CAJ | 학술논문

针对瓦斯涌出量受多因素影响，传统的预测方法难以建立准确的数学模型，导致预测精度低这一问题。提出一种经改进的粒子群算法( MPSO)优化的基于柯西分布加权的最小二乘支持向量机( CWLS-SVM)算法来预测非线性动态瓦斯涌出量。柯西分布加权的最小二乘支持向量机根据预测误差的统计特性，确定加权规则参数，以达到赋予训练样本不同权值的目的。并用MPSO算法对CWLS-SVM模型的正则化参数λ和高斯核参数σ寻优。利用无线传感器网络采集到的各项历史数据进行实例分析。结果表明，该算法有效的提高了瓦斯涌出量的预测精度，降低了预测误差，为煤矿瓦斯防治提供理论支持。
침대와사용출량수다인소영향，전통적예측방법난이건립준학적수학모형，도치예측정도저저일문제。제출일충경개진적입자군산법( MPSO)우화적기우가서분포가권적최소이승지지향량궤( CWLS-SVM)산법래예측비선성동태와사용출량。가서분포가권적최소이승지지향량궤근거예측오차적통계특성，학정가권규칙삼수，이체도부여훈련양본불동권치적목적。병용MPSO산법대CWLS-SVM모형적정칙화삼수λ화고사핵삼수σ심우。이용무선전감기망락채집도적각항역사수거진행실례분석。결과표명，해산법유효적제고료와사용출량적예측정도，강저료예측오차，위매광와사방치제공이론지지。
For gas emission influenced by multiple factors, the traditional forecasting methods are difficult to establish accurate mathematical model, leading to the problem of low prediction accuracy. An modified particle swarm optimization ( MPSO ) algorithm based on cauchy distribution by weighted least squares support vector machine( CWLS-SVM) algorithm to predict the nonlinear dynamic of gas emission. Cauchy distribution weighted least squares support vector machine based on the statistical properties of the prediction error to determine the weighting rule parameters in order to achieve the purpose of giving different weights training samples. MPSO algo-rithm with regularization parameter CWLS-SVM model and the gaussian kernel parameter optimization. Using wire-less sensor networks to collect the historical data to analyze an example. The results show that the algorithm effec-tively improved the prediction accuracy of gas emission,reducing forecast error,providing theoretical support for coal mine gas prevention and control.