Q学习对制糖结晶遗传神经网络收敛性的改进(英文)

Improving the Convergence of the Genetic Neural Network in the Crystallizing of Sugar Using Q-Learning

采用多层前馈遗传神经网络模型对甘蔗制糖结晶速度进行学习和预测 ,并针对该模型存在的计算量大 ,收敛慢的问题 ,采用具有强化作用的Q学习确定遗传算法的变异概率 ,以提高学习的收敛速度 。

The crystallizing speed of cane sugar is learned and predicted by the model of feedforward neural network using genetic algorithms. To counter the problem in the model which needs a lot of calculations but has slow speed of convergence, we use Q learning with reinforcement to decide on the variation probability of genetic algorithms and to increase the convergence speed of learning. The results of the simulation show the effectiveness of the method.