基于蚁群粒子群优化的卡尔曼滤波算法模型参数辨识

针对复杂的低压配电网通信环境,提出一种基于蚁群粒子群融合的无先导卡尔曼滤波(UKF)算法的模型参数辨识方法。对于电力线多径信道传输模型,采用具有最小均方误差估计效果的UKF辨识算法。针对UKF算法通过试验调节难以取得最佳滤波效果的问题,提出基于蚁群粒子群算法优化UKF噪声矩阵的方法,同时引入蚁群算法将惯性权重离散化以提高粒子群算法的搜索效率,克服其容易发生早熟收敛的缺点。试验和仿真结果表明,采用该优化算法辨识电力线信道模型可克服参数的分散性,提高拟合精度并缩短辨识时间。

Model Parameters Identification of UKF Algorithm Based on ACO-PSO

In view of the complicated low voltage communication environment of distribution network, an unscented Kalman filter (UKF) algorithm is proposed based on ant colony optimization-particle swarm optimization (ACO-PSO) to identify multipath channel transmission model parameters. The UKF algorithm using the minimum mean square error is adopted to estimate the identification effect for the power line multipath channel transmission model. Considering the difficulty in obtaining the optimal filtering effect through test adjustment by the UKF algorithm, this paper proposes an ACO-PSO algorithm based method for optimizing the noise matrix of the UKF algorithm, while introducing the ant colony algorithm in discretizing the inertia weight parameters to improve the searching efficiency of the PSO algorithm and overcome the shortcomings of being prone to premature convergence. Test and simulation results show that identifying the power line channel model with this optimization algorithm can overcome the dispersion of model parameters, improve the fitting accuracy and shorten the identification time.