基于泛平均运算的矿井瓦斯浓度组合预测模型

Coal Mine Gas Concentration Combination Prediction Model Based on Universal Average Operation

为有效分析煤矿瓦斯监测数据以实现准确、可靠的瓦斯浓度预测,基于不等权泛平均运算模型,研究瓦斯浓度时间序列组合预测的方法,提出一种新的矿井瓦斯浓度组合预测模型,并证明最优组合预测模型是其特例。采用自回归(AR)模型和径向基函数(RBF)神经网络预测模型作为组合预测模型的单项预测模型;以遗传算法和最小二乘法确定新组合预测模型的参数,实现瓦斯浓度预测单项模型的最优组合。试验分析表明:新模型在平方和误差、平均绝对误差、均方误差、平均绝对百分比误差、均方百分比误差等评价指标上,均取得比自回归模型、径向基函数神经网络模型和最优组合预测模型更低的误差。

For the purpose of achieving accurate and reliable gas concentration prediction through effec- tive analysis of gas measuring data in mines, based on unequal weighted universal average operation model, a method for gas concentration time series combination prediction was studied. A new coal mine gas con- centration combination prediction model was proposed, and it was proved that the optimal combination forecasting model is a special case. Autoregressive model and radial basis function neural network predic- tion model were thought as single predictive models for combination prediction model, parameters of this model were determined by combining genetic algorithms with least squares method to achieve the optimal combination of individual models to predict gas concentration. The results show that in terms of sum of square error, mean absolute error, mean square error, mean absolute percentage error, mean square per- centage error evaluation, this model achieved less error than autoregressive model, radial basis function neural network model and optimal combination prediction model.