Research on the Prediction of Maintenance Spare Parts Consumption based on Extreme Learning Machine

Aiming at the problems of slow response to fault disposal of operational equipment, long ordering cycle of spare parts, and large shortfall of spare parts prepositioning in special vehicle companies, a method of predicting the consumption of maintenance spare parts based on empirical modal decomposition to extract the fault occurrence factor and based on the Extreme Learning Machine (ELM) is proposed. The method firstly extracts the failure occurrence characteristics through empirical modal decomposition of a certain type of faulty maintenance data in the past 2 years, and then uses the limit learning machine to predict the occurrence of equipment failures in the following year, obtains the number of maintenance spare parts pre-positioned in each month, and pre-positioned the spare parts in place in advance, and finally, verifies it by using the maintenance data of a certain type of equipment in the past 2 years, and the experimental results show that: compared with the traditional practice of repairing spare parts ordering, the equipment repairing spare parts consumption prediction method using the limit learning machine is more effective than that using traditional practice. The experimental results show that, compared with the traditional maintenance spare parts ordering practice, the equipment maintenance spare parts prediction using the limit learning machine is more accurate, and the guarantee efficiency is higher.