摘要
在电热镦粗过程中,离线、在线两套网络同时运行。离线网络将神经网络与数学模型相结合,用来对样本数据进行训练、测试,存储最优的知识参数,并建立加热电流的高精度预报模型。在线网络是根据传感器实时采集的镦粗缸速度、砧子速度、镦粗压力,按照知识库中的网络知识参数预报出加热电流,较好地实现了在线跟踪控制。
Off-line network and on-line network are adopted in the electric upsetting. Off-line network employed training and testing sample data and Storing optimum knowledge parameters, and then a high precision prediction of the heating current are realized, which combines BP NN with mathematic model. On-line control could predict heating current according to network knowledge parameters in knowledge base, based on real-time acquisition velocities of upsetting cylinder and anvil cylinder, pressure of upsetting cylinder from sensors and realize tracking control current of electric upsetting.
出处
《微计算机信息》
北大核心
2006年第12S期60-62,共3页
Control & Automation
基金
广东省自然科学基金资助项目(990141)
关键词
综合神经网络
预报
离线
在线控制
synthetic neural network,prediction,off-line,on-line control.