论文部分内容阅读
在基于支持向量机的故障预报过程中,故障样本数据的不相关变量会影响支持向量机的性能;加权支持向量机中直接选择加权系数存在很多不足;支持向量机参数主要凭人的经验或通过多次实验获得,还没有一个确定而有效的方法。针对这三种问题,提出了采用改进的人工鱼群算法将特征选择、加权系数、支持向量机参数进行并行优化的方法,并将此方法应用于船舶动力装置冷凝器的故障预报中。仿真结果表明:相对于单独优化,并行优化能够在更短的时间内进行最有效的故障特征提取,并且提高支持向量机的性能;相对于遗传算法,改进人工鱼群算法能够以更快的速度达到最终的优化结果。
In the process of fault prediction based on SVM, the irrelevant variables of the fault sample data will affect the performance of SVM. There are many deficiencies in the direct selection of weighting factors in SVM. The parameters of SVM mainly depend on human experience Obtained many experiments, there is no a sure and effective method. In view of these three problems, a method of optimizing the parameters of feature selection, weighting coefficient and support vector machine by using the improved artificial fish-swarm algorithm is proposed. The method is applied to the fault prediction of condenser of marine power plant. The simulation results show that compared with single optimization, parallel optimization can extract the most effective fault features in a shorter time and improve the performance of SVM. Compared with genetic algorithm, the improved artificial fish swarm algorithm can be used at a faster speed Achieve the final optimization result.