基于时序数据的电梯数据清洗方法研究

Research on Data Cleansing Method of Elevator Based on Time Series Data

为了对非周期性、非高斯性及间歇性的曳引式电梯数据进行数据清洗,对电梯运行过程中的异常数据进行排查,提出一种改进后的长短期记忆网络的数据清洗模式。在对基于物联网技术使用数据库存储的时序数据进行异常数据的清洗时,提取不等长的时间序列数据进行划分与填充,利用长短时间神经网络对其进行建模,进行初期的异常数据检测清洗。在系统中实现电梯故障系统的故障预测、寿命分析、可视化前的数据清洗工作,完成数据优化。

In order to perform the data cleaning of aperiodic,non-Gaussian and intermittent traction type elevator and troubleshoot the abnormal data in the process of elevator operation,this paper proposes a data cleaning mode of modified long short-term memory network.Based on the IoT technology using a database to store the time-series data of abnormal data cleaning,the time series data of different lengths are extracted for division and filling,the long and short time neural network is used for modeling,and the initial abnormal data detection and cleaning are carried out.Data cleaning and data optimization are completed before the realization of fault prediction,life analysis and visualization of the elevator fault system.