基于无迹粒子滤波的电网动态谐波估计

Estimation of dynamic harmonics in power systems based on unscented particle filter

提出一种基于无迹粒子滤波(unscented particle filter,UPF)算法的电网动态谐波估计方法.通过无迹卡尔曼滤波算法得到电网动态谐波状态量的估计值和协方差,运用这些结果改进传统粒子滤波算法的重要密度函数,采用粒子滤波算法得到电网动态谐波的最优估计值.该方法克服了无迹卡尔曼滤波算法(unscented Kalman filter,UKF)对噪声要求为高斯分布的限制和传统粒子滤波(particle filter,PF)算法易退化的缺点,保留了UKF对非线性问题的较好处理和PF强抗干扰性能力.仿真结果表明,在高斯噪声和非高斯噪声情况下,UPF算法得到的电网动态谐波幅值、相位的估计值都更接近真实值.

This paper proposes a new method for estimating dynamic harmonics in power systems based on the unscented particle filter（ UPF） algorithm. The UPF algorithm is used to estimate the values and covariance of the state variables of the dynamic harmonics in power systems. These estimated values are used to generate the importance density function of particle filter（ PF） algorithm. The optimal estimation of dynamic harmonics in the power system is achieved by using the derived PF algorithm. The proposed method overcomes not only the restriction of Gaussian distribution noise required in unscented Kalman filters（ UKF） but also the drawback of easy degeneration for conventional PF. In addition,it retains the high performance of UKF in processing nonlinearity and the strong anti-interference capability of PF. Experiments show that the estimates of dynamic harmonic amplitude and phase by the proposed method are closer to the true values under both Gaussian noise and non-Gaussian noise situations.