SSA优化深度双向门控循环单元网络的轴承性能退化趋势预测

Bearing performance degradation trend prediction sparrow search algorithm optimization bidirectional gating cycle unit

为在非经验指导下获取双向门控循环单元网络中最优隐藏层单元数,实现滚动轴承性能退化趋势预测,提出基于麻雀搜索算法优化深度双向门控循环单元的轴承性能退化趋势预测方法。首先,在正向门控循环单元网络基础上,增加反向门控循环单元网络,以构建深度双向门控循环单元预测网络;然后,将预测值与真实值的均方误差作为适应度值,根据麻雀发现者和捕食者进行参数更新,经优化后获得最优隐藏层单元参数下的深度双向门控循环单元网络预测模型;最后,通过全连接层实现性能退化趋势预测。在公共数据集与实测数据集上进行试验验证,验证了所提方法的有效性与可行性。

In order to obtain the optimal number of hidden layer elements in bidirectional gated cyclic element network under non-empirical guidance and realize the performance degradation trend prediction of rolling bearings,a bearing performance degradation trend prediction method was proposed based on sparrow search algorithm to optimize the depth bidirectional gated cyclic element.Firstly,the forward gating loop unit network is added to the reverse gating loop unit network to construct the deep bi-directional gating loop unit prediction network.Then,the mean square error between the predicted value and the true value was used as the fitness value,and the parameters were updated according to the sparrows finder and predator.After training,the depth bidirectional gated cyclic element network with the optimal hidden layer element parameters was obtained.Finally,the performance degradation trend is predicted by the full connection layer.The validity and feasibility of the proposed method are verified by experiments on public and measured data sets.