【摘要】 以一台单级GM脉管制冷机为对象,基于前馈神经网络对阀门开度与无负荷制冷温度的关系展开探究。利用实验数据集训练神经网络模型,预测不同阀门开度下的无负荷制冷温度。结果表明,模型预测值与实际测量值平均相对误差为5.1%,证明了模型的有效性和可靠性。优化后的制冷机无负荷制冷温度15.2 K,性能为40.6 W@30.0 K,相对卡诺效率为4.3%。更多还原

【Abstract】 The relationship between valve opening and no-load refrigeration temperature in a GM single-stage pulse tube cryocooler was studied using a feedforward neural network approach.An experimental dataset was employed to train the neural network model,enabling the prediction of the no-load temperature achievable under different valve opening configurations.The results reveal an average relative error of 5.1 % between the model’ s predicted values and the experimentally measured values,thereby validating the model s effectiveness and reliability.Following optimization,the cryocooler attained a no-load temperature of 15.2K and a cooling capacity of 40.6 W at 30K.The relative carnot efficiency is 4.3%.更多还原