基于人工神经网络的P20钢热处理工艺

Heat treatment process of P20 steel using artificial neural network

用BP人工神经网络及材料微观分析方法研究了热处理工艺对P20钢硬度的影响。结果表明,BP网络能根据淬火及回火温度精确预测P20钢热处理后的硬度;BP网络预测结果表明,P20钢经800～920℃淬火及530～650℃回火,在给定的淬火温度下,随回火温度的增加硬度急剧降低;在给定的回火温度下,随淬火温度的增加硬度略有增加。材料微观分析表明:这主要归因于回火温度升高造成的碳化物长大和α相的回复程度的加剧及淬火温度升高造成的碳及合金元素固溶量的增加。

The effects of heat treatment process on hardness of P20 steel were studied using an artificial neural network model with back propagation algorithm（BP network） and microstructure analysis method.The results show that BP network can be used to predict the hardness of P20 steel accurately according to quenching and tempering temperature.Results of the prediction indicate that hardness of the steel decreases with increasing tempering temperature at a given quenching temperature,and increases slightly with increasing quenching temperature at a given tempering temperature,when the steel is quenched at 800～920℃ and tempered at 530～650℃.It is confirmed that the change of hardness is attributed to carbide coarsening and α phase recovering with increase of tempering temperature and higher content of carbon and alloy elements in austenite with increasing quenching temperature based on microstructure observation.