Research On Forecasting Methods Of Natural Gas Load Based On Particle Swarm Optimization-Least Squares Support Vector Machine

As a green, economical, efficient, convenient, safe and environmental friendly energy, Increase the proportion of natural gas in the energy consumption structure, not only conducive to the promotion of energy conservation, but also promote the sustainable development of economic and social, while natural gas as the citygas, is one of the main energy of the modern city, and closely related to people’s lives and industrial production.Therefore, It’s very important to have a scientific forecast for natural gas load, which has a very important significance to analysis and formulate the natural gas pricing policy and the residents’ consumer strategy.Primarily, based on the background of the natural gas load forecasting, the research status, and the significance of research, the paper analyzes the characteristics of natural gas, and the non-linear relationship between the gas load forecasting and the various affecting factors. Then Support Vector Machine (SVM) which has many advantages on solve nonlinear, high dimension and other problems is proposed in this paper. Based on the research, we change the inequality constraints to equality constraints, the training error square to slack variables, and Least Squares Support Vector Machine (LS-SVR) is proposed, which greatly accelerate the solving process and need less optimized parameters.While as the lack of robustness, finally, Weighted LS-SVR algorithm is proposed. Considering it’s crucial of the model parameters to the prediction accuracy, we proposed the Particle Swarm Optimization (PSO) to optimize the parameters in WLS-SVR, in order to improve the accuracy of the prediction. The simulation results show that WLS-SVR is better than SVR, and PSO-WLS-SVR model is the best, which shows the effectiveness and superiority of the PSO-WLS-SVR, and also shows it has certain research value and social significance.