基于烟花算法与长短时记忆网络的模温预测方法研究

针对传统神经网络模型预测模具温度准确度低,网络超参数选取困难等问题,提出一种基于烟花算法优化长短时记忆网络的模温预测模型,为铸造成型模温自动控制提供基础。首先根据铸造过程生产工艺选取影响铸造系统的主要变量,利用灰关联分析得出各变量灰色关联度并去除关联度小的变量,建立模具温度影响因子变量的数据集;其次采用烟花算法对长短时记忆网络进行优化,建立模具温度预测模型;最后与BP神经网络和长短时记忆网络预测效果进行对比。实验结果表明基于烟花算法优化的长短时记忆网络的模温预测方法绝对误差小于2.4℃,平均绝对百分比误差小于0.12。

Research on Model Temperature Prediction Method Based on Fireworks Algorithm and Long and Short Time Memory Network

Aiming at the problems of low precision of mold temperature and difficulty in selecting network parameters, based on fireworks algorithm to optimize the long and short time memory network, a mold temperature prediction model is proposed, which provides a basis for automatic control of casting mold temperature. Firstly, according to the production casting process, the main variables affecting the casting system are selected. The grey relational analysis is used to obtain the grey correlation degree of each variable and remove the small correlation degree, and the data set of the mold temperature influence factor variable is established. Secondly, the fireworks algorithm is used to optimize the long and short time memory network to establish a mold temperature prediction model. Finally, the prediction results are compared with BP neural network and long-term short-term networks. Experiments show that the absolute error of the model temperature prediction method based on the long-term and short-term memory network optimized by the firework algorithm is less than 2.4℃ and the mean absolute percentage error is less than 0.12.