Study On The Calculating Method For Effective Diameter Of Pipeline In Water Network Based On The Anti-Analysis Theory

Water supply network modeling technology which is widely used at present can realize the real-time operating conditions of urban water supply pipe network, more convenient for the actual water supply network management and optimization scheduling, etc. There are a lot of old pipes in the city water supply network, these pipes have been laid for a long time, due to various reasons lead to the effective diameter occurred relatively large changes, which will directly affect the accuracy of the model of water supply network. Therefore, it is of great practical significance to calibrate the effective diameter of the urban water supply network.In this paper, the inverse problem method is used to calibrate the effective pipe diameter of the water supply network. Using BP neural network powerful nonlinear mapping ability to approximate the function relationship between finite node pressure, pipe flow rate and effective pipe diameter.Using MATLAB to prepare the effective pipe diameter calculation program for the experimental pipe network platform, The accuracy of the trained BP neural network is verified by the test samples which is generated by EPANET, comparing the simulation samples’ true value and predictive value of BP, the average absolute error is 0.841mm, the average relative error is 1.625%.Analysis the shortcomings of BP neural network algorithm, Proposed using PSO algorithm to optimize BP neural network, comparing the simulation samples’ true value and predictive value of PSO-BP, the average absolute error is 0.086mm, the average relative error is 0.164%.Finally, the test measured data under various working conditions were brought into the trained BP neural network and PSO-BP neural network, get the calculated value of corresponding effective diameter, then bring the calculated value of effective pipe diameter into EPANET to simulate the test pipe network model. Comparison with the error of node pressure and pipe flow between the measured results and the calculated results under the effective diameter. The average absolute error under various working conditions pipe flow of BP algorithm is 0.384m3/h, The average relative error is 3.922%, The average absolute error of node pressure is 0.203m, The average relative error is 2.929%, The average absolute error under various working conditions pipe flow of PSO-BP algorithm is 0.152m3/h, the average relative error is 1.535%, the average absolute error of nodespressure is 0.140m, the average relative error is 1.503%.Based on the experimental study on the calculation of the effective diameter of the pipe network shows that, using pure BP algorithm to solve the problem of effective diameter can achieve the required accuracy. When using the PSO-BP algorithm not only improve the accuracy and the training speed of the network is also improved.But this method Coding complexity and hard to operate. To solve the small water supply network can be directly used BP neural network algorithm, this method is fast and easy to implement. When dealing with the problem of large water supply pipe network, the PSO-BP algorithm is considered., in order to improve the global searching ability of the algorithm. Although the encoding operation is more complex than the simple BP neural network, but it can get higher accuracy and more stable result, and can effectively avoid falling into local optimal solution when solving complex problems. Combined with the above research results to provide reference for the effective pipe diameter calibration of pipe network.