多频率方波激励阻容解耦软测量的数值模拟

Numerical simulation of resistance-capacitance decoupling soft-sensing with multi-frequency AC square excitation

诸如电导率等许多参数测量问题可转化为阻容网络模型。本文采用多个频率的交流方波激励阻容网络,在响应信号与激励、阻容参数之间非线性关系的基础上,将阻容解耦测量问题转化为非线性最小二乘参数估计问题。为在线实时应用,采用最速下降法求解,并辅以仿真分析的方法,具体讨论了求解的全局最优性及迭代收敛速度等问题。针对电导率测量问题的阻容模型,实验结果表明了所提方法的有效性。该方法将作为干扰的电容效应建入模型,通过估计干扰来消除干扰能够提高测量精度,并可有效抑制测量过程中多种不确定性的影响。

Many parameter measurement problems, such as conductivity measurement, can be transformed into a resistance-capacitance （RC） network model. In this paper, multi-frequency AC square wave is adopted to excite RC network, and the measurement of RC decoupling can be converted to parameter estimation with nonlinear least squares on the basis of the nonlinear mathematical model of the response signal, excitation signal and RC parameters. Due to on-line application, steepest descent method is used to solve the parameters. Assisted by simulation analysis, the global optimality of the solution and the iterative convergence rate of the parameter calculation are discussed concretely. Aiming at the model of RC network in conductivity measurement, the experimental results show the effectiveness of the proposed method. When measuring conductivity, capacitance effect is considered as disturbance and can be eliminated by being constructed into the mathematical model for further estimation, which can improve the precision of the measurement and restrain the effects of measurement uncertainty effectively.