Research On Voltage Sag Source Identification Of Microgrid Based On Improved Lion Swarm Optimization Algorithm

Transient voltage disturbances are prevalent issues encountered in power grids with diverse characteristics resulting from varying sources of disturbance.These disturbances have varying impacts on power users,and thus,the protective and compensatory measures required vary.Accurately detecting and identifying the sources of transient voltage disturbances is crucial to enhance power quality and guarantee the safety and stability of power grid operation.In this study,we propose the application of an enhanced lion swarm optimization algorithm to extract feature vectors and classify voltage transients,thereby improving the capability to recognize transient voltage sources.The principal thrusts of this study are delineated as below:(1)This paper presents two enhancements to overcome the limitations of the lion swarm optimization algorithm in addressing certain problems,including inadequate convergence accuracy,late convergence into local optimal solutions,and numerous ineffective cycles in the later stages.Firstly,this study introduces weighting and adjustment factors to balance the algorithm’s capability in local and global optimization.Secondly,the Levy flight strategy is incorporated to introduce randomness to the king lion’s position to direct the convergence of the entire lion swarm,thereby strengthening the algorithm’s robustness.To validate the proposed modifications,the improved lion swarm optimization algorithm is compared against several classical population intelligence algorithms using test functions.(2)To mitigate the issue of suboptimal voltage transient feature extraction in the presence of noise,this study integrates the improved lion swarm optimization algorithm into wavelet threshold denoising.The proposed approach optimizes the threshold size and pre-processes the voltage transient signal to enhance its denoising capacity.Moreover,we conduct a comparative study with the conventional threshold selection technique to validate our proposed approach.Additionally,a novel wavelet-based theory is introduced that leverages the improved lion swarm optimization algorithm to extract voltage transient features.(3)We construct a simulation model for microgrid voltage transients to generate relevant data,which is subsequently subjected to denoising preprocessing for feature vector extraction.To enhance the accuracy of BP neural network classification and recognition,we integrate the improved lion swarm optimization algorithm into the network architecture.Specifically,the algorithm is utilized for the adaptive optimization of initial weights and thresholds,thereby enhancing the network’s classification and recognition capabilities.(4)This study addresses the issue of the time-consuming and challenging nature of voltage transient denoising,detection,feature extraction,and classification.To mitigate these issues,we develop a software platform for microgrid voltage transient analysis.The proposed platform packages the aforementioned steps and creates a graphical user interface to enhance human-computer interaction.(5)The outcomes of the study demonstrate that the enhanced denoising technique is more efficient and retains the underlying characteristics of the voltage transient signal.Additionally,the improved classifier exhibits higher correct recognition rates,further affirming the efficacy of the proposed approach.Furthermore,we corroborate the superiority of the proposed method using actual measurement data.