A New Method for Rolling Bearing Condition Evaluation Method Under Varying Speeds

Current rolling bearing condition monitoring methods based on health indicator (HI) under constant speed have achieved promising results. However, the development of condition monitoring methods under variational speeds remains slow due to two main drawbacks: 1) varying speeds bring too much redundant information with excessive sampling frequency and 2) under the influence of varying speeds, current methods are insufficient in properly reflecting the correlation with damage degree. To address these problems, a genetic algorithm-based feature-fusion health indicator (GA-FF-HI) is proposed to evaluate the condition for bearings under variable speeds. In the proposed method, two core issues are considered. The first one is to extract a frequency band, which can effectively reflect the shock in vibration signals by an optimal bandpass filter. Another issue is the searching of the compatible weights of features for HI construction. First, a GA is applied to ascertain the central frequency and bandwidth for the filter where the amplitude on fault orders in order spectrum is considered as the optimization objective. The features from the filtered signal are combined with corresponding weights obtained by the convex optimization to form the final HI. Two cases are implemented to validate the effectiveness of the GA-FF-HI in condition monitoring. The experimental results prove that the GA-FF-HI can reveal the damage degree of bearing signal under varying speeds.