A novel multi-agent based crisscross algorithm with hybrid neighboring topology for combined heat and power economic dispatch

Combined heat and power economic dispatch (CHPED) is a challenging optimization problem with character-istics like non-convexity, discontinuity, and non-differentiability. Although the crisscross optimization (CSO) algorithm can alleviate the premature convergence faced by most swarm optimization algorithms, it has a slow convergence speed to approximate the global optimum, especially at the late period of evolutionary process. To address the issue, a novel hybrid neighboring topology based multi-agent crisscross algorithm (HNT-MACSO) is proposed to enhance the balance ability of exploration and exploitation. First, based on the graph theory, the population particles of CSO are structured with two topologies, i.e., the random topology and the small world topology respectively. Second, a hybrid neighboring topology is established by applying the information relay register, aiming to improve the robustness of CSO. Third, the separate CSOs assigned to different topologies are deployed on a multi-agent system (MAS), which enables a flexible and robust distributed evolving environment for all agents to search in an independent and asynchronous optimization manner. Furthermore, five cogene-ration systems are tested, and experimental results show that the proposed HNT-MACSO outperforms other state-of-the-art algorithms in terms of solution accuracy and runtime, which confirms the effectiveness and superiority of HNT-MACSO for large-scale CHPED problems.

Automated summary

This paper proposes a novel hybrid neighboring topology based multi-agent crisscross algorithm (HNT-MACSO) to address the slow convergence speed of the global optimum in large-scale CHPED problems.