论文部分内容阅读
铝液中夹杂物的含量与大小会直接影响铝液浇注过程及成品质量。用电阻法在线检测电气回路的电流,该电流的变化与铝液的夹杂物密切相关,同时考虑在电压、气压、小孔直径、铝液温度等条件不同的情况下,夹杂物直径大小会直接影响到电流的幅值和宽度。利用BP神经网络方法建立铝液夹杂物直径预测模型,结合在线测量的实验样本数据,对预测模型进行训练和测试,并且通过训练效果确定隐含层的神经元个数,从而确定最好的BP神经网络的夹杂物预测模型。实验和仿真结果表明,基于BP神经网络的铝液夹杂物预测模型的预测数据符合实际情况。
Aluminum liquid inclusions in the content and size will directly affect the aluminum casting process and product quality. The current of the electric circuit is detected online by the resistance method. The change of the current is closely related to the inclusions in the aluminum liquid. In consideration of the different conditions of the voltage, the air pressure, the hole diameter and the temperature of the molten aluminum, the diameter of the inclusions will be directly Affects the current amplitude and width. The prediction model of aluminum liquid inclusions diameter was established by BP neural network method. The prediction model was trained and tested by combining with the experimental sample data of on-line measurement, and the number of neurons in hidden layer was determined by training effect to determine the best BP Inclusion prediction model based on neural network. The experimental and simulation results show that the prediction data of aluminum liquid inclusion prediction model based on BP neural network is in accordance with the actual situation.