Some simplified modeling parameters in the traditional rolling bearing dynamicsmodel led to poor consistency between the simulated and measured vibration signals. In addition, the simulated signal has low accuracy, w...
详细信息
Some simplified modeling parameters in the traditional rolling bearing dynamicsmodel led to poor consistency between the simulated and measured vibration signals. In addition, the simulated signal has low accuracy, which in turn reduces the accuracy of bearing fault diagnosis based on the simulated signal. Thus, to resolve the above-mentioned issues, this study presents a method for optimizing rolling bearing dynamicsmodels based on an improved golden jackal optimization (IGJO) algorithm and sensitive feature fusion. Firstly, this study proposes an IGJO algorithm using a dimension-by-dimension reverse learning strategy and adaptive weights to address the interplay of dimensions in the multidimensional optimization process and the imbalance between the global and local search abilities of the GJO. Secondly, a fusion strategy for bearing faultsensitive features is proposed based on the binary's improved golden jackal optimization (BIGJO) algorithm and principal component analysis (PCA). In this strategy, the Sigmoid function discretization method is used to obtain the B-IGJO algorithm, which is then applied to the measured signal to select the bearing fault sensitivity features. These features are analyzed using PCA to obtain the fused sensitivity feature expression. Finally, the fusion-feature expressions are used to calculate the fusion-sensitive features of the measured and simulated vibration signals, and then the residuals of the two are used as the objective function for modeloptimization. The parameters of the rolling bearing dynamicsmodel are updated using the IGJO. The proposed method is experimentally verified through a single pitting fault dynamicsmodel of the outer ring of the rolling bearing. In conclusion, our results confirm the effectiveness and feasibility of the proposed method.
暂无评论