论文部分内容阅读
为提高冷连轧机轧制力的预报精度和预报速度,用蚁群算法和神经网络相结合的方法进行轧制力预报模型设计。根据轧制原理建立了BP神经网络冷连轧机轧制力预报模型,以网络权值和阈值为自变量,网络预报误差为目标函数,通过蚁群多代运算,找出预报误差全局最小值,再将相应的权值和阈值输入网络进行训练。应用某厂1 450 mm冷连轧机的实测数据进行离线计算的结果表明,该方法能够防止BP网络陷入局部极小点,且收敛速度快,可作为轧制力预报的新方法在实际应用中加以推广。
In order to improve the prediction accuracy and forecasting speed of rolling force of tandem cold mill, the design of rolling force prediction model is carried out by a combination of ant colony algorithm and neural network. According to the rolling principle, the rolling force prediction model of the tandem cold mill is established. Taking the network weights and thresholds as independent variables and the network forecasting error as the objective function, the global minimum of forecast error is found by multi-generation ant colony algorithm. Then the corresponding weights and thresholds are input into the network for training. The results of offline calculation using the measured data from a 1 450 mm tandem cold mill in a certain plant show that this method can prevent the BP network from falling into a local minimum and has a fast convergence rate and can be used as a new method for forecasting rolling force in practical applications Promotion.