Abstract: The optimization control and the optimal scheduling of water supply pumping stations belong to the complicated nonlinear bounded programming category. It is difficult to model and solve this problem due to the multi-variable coupling and high nonlinear pump model as well as the synthesis of continuous and discrete variables. To remedy the defects of the flow difference model proposed by G.Cohen, an improved model on flow difference minimization is given out which satisfies the parallel characteristics of pumping stations. It proposes the simplified constraints to solve the solution complexity of precise constraints of high-efficiency operational areas, re-ducing the solution time greatly and maintaining the precision. A effectual approach is used to predict the short-term load for urban water supply system, which can predict the load in the next three or four days with the error of only 1.7%. It’s easy to satisfy the demand of real time controlling in engineering with high appliance value. 0-1 enumeration and Wolfe reduced gra-dient method is used to solve the model, the can obtain the satisfying control strategy. A lot of experimental results illumi-nated the validity of the optimization model. Some research fruits are applied to real engineering with good impression.

Key words: Load forecasting, Nonlinear programming, Optimal scheduling, Optimization control, Water supply pumping stations