基于概率神经网络(PNN)的汽轮发电机组故障诊断
1.无需注册登录,支付后按照提示操作即可获取该资料.
2.资料以网页介绍的为准,下载后不会有水印.资料仅供学习参考之用.
密 惠 保
基于概率神经网络(PNN)的汽轮发电机组故障诊断(论文14000字)
摘要:汽轮发电机组变得越来越智能,功能也增强,但不确定性因素和不确定性信息仍然大量存在。为保证汽轮机的安全性,应着重解决它的这些问题。传统的BP神经网络不能有效的识别汽轮发电机组的不确定性,而且,故障出现的误差结果也没有很好的参考价值。因此,诊断的准确性也有待提高。针对汽轮发电机组出现的问题,出现了基于概率神经网络(PNN)的汽轮发电机组故障诊断。PNN优点多,机器学习算法简易,方便训练,相比于传统的样本处理方式,PNN对训练样本训练,并引入训练网络,更好的确保了诊断结果的正确和可信程度。在MATLAB仿真结果表明,PNN在保证诊断结果准确率的基础上,速度加快,分类性能大大提高,诊断效率也大大提高。
关键词:故障诊断;汽轮发电机组;神经网络;概率神经网络
Fault Diagnosis of Turbo-generator Based on Probabilistic Neural Network
Abstract:Turbo-generator units are becoming increasingly intelligent and functional.But uncertainty and uncertain information exist. In order to ensure the safety of turbines, the uncertainties are solved first. Because the traditional BP neural network has limited ability to recognize this kind of uncertainty, and the error convergence rate is slow, the diagnosis accuracy is low. In order to ensure the accuracy of fault diagnosis of turbo - generator, a fault diagnosis of turbo - generator based on probabilistic neural network (PNN) is proposed. PNN has many advantages, and the machine learning algorithm is sample and convenient to train. Compared with the traditional sample processing method, PNN training sample and introduce to training network, so as to ensure the correct and reliable diagnosis result. The results of MATLAB simulation show that PNN ensure the accuracy of diagnostic on the basis of speed, classification performance greatly improved diagnostic efficiency is also greatly improved. [资料来源:http://THINK58.com]
Keywords:fault diagnosis; turbo-generator sets; neural network; probabilistic neural network
[资料来源:THINK58.com]
目录
1.绪论 1
1.1 选题背景及意义 1
1.2 国内外的故障诊断技术研究现状 1
1.3 故障诊断原理及发展 3
1.4 本文的主要研究内容 4
2.神经网络 5
2.1 神经网络的概述 5
2.2 神经网络的基本原理 5
2.3 神经网络的结构与特性 6
2.3.1 神经网络的结构 6
2.3.2 神经网络的特性 7
2.4 神经网络在汽轮发电机组故障诊断中的应用 7
[资料来源:http://THINK58.com]
3. 基于反向传播(BP)神经网络的汽轮发电机组故障诊断 8
3.1 BP神经网络模型 8
3.2 BP神经网络的学习流程 9
3.3 基于BP神经网络的汽轮发电机组故障诊断 10
4. 基于概率神经网络(PNN)的汽轮发电机组故障诊断 14
4.1 概率神经网络模型 14
4.2 概率神经网络模型的理论与方法 15
4.2.1 Bayes分类 15
4.2.2 Parzen窗口法 16
4.3 概率神经网络的数学描述 17
4.4 基于概率神经网络的汽轮发电机组故障诊断 17
5. 总结与展望 23
附录 26
致谢 30