基于遗传神经网络的真空绝热板导热系数测量方法研究

Abstract

传统采用大平板热保护法测量真空绝热板(VIP)导热系数,该方法测量精度高,但检测速度慢。国外HC-120 VIP快速检测仪大大提高了检测速度,但价格昂贵。针对以上问题,本文提出了基于埋入式热流计法快速高精度测量VIP导热系数的方法。通过理论分析和仿真分析,证明了测量原理的可行性。在测量模型的标定中,针对VIP导热系数与频率变化特征值之间的非线性问题,巧妙结合遗传算法和bP神经网络,从而进行非线性补偿,使测量精度优于1%。与传统基于最小二乘法进行非线性补偿的方法相比,遗传神经网络具有误差小、精度高和全局寻优能力等优点。实际应用表明,本文提出的测量方法满足了高精度、低成本的测量要求,具有广泛推广的应用价值。

A novel technique was developed to rapidly and precisely measure the thermal conductivity of largearea,vacuum insulated panel( VIP).In the newly-developed technique,the heat-flow sensors were embedded inside the VIP.The theoretically calculated,numerically simulated and experimentally evaluated results strongly support the new method.In measurement calibration,genetic algorithm and BP neural network were used to solve the nonlinear problems involved in dependence of VIP's thermal conductivity on frequency characteristics; and the non-linear compensation results in a precision better than 1%.When it comes to nonlinear compensation,genetic neural network outperforms the conventional least-square method in many ways,including smaller error,higher precision and better global optimization ability.The test results show that the novel technique is capable of measuring the VIP's thermal conductivity rapidly,precisely,easily and at a low coat.