Jump to Main Content
A micro-market module design for university demand-side management using self-crossover genetic algorithms
- Xu, Fangyuan, Wu, Wanli, Zhao, Fei, Zhou, Ya, Wang, Yongjian, Wu, Runji, Zhang, Tao, Wen, Yongchen, Fan, Yiliang, Jiang, Shengli
- Applied energy 2019 v.252 pp. 113456
- algorithms, case studies, cost control, economic incentives, electricity costs, energy use and consumption, models
- Demand Side Management (DSM) is an effective measure in load configuration for microgrid power cost control and power system operation. In most extant studies, DSM in microgrid only consider directly controllable devices for load modification. The load triggered by non-controllable devices with sub-decision-makers are regarded as unchangeable load and generally not considered in DSM. A critical reason for unchangeable load is that the sub-decision makers in these microgrids may not sense and react to external dynamic electricity prices. However, these non-changeable loads in some microgrids contribute significantly to the overall power consumption of the system. Thus, a new demand side management scheme is required for these special microgrids so that the load triggered by these sub-decision makers can also response to external dynamic electricity prices. Based on a case study of a university campus, this study proposes a micro-market module to facilitate the participative behaviours of sub-decision makers in a microgrid with extra financial incentives. A university microgrid DSM optimization model is formulated to optimize the total system cost, the control of the microgrid controllable load, the behaviour of sub-decision makers and the micro-market operations are modelled. A new optimization algorithm, the self-crossover genetic algorithm, is proposed. Empirical data from a university is used to conduct a numerical study to test the proposed module and algorithm. The results show that DSM with the micro-market module can reduce the overall electricity cost of the system, and the proposed self-crossover genetic algorithm out-performs traditional optimization algorithms for the proposed model.