基于生态遗传算法的煤矸石自适应模糊模式识别

给出煤矸石组分模式识别的模糊神经网络模型，提出一种实用生态算子，同时将此基础上构建的生态遗传算法用于模糊神经网络的离线学习，能有效避免传统BP算法学习速度慢、易陷入局部极小的缺陷和基本遗传算法的遗传滑脱现象．仿真和实验结果显示新算法使离线训练的网络具有良好的收敛性能，而且从训练好的定量网络中提取模糊规则用于原煤的在线自动分选，不仅能提高煤中矸石的识别率，而且有效解决了系统识别精度与实时分选之间的矛盾．

Adaptive fuzzy pattern recognition of coal and gangues based on niche genetic algorithm

A fuzzy neural network (FNN) used for the identification of coal and gangues was described. The niche genetic algorithm (NGA) based on a proposed practical niche operator was responsible for the off-line training of FNN. It was better than the most common BP algorithm with low learning speed and the general local minimum solutions, and it was predominant over the simple genetic algorithm with genetic drift. It was shown that the off-line trained FNN by NGA had a global convergence, and the fuzzy rules from the resultant quantitative FNN were used for the on-line separation of coal and gangues. The resulting separator was provided with higher recognition accuracy and excellent real-time performance.