基于IPSO优化LSSVM的水轮发电机组振动故障诊断

提出改进的粒子群(IPSO)算法，并与最小二乘支持向量机(LSSVM)相结合，得到基于IPSO-LSSVM的水轮发电机组故障诊断方法。改进后的粒子群算法能较好地调整算法在全局与局部搜索能力之间的平衡，将其应用于LSSVM的参数优化，可以提高故障诊断的精度和效率。实例分析结果证明，IPSO-LSSVM模型不仅能够取得良好的分类效果，而且诊断速度与精度均高于采用BP神经网络、LSSVM以及PSO-LSSVM等方法，适合在实际工程中应用。

Vibration fault diagnosis of hydroelectric generating unit by Least Squares Support Vector Machine based on Improved Particle Swarm Optimization

In order to diagnosis the vibration faults of Hydroelectric Generating Unit quickly and accurately, an Improved Particle Swarm Optimization (IPSO) algorithm is proposed. The new method of fault diagnosis has been found by combining it with the Least Squares Support Vector Machine (LSSVM) to form Improved Particle Swarm Optimization and Least Squares Support Vector Machine (IPSO-LSSVM) algorithm. IPSO algorithm can adjust the balance between global and local search capabilities suitably, and which optimize parameters of LSSVM in order to improve the precision and efficiency of the faults diagnose. By discussing the experiment results, the IPSO-LSSVM method has very good classification results, and the precision and rate of diagnostic is better than BP network, LSSVM and PSO-LSSVM. Consequently, the IPSO-LSSVM model is a proper alternative for vibration fault diagnosis of hydroelectric generating unit.