正交多智能体算法求解管网直接优化调度模型

An orthogonal multi-agent algorithm for solving the direct optimal operation model of water distribution network

针对传统遗传算法存在高维空间寻优能力较差的问题,提出采用正交多智能体算法求解管网直接优化调度模型.对智能体的随机初始种群进行正交操作,得到较优初始种群,通过智能体的竞争和自学习,找到较优解.结果表明:与正交自适应遗传算法相比,正交多智能体算法具有更强全局搜索能力和更快寻优速度,并且正交多智能体算法优化方案较大提高了水泵运行效率,可节电2.96%.

As traditional genetic algorithm （GA） can＇ t find optimal solution in a high dimension space, an orthogonal multi-agent algorithm was proposed to solve the direct optimal operation model of water distribution network. Orthogonal operation was applied to the random initial population of agents to achieve the optimal initial population, then the operations of competition and self-learning were performed among the agents to find the optimal solution. Case study shows that the orthogonal multi-agent algorithm has better global search performance and higher convergence speed than the orthogonal self-adaptive GA, and that the pump schedule policy determined by orthogonal multi -agent algorithm is more efficient and the electricity expanse is reduced by 2. 96%.