Abstract: To accurately and efficiently diagnose the health state of rolling bearings,a rolling bearing fault diagnosis model based on Improved Ant-lion Optimization (IALO) algorithm was proposed to optimize BP neural network.In IALO algorithm,the mutation operator was used to enhance the diversity of population.The dynamic proportional coefficients and nonlinear dynamic weights were used to balance the wandering weights at different periods in the iterative process,and the possibility of the algorithm falling into local extreme values was reduced.The benchmark function test results showed that IALO algorithm had better optimization performance compared with other algorithms.In addition,to improve the classification performance of BP neural network,the IALO algorithm was used to optimize the weight and threshold of BP neural network,and the rolling bearing fault diagnosis model was built.The experimental results of Paderborn bearing data showed that the BP model optimized by IALO had better fault diagnosis performance.

Key words: bearing fault diagnosis, ant-lion optimization algorithm, dynamic proportionality coefficient, nonlinear dynamic weight, BP neural network

摘要： 为了准确高效地对滚动轴承的健康状态进行诊断,提出一种基于改进蚁狮优化(IALO)算法优化BP神经网络的滚动轴承故障诊断模型。在IALO算法中,采用变异算子,增强了种群的多样性;采用动态比例系数和非线性动态权重,平衡了迭代过程中不同时期游走的权重,降低了算法陷入局部极值的可能性。基准函数测试结果表明,与其他算法相比,IALO算法具有更好的优化性能。另外,为了改善BP神经网络的分类性能,利用IALO算法优化BP神经网络的权值和阈值,构建滚动轴承故障诊断模型。帕德伯恩轴承数据集的实验结果表明,采用IALO算法优化后的BP模型具有较好的故障诊断性能。

关键词: 轴承故障诊断, 蚁狮优化算法, 动态比例系数, 非线性动态权重, BP神经网络