| The smart operation and management of water distribution system(WDS)is an inevitable trend of the development of water supply industry.The real-time modeling of water distribution system is widely used in the analysis,design and operation of water distribution network.The fundamental characteristic of the real-time hydraulic model is that the nodal water demand can be adjusted immediately with the real-time measured data and thus the model can represent the real state of the WDS in real time.The sampling design of measurements,the quality of measured data and the accuracy and efficiency of nodal water demand calibration algorithm are the main factors affecting the practical application of real-time hydraulic model.First of all,a greedy algorithm to optimize the sampling design for the calibration of WDS hydraulic model is developed.The assumption of pressure sensors coverage is proposed by analyzing the sensitivity relationship of the nodal pressure and the coverage degree is quantified by the nodal pressure calibration erros and the sensitivity weight coefficient.The greedy algorithm starts from the existing sensors and sequentially adds one new sensor at each step.The new sensor is selected from the candidate sensor locations that can cover the network nodes in the maximum range,thus improving the calibration accuracy.Given the number of sensors,the proposed algorithm can optimize the placement of the senoes and improve the model accuracy.The measured data of the water distribution network is inevitably polluted by background noise,leading to the degradation of the signal quality.The research review some widely used denoising methods in hope that they can be used to denoise the measurements in the water distributioin network.Methods include low pass filtering algorithm(LP),moving average algorithm(MA),Savitzky-Golay filtering algorithm(SG)and the wavelet filter algorithm(DWT).Besides,based on the feature of correlation between the hydraulic measurements,the performance evaluation index is designed to guide the parameter selection of the above denoising algorithm.The ill-posed problem caused by insufficient information to uniquely determine the nodal watere deamnd is the mainy difficulty of the nodal water demand calibration.In this research,a real-time nodal water demand calibration algorithm which can simultaneously couple multiple prior probabilistic constraints is proposed.According to the different user types,the algorithm can adopt different probability distribution as the prior information,and uniquely solve the node water demand.The algorithm can make the model output match the measeured data and ensure that the nodal water demand has a large probability density in the prior probability distribution..Computation efficacy is another factor that restrict the real application of real-time calibration algorithm,especially when dealing with large-scale network.Based on the analysis the bottleneck factors restricting the operational efficiency of the model,this study found that the process that dominates the main computation time locates in the calculation of the inverse of the Hessian matrix.According to the characteristics of the Hessian matrix involved in the WDS calibration process,we design a more efficient algorithm to obtain the covariance matrix of node water demand based on the success use of Sherman-Morrison Formula,for which the computational complexity of the inverse algorithm is significantly reduced.Numerical experiments show that the algorithm can significantly improve the calibration efficiency especially when combining with parallel programming.Through the above work,this paper forms a technical system covering sampling design,noise removal and model calibration,which provides theoretical and technical support for the establishment and application of real-time hydraulic model of WDS.This work has important theoretical and practical significance for the operation management,optimal scheduling and accident prevention of WDS. |
PDF Full Text Request