Application of a Combined Model to Spare Parts Consumption Prediction

Abstract

Spare parts consumption prediction lays a foundation for spare parts support. This paper combines the expo- nential smoothing method with a grey model, and establishes a combined model. The combined model solves the problem of spare parts consumption prediction. The example indicates that the combined model is much more accurate than a sin- gle model. The spare parts consumption is the categories and the amounts of spare parts used to maintain the specific amount of equipments at the specific states under specific times and conditions. Nearly all the segments about spare parts include acquisition, storage, supplying and management have close connections with the analysis of the spare parts consumption information. We can project reasonable spare parts support scheme and enhance the scientific lever of spare parts support work only if we master the law of spare parts con- sumption. Many scholars have made scientific researches on spare parts consumption prediction. Zhao Jianzhong improves on search mode of APSO and weighted method of least squares support vector machine. Then the consumption forecasting model of missile spare parts is established based on RS,EW and WLS-SVM with APSO, and realization proc- ess is analyzed. The example results show that the combi- natorial forecasting model has better forecast precision and important applied value in the course of consumption fore- casting of missile spare parts (1). Ni Xiancun uses the con- cept of the repair degree and improves the proportional haz- ards model based on general renewal process. The parameter value is estimated by analyzing failure data and then the number rotables are calculated based on Monte Carlo simu- lation. An example is given and the results of various main- tenance policies with and without considering covariates are compared and analyzed. Result s show that the model has a larger practical value (2). Li Dawei uses the initial spare parts scheme as prior information and proposes the regulate method of spare parts in incipient operation based on the Bayes method. Finally, the simulation example shows that the method proposed is feasibility. Comparing with classical method, the method proposed can improve the accuracy of spare parts consumption estimation and has good steadiness. So the more rational spare parts scheme can be formed (3). In spare parts consumption predicting practice, there are not enough historical data to be used to forecast by variety of reasons, and the small sample data of spare parts consump- tion are the only available information. The traditional fore- casting methods produce little effect on small sample data. Therefore, a combined model for spare parts consumption prediction based on exponential smoothing method and grey model is developed to avoid the limitations of single predic- tion method, utilize all the information, and improve the pre- diction of spare parts precision consumption. This combined model works well with small sample data. 2. MODELING 2.1. Exponential Smoothing Method The exponential smoothing method has some specific models such as the single exponential smoothing model, the double exponential smoothing model and the cubic exponen- tial smoothing model (4, 5). The single exponential smooth- ing model could be used when the time series are stable, and the trend shows a horizontal direction. The double exponen- tial smoothing model could be used when the time series shows a linear direction. The cubic exponential smoothing model could be used when the time series shows a non-linear direction.