Research On Market Price Construction Of Natural Gas Based On Multi-objective Optimization Algorithm

In the context of increasing global environmental pollution and energy demand,as an environment-friendly and clean energy,natural gas has gradually become the best choice for countries to adjust their energy structure and transition to low-carbon development.Over the years,the global natural gas market has been tight in supply,and the continuous increase in global natural gas demand has led to rising natural gas prices in the three major regions of North America,Europe and Asia-Pacific.In addition,the price of natural gas is affected by the international situation,supply and demand relationship and other factors,which makes the trend analysis of natural gas price become difficult.In this context,it is very important to study the prediction of natural gas price.Scholars from home and abroad have done lots of research on the price prediction of natural gas,and achieved certain research results,which promoted the development and innovation in the field of energy prediction.However,natural gas series have strong nonlinearity and randomness,so the statistical prediction method based on linear hypothesis has certain limitations.With the development and application of artificial intelligence,artificial intelligence forecasting method has gradually attracted the attention of scholars because of its good nonlinear fitting ability.Single artificial intelligence prediction methods can effectively extract data features and improve prediction accuracy to a certain extent,but the prediction process is prone to over-fitting and local optimal conditions,leading to distortion of prediction results.In contrast,the combined prediction model including data preprocessing technology and optimization algorithm can better improve the prediction performance of the model.At present,most of the proposed combination models use single objective optimization algorithm,focusing on the improvement of prediction accuracy,but ignoring the stability of prediction results.Therefore,in order to improve the prediction accuracy and enhance the prediction stability,this paper proposes a novel combined prediction model based on the improved complete ensemble empirical mode decomposition with adaptive noise ICEEMDAN and multi objective grasshopper optimization algorithm MOGOA.This paper mainly includes the following aspects:first,the original sequence is decomposed and reconstructed by using ICEEMDAN data preprocessing technology to eliminate the interference information in the sequence,and then a relatively smooth new sequence is generated for subsequent model prediction;secondly,this paper uses BPNN algorithm,ELM algorithm,LSTM algorithm and CNN algorithm respectively to predict the natural gas price series after denoising,forming four groups of different prediction results.furthermore,MOGOA algorithm is applied to optimize the weight coefficient,and the above prediction results are integrated,so as to achieve the best overall prediction effect.Finally,this paper proposes a relatively complete index evaluation system,several simulation prediction experiments and three validity tests to comprehensively and scientifically evaluate and further verify the prediction performance of the combined prediction model proposed in this paper.The simulation results show that the ICEEMDAN-MOGOA combined forecasting model has better prediction performance in point prediction and interval prediction than 18 comparison models(i.e.,6 single prediction models and 12 combined prediction models),and the prediction results have higher accuracy and stability.Based on sorting out and summarizing many domestic and foreign researches on energy price forecasting,the innovations in this paper can be listed as:first,this paper proposes a novel combined prediction model,a combination of data pretreatment technology,multi-objective optimization algorithms and the advantage of multiple single forecasting model,which can be more effective,accurate data of natural gas price forecast.Secondly,the advanced data preprocessing technology,namely the improved complete ensemble empirical mode decomposition with adaptive noise ICEEMDAN,is used to reconstruct the original sequence,eliminating the high-frequency noise of the data and extracting more information from the data.In addition,the stability and accuracy of the prediction are also paid attention,the multi-objective optimization algorithm is designed to weight the results of multiple single prediction models,which improves the prediction accuracy and stability of the model.Furthermore,this paper designs an integrated index evaluation system and some different simulation experiments to comprehensively and scientifically evaluate the performance of the combined prediction model.At the same time,three sets of validity tests are introduced to further verify the performance of the combined forecasting model,making the experimental results more reliable.Finally,the combined prediction model proposed in this paper has strong applicability and stable performance in point prediction and interval prediction experiments,which provides a new idea for the field of energy price prediction.The shortcoming of this paper is that due to data limitation,it fails to consider various factors affecting natural gas price.In addition,this paper only selected some representative prediction methods for comparison,without fully considering other excellent algorithms or models.Therefore,the prediction system proposed in this paper still needs to be continuously improved and perfected in the future research work.