In view of the topological designed problem of power collection system for photovoltaic power plants, the multi-objective and hierarchical optimization based on the multi-elite co-evolutionary partheno-genetic algorithm (MCPGA) is proposed. Firstly, the composition of power collection system is expounded, the corresponding mathematical optimization model is established. Secondly, the MCPGA is used to solve the model, the reliability of collection system is evaluated by Monte Carlo. The economy goal and reliability index are used to evaluate the design of power collection system. At last, simulation results show that this method is effective and practical, convergence of the MCPGA is fast, which can give a coordinate plan of power collection system for photovoltaic power plants.