【摘要】 以光电效应为基础的一维位置敏感位移传感器（position sensitive detector, PSD）受温度影响很大，导致测量误差大。本文针这一问题，提出免疫算法与最小二乘支持向量机（IA-LSSVM）相结合，通过免疫算法迭代优化最小二乘支持向量机中的惩罚因子以及核函数参数来构建补偿模型。由二维标定实验得到在不同温度下传感器的输入输出数据，将其分为测试集与训练集，以测试集的最小二乘均方差为目标函数，将其转化为凸优化问题来达到温度补偿的效果。试验结果表明，传感器零位温度系数由补偿前的3.0×10^-4/℃提升到1.6×10^-4/℃,提高了约两倍；灵敏度温度系数由之前的2.8×10^-3提升到2.8×10^-4/℃,提高了十倍；温度附加误差由补偿前的12.5%提升至1.27%,提高了近十倍，从而有效削弱了温度对测量结果的影响。更多还原

【Abstract】 A one-dimensional Position Sensitive Detector based on photoelectric effect is greatly affected by temperature, resulting in large measurement error. In this paper, a combination of immune algorithm and least squares support vector machine（IA-LSSVM）is proposed to build a compensation model by iteratively optimizing the penalty factor and kernel function parameters of LSSVM. The input and output data of the sensor at different temperatures were obtained by two-dimensional calibration experiments, which were divided into test set and training set. The variance of test set was taken as the objective function, and it was transformed into a convex optimization problem. The test results show that the zero temperature coefficient increases from 3.0×10^-4/℃ to 1.6×10^-4/℃ after compensation, about two times lower; the sensitivity temperature coefficient drops to 2.8×10^-4/℃,about ten times lower than the 2.8×10^-3/℃;the temperature additional error increases from 12.5% to 1.27%,nearly ten times higher. The effect of temperature is almost effectively eliminated.更多还原