基于卷积稀疏组合算法的轴承性能衰减评估

为有效监控与评估轴承工作状态,提出一种基于卷积稀疏组合算法评估方案。基于卷积神经网络框架建立轴承性能稀疏表示判别准则,并预测轴承的性能衰减程度；利用轴承衰减自相关函数,预判与轴承谱相关的密度条件,并在分析其他模型数值参量的基础上,验证评估方法的应用平稳性。选取退化指标作为实验对象,并通过分析相关的指标参数值可知,提出算法的评估结果可解释性强,能够较好维护轴承性能的衰减机制,影响系数值被控制在[-1,1]之间；在与传统算法的预测性能对比中,提出算法在两种状态下的偏差值分别为0.02和0.01,优于传统的轴承性能评估算法,同时在评估预测效率方面也具有一定优势。

Evaluation of Bearing Performance Attenuation Based on Convolutional Sparse Combination Algorithm

In order to effectively monitor and evaluate the working state of bearings, an evaluation scheme based on convolutional sparse combination algorithm was proposed. Based on the convolutional neural network framework, the sparse representation criterion of bearing performance was established and the attenuation degree of bearing performance was predicted. The bearing attenuation autocorrelation function is used to predict the density conditions related to the bearing spectrum, and on the basis of analyzing the numerical parameters of other models, the application stability of the evaluation method is verified. Degradation index is selected as the experimental object, and through the analysis of related index parameter values, it can be seen that the evaluation results of the proposed algorithm have strong interpretability, which can better maintain the attenuation mechanism of bearing performance, and the influence coefficient value is controlled between [-1,1]. In comparison with the prediction performance of the traditional algorithm, the deviation values of the algorithm in two states are 0.02 and 0.01 respectively, which is superior to the traditional bearing performance evaluation algorithm, and also has certain advantages in the evaluation and prediction efficiency.