Abstract: Optimizing the water distribution process of the canal system is an important measure to alleviate the pressure of water resources in the irrigation area and achieve efficient water saving in agriculture. A canal system optimization model was established by taking the general main canal and main canal system in Hetao irrigation district as the research object and taking the steady flow transition and the minimum water loss as two optimization objectives. The model was solved by backtracking search algorithm（BSA）, multi-objective particle swarm optimization（MOPSO） and vector evaluation genetic algorithm（VEGA）. The results show that the water distribution time of BSA, MOPSO and VEGA is 30.96, 11.65 and 29.96 d respectively, and the skewness coefficients of the valve opening time points are-0.048, 0.068 and 0.566, which means that BSA and MOPSO can guarantee the stability of irrigation water in the canal transportation. Considering the change of the water level at the head of the canal, BSA and MOPSO are applicable to the situation where the water intake of the canal head is less and sufficient respectively. Considering the concentration of water distribution time and the geographical distribution of main canals, BSA and MOPSO are conducive to the implementation of zoning management measures.