基于人工神经网络的QCr0·5热反挤压力预测模型

Prediction model of QCr0.5 hot invert extrusion force based on artificial neural network

以铬青铜热反挤压过程中凸模锥角、断面缩减率以及温度和挤压力的关系为研究对象,在Matlab语言环境下,以高温反挤压试验数据作为训练和预测样本,用2、3节点的双隐层BP型神经网络对钢材单位挤压力进行了预测。结果表明此方法预测铬青铜反挤压力是有效和可行的。

The paper takes the relationship between punch angle, ratio of section reduction, extrusion temperature and extrusion force as the research object. Taking experimental data of backward extrusion as the samples for training and predicting, it predicts extrusion pressure using double hidden layer BP model with two and three nodes under Matlab language environment, The results indicate that this way is a valid and feasible way for predicting the extrusion load.