior and material properties as well as other factors. Among them the water pressure and temperature are external factors (source factors) that cause dam deformation, and dam’s structural behavior and material properties are the internal factors of deformation (resistance factors). The dam deformation is the result of the mutual game playing between source factors and resistance factors. Therefore, resistance factors of structure and materials that reflect resistance character of arch dam structure are introduced into the traditional model, where structure factor is embodied by the flexibility coefficient of dam body and the maximum dam height, and material property is embodied by the elastic modulus of dam. On the basis of analyzing the correlation between dam deformation and resistance factors, the game model of safety monitoring for arch dam deformation is put forward.

     

Oil monitoring methods based on information theory

To evaluate the wear condition of machines accurately,oil spectrographic entropy,mutual information and ICA analysis methods based on information theory are presented. A full-scale diagnosis utilizing all channels of spectrographic analysis can be obtained. By measuring the complexity and correlativity,the characteristics of wear condition of machines can be shown clearly. The diagnostic quality is improved. The analysis processes of these monitoring methods are given through the explanation of examples. The availability of these methods is validated and further research fields are demonstrated.     

Assessment model of dam operation risk based on monitoring data

Although the dams produce remarkable social and economic benefits, the threat made by unsafe dams to the life and property of people who live in the lower river area is un-negligible. Based on the monitoring data which reflect the safety condition of dams, the risk degree concept is proposed and the analysis system and model for evaluating risk degree (rate) are established in this paper by combining the reliability theory and field monitoring data. The analysis method for risk degree is presented based on Bayesian approach. A five-grade risk degree system for dam operation risk and corresponding risk degree is put forward according to the safety condition of dams. The operation risks of four cascade dams on some river are analyzed by the model and approach presented here and the result is adopted by the owner.     

Assessment model of dam operation risk based on monitoring data

Although the dams produce remarkable social and economic benefits,the threat made by unsafe dams to the life and property of people who live in the lower river area is un-negligible.Based on the monitoring data which reflect the safety condition of dams,the risk degree concept is proposed and the analysis system and model for evaluating risk degree (rate) are established in this paper by combining the reliability theory and field monitoring data.The analysis method for risk degree is presented based on Bayesian approach.A five-grade risk degree system for dam operation risk and corresponding risk degree is put forward according to the safety condition of dams.The operation risks of four cascade dams on some river are analyzed by the model and approach presented here and the result is adopted by the owner.     

大坝监测空间位移统计模型研究

突破大坝监测资料分析中各方向位移单独建模分析的局限,综合测点三方向位移测值信息建立了合理的空间位移拟合模型,对正确描述、预报测点的总体变形提供了可行途径．     

基于多传感器信息融合的大坝监测数据分析

大坝中的传感器(测点)在信息上存在相互印证关系.首先通过对具有相似变化规律的多个测点的监测数据进行标准化处理,将不同测点在同一时间的一次观测近似看作同一测点在相同观测条件下的重复观测.然后,运用常规统计理论,在大坝安全监控专家系统中实现异常值检测以及异常趋势识别.从而,及时发现大坝潜在的问题,为大坝安全监控提供可靠依据.     

大坝监测数学模型若干问题的研究

基于事件出现频率一致收敛于其概率理论,对大坝监测数学模型类数和模型的优选及样本数、样本点的分布对模型性能的影响做了研究,并在概率意义上给出了确保所有模型性能的样本数确定方法,还就研究结果给出了算例．     

基于混沌优化支持向量机的大坝安全监控预测

首先介绍了基于统计学习理论的一种新的机器学习技术——支持向量机(Support Vector Machine,SVM),并针对目前支持向量机参数选择时人为选择的盲目性,将具有良好优化性能的混沌优化(Chaos Optimi-zation)技术应用到支持向量机惩罚函数和核函数参数的优化,提出了混沌优化支持向量机(Chaos Optimization Support Vector Machine,COSVM)方法.根据丰满大坝1997-2004年的实际监测数据,建立了混沌优化支持向量机大坝安全监控预测模型,进行了与统计回归模型和BP神经网络模型的分析比较,结果表明,COSVM模型具有更高的预测精度,同时在较长时段的预测中,COSVM模型也表现出更好的泛化推广性能.