A multi-strategy improved particle swarm optimization algorithm and its application to identifying uncorrelated multi-source load in the frequency domain

As a type of indirect method, the traditional frequency approach to load identification establishes a systems frequency response function (FRF) and calculates loads using its inverse and the responses. Based on the established FRF method, a novel identification approach that transforms the inverse problem into a forward single-objective optimization model is proposed. Furthermore, a multi-strategy improved particle swarm optimization algorithm (MsiPSO) for identifying uncorrelated multi-source load in the frequency domain is also proposed. Depending on the specific application, MsiPSO initializes the swarm based on domain knowledge. It applies asymmetric and nonlinear strategies to adaptively set the control parameters and genetic operators to strengthen the diversity of the population and avoid local optima. In the experiments, MsiPSO is compared with the general particle swarm optimization (PSO) algorithm and some well-performing variants. A simulated model is then defined to validate the load recognition accuracy of the proposed approach. The experimental results show that MsiPSO is competitive with other methods in terms of convergence, stability, and precision. It has a higher recognition accuracy and is faster than traditional load identification methods and standard PSO.