Based on the distribution of wind or solar resources in northwest,a small wind-solar-battery hybrid power system was proposed. Based on the statistical situation of perennial climate changes,the mathematical model was built considering multi optimization objectives including investment cost, system reliability and utilization of wind-solar resources. The different situations in both stand-alone and grid-connected modes were optimized. The vector-evaluated genetic algorithm(VEGA) was used to solve the optimal problems. The optimal allocations of distributed source and batteries were got in micro-grid.