压电作动器的高斯过程回归建模

Gaussian Process Regression Modeling of Piezoelectric Actuators

针对压电作动器所存在的迟滞非线性问题,提出了一种基于高斯过程回归的压电作动器建模方法,用高斯过程来表征压电作动器的电压与位移之间的非线性关系,采用粒子群优化算法对高斯过程中协方差函数所包含的三个超参数进行辨识,并与传统的Bouc-Wen模型下的压电作动器的迟滞性结果进行对比。实验仿真结果表明,在信号频率(1~100)Hz范围内,高斯过程回归模型所建立的压电迟滞模型最大均方根误差为0.5329μm,最大相对误差为3.23%,相比于传统的Bouc-Wen模型,高斯过程回归模型模型在10 Hz、50 Hz、80 Hz、100 Hz处的建模误差分别降低了约15%、29%、67%、72%.

For piezoelectric actuator hysteresis nonlinearity existing problem,a kind of piezoelectric actuator based on gaussian process regression modeling method is put forward,the gaussian process is used to characterize the nonlinear relationship between voltage of piezoelectric actuators and displacement,the three super parameters which are contained in the process of Gaussian covariance function are identified using the particle swarm optimization algorithm.Comparing with the traditional Bouc-Wen under the model of piezoelectric actuators,the experimental simulation results show that within the signal(1~100)Hz frequency range,the maximum root mean square error of gaussian process regression model of piezoelectric hysteresis model is 0.5329μm,the maximum relative error is 3.23%.Comparing with traditional Bouc-Wen model,the gaussian process regression model in 10 Hz,50 Hz,80 Hz and 100 Hz modeling errors were reduced respectively by about 15%,29%,67%and 72%.