基于PSO-LSSVM的水分仪称重传感器非线性补偿研究

发布时间：2018-05-21 03:10

本文选题：应片式传感器 + 非线性补偿　；参考：《仪器仪表学报》2017年03期

【摘要】：水分仪称重传感器的输出与待烘干物重量存在一定的非线性关系,且烘干失重法对传感器工作环境温度影响也会造成称重传感器输出的非线性变化。在分析传统应片式传感器非线性输出的产生机理基础上,提出一种基于变异粒子群-最小二乘支持向量机(PSO-LSSVM)的新型称重传感器非线性补偿方案:首先采用噪声协方差可变的卡尔曼滤波算法对称重传感器输出数据进行滤波,减小噪声的影响;再将滤波后的数值与环境温度值作为回归参量,建立基于LSSVM算法的水分仪称重模型;利用变异PSO算法对模型进行参数寻优。实验表明,在220 g/0.001 g水分仪中采用本方法对传感器输出进行非线性补偿后,称量精度明显优于国家检定规程标准;此外,该方法满足在小训练样本条件下,具有很好的泛化性。
[Abstract]:There is a certain nonlinear relationship between the output of the moisture sensor and the weight of the object to be dried, and the influence of the drying weight loss method on the temperature of the sensor working environment will also cause the nonlinear change of the output of the weight sensor. On the basis of analyzing the mechanism of nonlinear output of traditional chip sensor, A novel nonlinear compensation scheme based on PSO-LSSVM-based variable particle swarm optimization (VPS-LSSVM) for weighing sensors is proposed. Firstly, the symmetric weight sensor output data are filtered by Kalman filter algorithm with variable noise covariance. The influence of noise is reduced, then the filtered value and ambient temperature are taken as regression parameters, and the weighing model of moisture meter based on LSSVM algorithm is established. The parameters of the model are optimized by mutation PSO algorithm. The experimental results show that the weighing accuracy of this method is obviously better than that of the national verification regulation standard after the nonlinear compensation of the sensor output in 220 g / 0.001 g moisture meter. In addition, the method can meet the requirements of small training samples and has a good generalization.
【作者单位】：湖南大学电气与信息工程学院;湖南师范大学工学院;
【基金】：国家自然科学基金(51205127) 湖南省科技计划(2015WK3002)项目资助
【分类号】：TP212

【相似文献】