压电陶瓷传感器的灵敏度温漂误差补偿研究

针对微测系统中压电陶瓷传感器的灵敏度温漂会使其在变化的温度环境中工作时性能不稳定,进而影响检测精度问题,提出了一种基于改进Elman神经网络的压电陶瓷传感器灵敏度温漂误差补偿控制方法。分析了压电陶瓷传感器产生灵敏度温漂现象的原因。以压电陶瓷切削力测量传感器为对象,在不同温度下对传感器的灵敏度进行了标定试验研究。研究结果表明,压电陶瓷传感器在同一温度下工作时具有良好的线性度,在温度变化的环境中工作会伴有灵敏度温漂现象。为了有效补偿灵敏度温漂附加误差,提高检测精度,建立了基于改进Elman神经网络的灵敏度温漂补偿模型,并对模型涉及的学习算法、激励函数、输入输出层节点以及承接层和隐含层节点数等相关内容进行了研究。对比试验验证结果表明,所建立的灵敏度温漂补偿模型对压电陶瓷传感器的灵敏度温漂误差补偿控制效果明显,未经灵敏度温漂补偿,直接按照常温下灵敏度标定结果预测的压电陶瓷传感器加载力和实际加载力之间误差较大,最大误差达到29.16 N,利用本文建立的基于改进Elman神经网路灵敏度温漂补偿模型补偿后,补偿模型的预测力和压电陶瓷传感器的实际加载力最大误差仅0.72 N,有效保证了检测精度。

Research of Sensitivity Temperature Drift Error Compensation on Piezoelectric Ceramic Sensors

In order to solve the problem that the sensitivity temperature drift of piezoelectric ceramic sensor will make its performance unstable in the changing temperature environment,and then affect the detection accuracy,a sensitivity temperature drift compensation control method of piezoelectric ceramic sensor based on improved Elman neural network was proposed.The reason of sensitivity temperature drift of piezoelectric ceramic sensor was analyzed.The sensitivity of piezoelectric ceramic cutting force sensor was calibrated at different temperatures.The results show that the piezoelectric ceramic sensor has good linearity when it works at the same temperature,and it will be accompanied with sensitivity temperature drift when it works in the environment of temperature change.In order to effectively compensate the additional error of sensitivity temperature drift and improve the detection accuracy,a sensitivity temperature drift compensation model based on improved Elman neural network was established,and the learning algorithm,the excitation function,the input and output layer nodes,the receiver layer and the hidden layer nodes were studied.The results of contrast test show that the sensitivity temperature drift compensation model has obvious control effect on the sensitivity temperature drift error compensation of piezoelectric ceramic sensor.Without the compensation of sensitivity temperature drift,the error between the predicted load force and the actual load force of piezoelectric ceramic sensor is large,the maximum error is 29.16 N.After the compensation based on the improved Elman neural network sensitivity temperature drift compensation model established in this paper,the maximum error between the predicted force of compensation model and the actual load force of piezoelectric ceramic sensor is only 0.72 N,which effectively guarantees the detection accuracy.