Rolling Bearing Fault Trend Prediction Based on Composite Weighted KELM

Abstract

The abilities of different degradation feature types to characterize rolling bearing fault trend are distinctive. And even the characteristic ability of the same degradation feature can change at various times. Thus these feature samples possess heteroscedasticity. However, traditional kernel extreme learning machine (KELM) model assumes that different input samples' effects on the predicted value are equal, which results in low prediction precision and low computing efficiency. To solve this problem, a novel composite weighted KELM (CWKELM) prediction model, which is fused with explicit weighting and implicit weighting, is proposed. In both feature type scale and sample time scale, the feature samples and the prediction model are weighted according to the prediction error. An adaptive mutation particle swarm optimization (AMPSO) algorithm is applied in optimizing the penalty factor and the kernel parameter in the model. Taking various entropy features as the input samples, the proposed model is adopted to conduct one-step and multi-step prediction for rolling bearing fault trend. Experimental results show that this prediction model has higher prediction accuracy and computing efficiency compared with the traditional KELM model.

Misalignment Fault Prediction of Wind Turbines Based on Combined Forecasting Model

“…After the fault signals are obtained, a reasonable and effective prediction method is necessary for accurately predicting the future operational status of the equipment faults. At present, many scholars have studied the prediction techniques for different types of faults [15][16][17]. In order to determine a suitable forecasting model of misalignment fault in wind turbines, the commonly used prediction methods are summarized in Table 1.…”