Multiplicative neuron models for very short-term load forecasting

The article presents an in-depth analysis of multiplicative neuron models for very short-term load forecasting in power systems. It begins by highlighting the importance of load forecasting in efficient power system planning and operation. The study compares conventional methods with advanced multiplicative neuron models, which offer advantages in representing nonlinear relationships between load and influencing factors. The research focuses on using only demand data to overcome the limitations of temperature-based forecasting. The paper discusses the design and training of various neuron models, including multiplicative spiking neuron models and quadratic integrate-and-fire neuron models. The performance of these models is evaluated using the Mean Square Error (MSE) and Akaike Information Criterion (AIC). The results demonstrate that multiplicative neuron models require fewer neurons and iterations, significantly reducing training time and computational complexity. The study concludes that these models can outperform conventional models like Multi-Layer Perceptron (MLP) in very short-term load forecasting, offering a more efficient and accurate solution. Future research directions are also suggested, including the exploration of other normalization techniques and the integration of additional factors such as weather conditions and time factors.