基于改进粒子群优化支持向量机的水利闸门卡阻故障诊断

Fault diagnosis of gates based on improved particle swarm optimizing support vector machine

针对水利闸门卡阻故障诊断主要依赖专家经验、故障样本少且难以自动识别的特点,提出了一种基于改进粒子群算法优化最小二乘支持向量机的水利闸门卡阻故障诊断方法。该方法以闸门的开度值、左右两侧荷重值及上下游水位差作为输入向量,通过最小二乘支持向量机进行故障分类,通过改进粒子群算法对支持向量机的参数进行优化。并将建立的诊断模型制作成Labview子VI程序模块,嵌入到闸门调度监控软件中,实现了闸门卡阻故障的快速自动诊断。通过在淮洪新河某泄洪闸实地实验及应用表明,该诊断程序具有较高的诊断准确率和诊断效率。

In order to overcome the disadvantages that gates’fault diagnosis mainly depends on experts’ experience,with less samples and difficult to recognize automatically, a method based on improved particle swarm optimizing support vector machine is proposed.The input vector of support vector machine is composed of gate opening value, water-head and load weight of both sides. The fault diagnosis model can be made into a Labview sub VI module which needs to be embedded into the gate monitoring and controlling software finally, to achieve fast and automatic fault diagnosis of gates.It shows that this fault diagnosis model has many advantages such as high accuracy and efficiency by using it at one sluice of Huaihong New River.