由于支持向量机的参数优化算法存在控制参数多、易陷入局部最优解的问题，并为了实现滚动轴承故障信号的诊断，提出一种基于单形进化算法的支持向量机滚动轴承故障诊断方法。首先，单形进化算法利用全随机的方式建立粒子的单形邻域搜索算子以减少算法控制参数，建立粒子多角色态搜索策略以避免算法陷入局部最优解；然后，将单形进化算法应用到支持向量机的参数寻优，并用滚动轴承信号完成故障诊断；试验中，采用滚动轴承信号的集总经验模态分解的能量特征作为输入，进行算法的性能分析与测试。结果表明该算法可以有效地缓解粒子陷入局部最优解且减少了控制参数，并能完成滚动轴承故障信号的诊断与识别。

Because the parameter optimization algorithm of support vector machine (SVM) has many control parameters and fall into the local optimum easily, and in order to realize diagnosis, rolling bearings fault diagnosis of SVM optimized by surface-simplex swarm evolution algorithm is proposed. First, surface-simplex swarm evolution (SSSE) establishes the particle's simple neighborhood search operator in a random way to reduce control parameters and the multi-role state search strategy to avoid falling into the local optimum. Then SSSE is applied to parameter optimization of SVM to realize fault diagnosis. In the test，the energy matrix of ensemble empirical mode decomposition(EEMD) of rolling bearing signal is used as the feature input to perform performance analysis and testing of the algorithm. The results show that the method not only effectively avoid the particles falling into the local optimal solution，but also reduce the control parameters, and can complete the signal diagnosis.

TP183

国家自然科学基金项目:提取重力固体潮信号中地球物理信息和地震前兆信息的关键信号处理算法研究