An Ant Colony Optimization-Enhanced LightGBM Algorithm

In recent years, the application of machine learning in non-intrusive load monitoring (NILM) for load identification has become increasingly widespread. Current research primarily employs traditional machine learning algorithms for load identification, such as Random Forest, Bayesian methods, and Deep Forest. However, these approaches often lack precision. The LightGBM (Light Gradient Boosting Machine) algorithm is known for its low memory consumption and computational complexity, yet it falls short in accuracy. To enhance the precision of machine learning in load identification, this paper introduces an improved LightGBM algorithm based on Ant Colony Optimization (ACO), termed ACO-LightGBM. Comparative experiments with other models demonstrate that the ACO-LightGBM algorithm achieves higher accuracy.