Aiming at problem of inaccurate predicting of traditional fuzzy C-means(FCM) clustering algorithm in coal and gas outburst nonlinear systems high-dimensional multivariate.A new hybrid approach includes an artificial fish swarm algorithm (AFSA) and FCM clustering algorithm based on principal component analysis screening(AFSA-PCA-FCM) is proposed.The approach uses AFSA to optimize initial parameters of FCM,and update the optimal solution of AFSA in bulletin board,so as to determine cluster center,through principal component analysis (PCA) algorithm find a best fit of cluster centers,and through wireless sensor networks (WSNs),real-time collect data of the main factors affecting underground coal and gas,the data are used as sample,and the preprocessed data are analyzed,modeling,and then compare the results with AFSA-BP,FCM and other methods.The evaluation results show that,for the prediction of coal and gas outburst strength,this method has higher accuracy,rapidity,and can stably converge to globally optimal solution.%针对传统模糊C均值(FCM)聚类算法在预测煤与瓦斯突出强度时不精确的问题,提出了一种基于人工鱼群算法(AFSA)和FCM聚类算法相结合的主成分分析(PCA)筛选的混合新方法.使用AFSA优化FCM聚类算法的初始参数,在公告板中更新AFSA的最优解,从而确定当次的聚类中心,通过PCA找到一个最佳聚类中心.通过无线传感网络系统实时采集井下影响煤与瓦斯突出的主要因素数据作为样本,将预处理后的数据进行分析、建模,并与AFSA-BP,FCM等方法进行比较、讨论.结果表明:对于煤与瓦斯突出强度的预测,该方法具有较高的准确性、快速性并能够稳定地收敛于全局最优解.
展开▼