摘要
影响矿井瓦斯浓度的地质因素间存在着模糊性,为了解决带有模糊信息的动态瓦斯浓度预测问题,建立了一种模糊自回归(Fuzzy-AR(P))时间序列预测模型。采用AIC,BIC和FPE准则来确定模型阶数(确定为22阶),将计算模型系数中心值及模糊幅度值的问题转化成约束优化问题,并利用MATLAB优化工具箱求解。利用所建模型对6个测试样本进行预测分析,平均模糊隶属度为0.85,平均绝对误差为0.040 3,预测效果明显。与其他预测模型相比,Fuzzy-AR(P)模型的预测结果是一个区间,扩大了相关量的适用范围,使预测结果更合理、更科学。
The geologic factors affecting mine gas concentration exists fuzziness, established a fuzzy regression （Fuzzy - AR （P）） time series prediction model in order to solve the problem of dynamic gas concentration prediction with fuzzy information. Adopting AIC, BIC and FPE criterion to determine the model order number （defined as 22 order） , at the center of the computing model coefficient and fuzzy amplitude value problem into a constrained optimization problem, and use the MATLAB optimization toolbox. Using the model to forecast the six test sample analysis, the fuzzy membership degree is 0.85 on average, the average absolute error is 0.040 3, prediction effect is obvious. Compared with other forecasting model, Fuzzy - AR （P） model to predict the result is a range, expand the scope of the amount of the relevant, the forecasting results more reasonable and more scientific.
出处
《煤》
2015年第6期4-6,25,共4页
Coal
基金
甘肃省科技厅项目资助(1204GKCA004)
甘肃省财政厅专项资金立项资助(甘财教[2013]116号)
