基于神经网络提高皮革收缩温度测量精度的研究  

Research on Improvement of Leather Shrinkage Temperature Measuring Precision Based on Neural Network

作　　者：宁铎[1] 石秉恒[1] 吕斌[2] 张欢[1] 

机构地区：[1]陕西科技大学电气与信息工程学院,陕西西安710021 [2]陕西科技大学资源与环境学院,陕西西安710021

出　　处：《中国皮革》2013年第21期29-33,共5页China Leather

基　　金：陕西省科学技术发展计划项目:"数字式皮革收缩温度测定仪研制"(2007xcp07)

摘　　要：针对皮革收缩温度测定仪中由于温度传感器自身客观存在的非线性特性而造成的测量误差,利用BP神经网络对其进行了有效补偿。在神经网络算法上结合模拟退火思想加以改进,从而使网络参数调整过程能够跳出局部最小点;将单片机模拟运算过程运用到皮革收缩温度测定系统中。补偿后最大相对误差从0.061下降到了0.010,有效提高了仪器的测量精度。On the basis of the poor measuring accuracy because of the er shrinkage temperature measurement system, a method was proposed to non - linear error of the temperature sensor in the leath- compensate the nonlinearity of temperature sensor with BP neural network. On the algorithm of the neural network , the method combined with the idea of simulated annealing, which can make the BP neural network escape from local minimum points. And then it accomplishes the nonlinear error compensation of tem- perature sensor ADS90 with the MCU to realize it. The greatest error of leather shrinkage temperature measurement system reduces from 0. 061 to 0. 010 after compensation. The accuracy is improved.

关 键 词：温度测量 非线性补偿 BP神经网络 皮革收缩温度 测定仪 

分 类 号：TS57[轻工技术与工程—皮革化学与工程]