The microscopic hydraulic model (MHM) of urban water distribution network(WDN) is the foundation in WDN control, and one of the important part of watersupply system inforaiationization. Also, the MHM is a most useful tool for networkdesign and optimal operation. Building an accurate MHM of WDN, is the keyproblem needs to be solved in process of water supply system informationization.In view of MHM of urban WDN application and WDN modeling technology requirement, This paper proposes a hybrid algorithm for water distribution model calibration based on water simulation techniques, which combined the Adaptive Genetic Algorithm with Simulated Annealing(AGASA). The algorithm is finally tested and approved with two simple networks and an actual WDN, and both the fixed-time simulation and the extended period simulation were performed.In this paper, the calibration of WDN parameters is based on an implicit calibration model, which takes mitigation of the differences between the simulated values and the observed values as the objective function. And the best value of parameters in MHM is achieved indirectly. The object of calibration is those parameters in the WDN model which are difficult to measure. In the paper, pipe roughness coefficient and node demand are selected as parameters to be calibrated. The value range of pipe roughness coefficient (the Hazen-Williams coefficient, C) is defined between 80～130, the node demand adjustment range is defined between 80% ～120% of the initial value.There are many kinds of objective function in the implicit calibration model,including the least-squares method ∑(H_{mi}-H_{ti})^{2} , the sum of differences absolutevalue between actual value and WDN model simulation value ∑|H_{mi}-H_{ti}|, the sum of relative differences absolute value between actual value and WDN modelsimulation value ∑|H_{mi}-H_{ti}/H_{ti}|×100%. The second one which summarization of... |