Prediction of Qualified Rate of a Military Product Based on Grey Weighted Markov Chain
基于灰色加权马氏链的某军品质量合格率预测

摘要

某军品的一次检验合格率系列同时具有内在规律性和随机波动性，选取灰色加权马氏链模型对序列进行预测。结果表明，组合模型的预测误差仅为1.08%，优于单纯GM（1,1）模型的4.91%，且前者的小误差概率及后验差比值分别为1.00和0.12，均优于后者的0.83和0.50，预测精度显著优于后者，并据此模型对未来6批次的合格率进行了预测分析。

Abstract

Through the statistics of the one-time inspection qualification rate of products produced in 12 consecutive batches of a military parts production line, it is found that the time series has inherent regularity and random volatility, GM(1,1) model and grey weighted Markov model are respectively selected to predict the qualification rate. The results show that the relative error of grey weighted Markov model for prediction is only 1.08%, which is obviously better than 4.91% of GM(1,1) model. The small error probability and posterior error ratio of the former are 1.00 and 0.12 respectively, which are better than 0.83 and 0.50 of the latter. The prediction accuracy of the former is significantly better than the latter. Finally, the excellent grey weighted Markov model is used to predict the qualification rate of the next 6 batches.