Research And Implementation Of Key Technologies Of Urban Water Supply Dispatching System

The urban water supply scheduling system gives the full range of operation and management on the whole water supply process.So that the water supply system can be run under economically reasonable conditions and sent drinking water to the citizen continuously.The water comsumption is one of the most important indicator in the process of water scheduling in urban.But the result of water forecast has the lower accuracy.The new method merges the BP neural network into demand forecasting has been proposed in this paper.This new method inherits the advantage from BP neural network,such as anti-interference,nonlinear and self-learning.The accuracy and adaptability are higher than in regression analysis method or time series analysis method.From a practical view,the error of BP neural network using in prediction is very small,only 1.43%.The data fit is also very close to the real situation.The scheduling solution the one of the method to adjust the state of network,and it is the necessary function in water supply scheduling system.But exhaustive way has been used in the current solution generation algorithm.More than 20-30 minutes will be cost in offline solution generation only once,that reduces the availability and timeliness.In this paper we combined pump characteristic curve and two stages schedule.Translates the exhaustive way into numerical calculation.This optimize solution ensures the performance and energy savings effects.In practical applications,ZZ City pump operation power can be saved about 15.7 million kw,saving the cost of water supply more than 4000 Yuan per day.With the improvement and expansion of the distribution network.The performance degradation makes the water supply scheduling system can not be normally used.The difficult problem has been solved by hydraulic computing service in this paper.The hydraulic computing service contains quasi-distributed structure.Trying to assign the compute tasks to which running on single server serial multiple servers parallel.The utilization of the resource on the server is improved,the computing time is cut and the stability and timeliness are increased.This greatly improves the computing performance,reducing the execution time of the entire scheduling process.The time by parallel execution can be shortened nearly 60%.Practical application shows,after the urban water supply network optimization and adjustment of the entire system,the accuracy and precision has been significantly improved,response period shortened,availability and timeliness have been fully reflected.It is can be seen that optimization method proposed in this paper is feasible and successful.