论文部分内容阅读
建立了控制ＣＯ２焊飞溅的神经网络控制系统并对该系统进行了仿真．构成仿真系统的逆模型、正模型和差值模型均采用了改进的ＢＰ算法，用典型的样本集对这三个模型进行了训练，分析了影响仿真效果和响应品质指标的因素，初步确定了一种最佳的控制方式．
A neural network control system to control CO2 welding spatter was established and the system was simulated. The inverse model, the positive model and the difference model which constitute the simulation system all adopt the improved BP algorithm. The three models are trained with the typical sample set. The factors affecting the simulation effect and the response quality index are analyzed. One of the best ways to control.