故障特征组合选择方法

特征选择方法主要包括过滤方法和绕封方法。为了利用过滤方法计算简单和绕封方法精度高的优点，提出一种组合过滤和绕封方法的特征选择新方法。该方法首先利用基于互信息准则的过滤方法得到满足一定精度要求的子集后，再采用绕封方法找到最后的优化特征子集。由于遗传算法在组合优化问题上的成功应用，对特征子集寻优采用了遗传算法。在数值仿真和轴承故障特征选择中，采用新方法在保证诊断精度的同时，可以节省大量选择时间。组合特征选择方法有较好的寻优特征子集的能力，能够节省选择时间，具有高效、高精度的双重优点。

Composite Fault Feature Selection Method

The feature selection can remove redundant features existed in the fault feature set to enhance diagnosis precision. Generally there are two different methods for the feature selection, i.e. filter method and wrapper method. A composite feature selection method for reaping the benefits of precision from the wrapper method and keeping the computational expense down from the filter method is proposed. Firstly the method filters original features to form a feature subset which can meet classification correctness rate, then the wrapper method is used to select an optimal feature subset. A technique for solving optimization problems is given by genetic algorithm (GA). GA is applied to the problem of the optimal feature selection. The composite method saves more computing time than the wrapper method with holding the classification accuracy in data simulation and experiment on bearing fault feature selection. So the method possesses excellent optimization property and saves more selection time, thus having the characteristics of high correctness rate and high efficiency.