改进和声搜索算法优化支持向量机的柴油机故障诊断研究

发布时间：2018-11-08 19:56

【摘要】：针对基本和声搜索算法在优化支持向量机参数时,其局部搜索能力不足且后期收敛速度比较慢的缺点,提出利用改进和声搜索算法对支持向量机相关参数进行选择优化(IHS-SVM)的方法。在这一方法中,将原算法中控制参数—记忆库取值概率(HMCR)、微调概率(PAR)和调节宽度(bw)由静态值改进为随迭代次数的不同而进行动态变化。通过对UCI中的2个数据集进行分类正确率测试,并与未优化的支持向量机(SVM)和基本和声算法优化的支持向量机(HS-SVM)测试结果对比,证明了该改进方法的优越性。最后,将其用于柴油机故障诊断,并将分类正确率与未优化SVM和HSSVM分类结果进行比较,进一步说明改进和声搜索算法优化的支持向量机(IHS-SVM)能获得更高的分类结果正确率,即证明了该改进方法的实用性。
[Abstract]:In order to solve the problem that the local search ability of the basic harmonic search algorithm is insufficient and the convergence speed is slow in the later stage when the parameters of support vector machine are optimized, An improved harmonic search algorithm is proposed to select and optimize the parameters of support vector machines (IHS-SVM). In this method, the control parameter-memory library value probability (HMCR), fine-tuning probability (PAR) and the adjusting width (bw) are improved from static value to dynamic change with the number of iterations in the original algorithm. By testing the classification accuracy of two data sets in UCI, and comparing the results with the unoptimized support vector machine (SVM) and the basic harmony algorithm optimized support vector machine (HS-SVM), the superiority of the improved method is proved. Finally, it is applied to diesel engine fault diagnosis, and the classification accuracy is compared with the results of unoptimized SVM and HSSVM classification. It is further proved that the improved harmonic search algorithm optimized by support vector machine (IHS-SVM) can obtain higher accuracy of classification results, which proves the practicability of the improved method.
【作者单位】：中北大学机械与动力工程学院;
【基金】：国家自然科学基金资助项目(51175480)
【分类号】：TK428;TP18

【相似文献】