基于遗传算法的模糊神经网络控制器在GTAW中的应用

Study in GTAW of fuzzy neural controller based genetic algorithm

GTAW (钨极气体保护电弧焊 )是一种能够很好控制线能量 ,进行高质量薄板焊接的方法。焊接过程是一个复杂的、多参数耦合的高度的非线性系统 ,在实际焊接过程中难以实现实时、有效的在线控制。模糊控制吸收了人的经验思维的特点 ;神经网络则对信息的处理具有自组织、自学习的特点 ;遗传算法是一种全局优化搜索方法 ,具有简单通用、普遍性强 ,适合并行处理和应用范围广的优点。作者将三者有机地结合起来 ,在模糊神经网络控制器的基础上利用了改进的遗传算法 ,并分析了其网络结构和离线学习的方法 ,协调利用三者的优势设计了一种新型的模糊控制器 ,并使之用于脉冲GTAW仿真中 ,结果证明了该新型模糊神经网络控制器比传统的模糊控制器具有一定的优越性。

GTAW controls linear energy easily and has high quality welding of sheet metal. Welding process is a strong nonlinear system of complexity and multi-parameter. Fuzzy control has characteristic of human experiential thought. Neural network is of self-control and self-study for information. The genetic algorithm is a new method of global optimal searching that has character of simpleness, generality, parallel processing and wide application. This paper designs a new fuzzy controller harmoniously using three methods' advantages. The result of simulation in pulsed GTAW indicates the new fuzzy neural controller is better than traditional fuzzy controller.