基于C++Builder的凝汽器故障诊断系统

Fault Diagnosis System of Condenser based on C++ Builder

为了有效地对电厂凝汽器进行故障诊断并给予操作指导,开发了凝汽器故障诊断系统。该故障诊断系统分别采用标准BP算法和改进BP变步长附加动量法进行网络的学习训练。以凝汽器故障知识库为依据,构造了17个输入层节点、13个隐含层节点、12个输出层节点的三层反向传播网络。经过对比,说明采用经过改进BP变步长附加动量法进行网络的学习训练,能够快速收敛,误差小于0.01。实例数据测试表明,训练的网络能够得到较为准确的故障诊断效果。软件系统的设计选用C++Builder6.0快速开发工具,该软件系统具有友好的用户界面,操作简单。

In order to accomplish the fault diagnosis of condenser efficiently,a condenser fault diagnosis system with operation guide is proposed.The Back-propagation network is training by a standard algorithm and an improved algorithm with variable step size adding momentum.According to the condenser fault database,the three-layer Back-propagation network with 17 input nodes,13 middle nodes and 12 output nodes is constructed. On the basis of comparison,the Neural Network uses variable step size adding momentum for training has quick convergence rate,and its error less than 0.01.Practical data testing shows that by training with variable step size adding momentum algorithm,the Neural Network can get accurate results.The whole software system is developed with C＋＋ Builder6.0.Its interface is friendly and simple.