关键词: 瓦斯涌出量, 主成分分析法, 改进的果蝇优化算法, 仿真预测

Abstract: In order to predict the gas emission of working face accurately and quickly, a prediction model for gas emission of working face is put forward based on Principal Component Analysis（PCA）, Modified Fruit Fly Optimization Algorithm（MFOA） and optimized Support Vector Machine（SVM）. PCA is used for dimensionality reduction of original data, eliminating data redundancy, MFOA is used to optimize SVM parameters, avoids the negative impact of the model prediction results affected by selection of SVM parameters. Eventually, coupling prediction model is established based on PCA-MFOA-SVM, and simulation forecast is done as example by taking actual monitoring?data. Results show that the mean absolute error of this model prediction is 0.077 5 m3/t, the mean relative error is 1.323 7%. Comparing with other models this model can realize the dynamic prediction of gas emission in the working face with its higher prediction accuracy and better comprehensive performance.

Key words: gas emission quantity, Principal Component Analysis（PCA）, Modified Fruit flies Optimization Algorithm（MFOA）, simulation and forecast