Electric energy price prediction based on EMD-PSO-LSTM improved algorithms

To enhance the precision and dependability of advanced algorithms operating in prolonged environments and to optimise the forecasting of electrical energy prices, it has been decided to employ SE, which stands for SampEn (Entropy) in this context, to filter EMD data sequences. On this basis, PSO-LSTM (Particle Swarm Optimization-Long Short-Term Memory, PSO-LSTM) combined algorithm model is introduced to predict electricity prices. This innovative model is capable of effectively addressing the challenge of non-stationarity of signals and adapting to the complex characteristics of current time series data, due to its excellent memory ability. Subsequently, the combined algorithm is applied to the obtained component sequence. A comparison with the actual data demonstrates that the proposed model can markedly enhance the precision of civil electricity price forecasting and sustain a high level of accuracy over an extended period.