电阻和电容信息融合的籽棉回潮率检测

Detection of Moisture Regain of Seed Cotton Based on Resistance and Capacitance Information Fusion

针对现有收购环节籽棉回潮率检测方法中单一电阻法检测精度差和单一电容法检测不稳定的问题,提出了一种电阻和电容信息融合的籽棉回潮率检测方法。首先,使用标准电阻和标准电容对所搭建试验平台的电阻检测板和电容检测板进行了可靠性验证;然后,使用试验平台以及八篮烘箱对待测棉样进行了检测并获得试验数据,利用该数据构建了基于电阻和电容信息融合的籽棉回潮率检测模型,并分别使用多元线性回归(MLR)、偏最小二乘回归(PLSR)、支持向量回归(SVR)和反向传播神经网络(BPNN)4种方法进行了模型构建及比较验证。结果显示:BPNN的评价指标优于其他3种模型,决定系数R^(2)为0.8966,均方根误差RMSE为0.0504%,程序运行时间t为0.6690s。最后,将信息融合算法与单一电阻法和单一电容法进行了对比分析,结果表明:该方法优于单一电阻法和单一电容法。因此,提出的电阻和电容信息融合的籽棉回潮率检测方法切实可行,能够为收购环节籽棉回潮率检测提供参考。

With regarding to the status quo ofpoor accuracy of single resistanceand instability of single capacitance detection on moisture regain of seed cotton in trade,this study proposed a detection method of moisture regain of seed cotton based on the resistance and capacitance information fusion.Firstly,the reliability of the resistance and capacitance detection board of the test platform was verified with using standard resistance and capacitance.Then,the detection platform and Eight Basket oven were used to test the cotton samples and acquire thedata.Based on the data,a testing model of moisture regain of seed cotton was built according to the information fusion of resistance and capacitance.In this process,multiple linear regression,partial least squares regression,support vector regression and back propagation neural network were used to construct and verify the models.The BPNNmodel achieved the best performanceamong the other three models,where the determination coefficient R^(2) was 0.8966,the RMSE was 0.0504%,and the timecostwas 0.6690s.Finally,the information fusion algorithm was compared with single resistance and single capacitance method,respectively,and the results showed that the method of information fusion of resistance and capacitancewas superior to the single resistance and single capacitance method.Therefore,this method was feasible and could provide insightful reference for detecting moisture regain of seed cotton in the purchase process.