Flexible energy load identification in intelligent manufacturing for demand response using a neural network integrated particle swarm optimization. (February 2022)

• Record Type: Journal Article
• Authors: Islam, Md Monirul; Sun, Zeyi; Qin, Ruwen; Hu, Wenqing; Xiong, Haoyi; Xu, Kaibo
• Other Names: Luo Xichun guest-editor.
• Is Part Of: Proceedings of the Institution of Mechanical Engineers. Volume 236:Number 4(2022)
• Page Start: 1943
• Page End: 1959
• Publication Date: 2022-02
• Subjects: Flexible load -- manufacturing system -- demand response -- neural network -- particle swarm optimization; Mechanical engineering -- Periodicals; 621.05
• DOI: 10.1177/0954406220933652
• Languages: English
• ISSNs: 0954-4062

Abstract

Various demand response programs have been widely established by many utility companies as a critical load management tool to balance the demand and supply for the enhancement of power system stability in smart grid. While participating in these demand response programs, manufacturers need to develop their optimal demand response strategies so that their energy loads can be shifted successfully according to the request of the grid to achieve the lowest energy cost without any loss of production. In this paper, the flexibility of the electricity load from manufacturing systems is introduced. A binary integer mathematical model is developed to identify the flexible loads, their degree of flexibility, and corresponding optimal production schedule as well as the power consumption profiles to ensure the optimal participation of the manufacturers in the demand response programs. A neural network integrated particle swarm optimization algorithm, in which the learning rates of the particle swarm optimization algorithm are predicted by a trained neural network based on the improvement of the fitness values between two successive iterations, is proposed to find the near optimal solution of the formulated model. A numerical case study on a typical manufacturing system is conducted to illustrate the effectiveness of the proposed model as well as the solution approach.