【摘要】 阐述为提高屏蔽门故障预测的精度，采用箱型图法对屏蔽门历史故障数据进行异常值处理，并将处理好的故障间隔时间作为LSTM神经网络的输入变量，提出基于遗传算法优化LSTM的屏蔽门故障预测方法。更多还原

【Abstract】 This paper describes the use of box plot method to process outliers in historical fault data of platform screen doors in order to improve the accuracy of fault prediction. The processed fault interval time is used as the input variable of LSTM neural network, and a platform screen door fault prediction method based on genetic algorithm optimization of LSTM is proposed.更多还原