A novel algorithm for energy market risk forecasting. (April 2022)

• Record Type: Journal Article
• Authors: Wang, Xiaofei; Pei, Pei
• Is Part Of: Computers & electrical engineering. Volume 99(2022)
• Publication Date: 2022-04
• Subjects: Artificial intelligence -- Firefly algorithm -- Forecasting -- Market price risk -- Optimization; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2022.107813
• Languages: English
• ISSNs: 0045-7906

Abstract

Highlights: Proposing a novel hybrid SVR based model for the market price forecasting considering the risk error in the analysis and evaluation. Proposing the firefly algorithm as a swarm based solution for optimizing the SVR hyperparameters according to the market price. Proposing a two-stage modification approach for the FA (MFA) to enhance the search capabilities and boost the diversity of solutions. Abstract: This paper proposes a novel artificial intelligence based approach for the accurate forecasting of the market cost incorporating the error risk. The proposed model makes use of a hybrid evolving solution based on firefly algorithm and support vector regression (SVR) for getting to the highest accuracy and precision. In contrast to the artificial neural network models which face the overfitting problem, the SVR would keep a limit on the forecast model complexity and thus would escape from the overfitting concerns. In order to get into the maximum performance of the SVR, one need to adjust the hyperparameters optimally. This article proposes the firefly algorithm which mimics the social behavior of these insects in their colony. In addition, a new modification method is introduced which would add up the algorithm population diversity and thus enhance the search results. The appropriate performance of the proposed hybrid AI base model is assessed on the typical market price datasets. Graphical abstract: Image, graphical abstract