Study On Leakage Characteristics Of Water Distribution Network Using Blind Source Separation Theory

Serious leakage of water distribution network makes water industry face hugechallenge, not only greatly reduces utility revenue, lowers the quality of water service,but also causes water resource wasting. How to quantify leakage of water distributionnetwork is the basis of leakage controlling, which gives direction for the leakagecontrolling of water distribution network. To achieve optimization of resource, themaximization of benefit, it has great practical significance to quantify physical leakage.This paper discusses three kinds of calculating methods of physical leakage whichare commonly used internationally, pointing out that the method of Minimum NightFlow needs to estimate difference coefficient and actual water consumption ofmeasurement area; The method of "water balance" exists the defect which the value ofapparent leakage water and free supplying water is speculated actually; The hydraulicmodel of pressure driven leakage has defects, such as the parameter selection is notinaccurate, which is more by experience, and the value of leakage by estimated can’t beverified, which is difficult to get the real leakage, and the hydraulic model is only backto two parameters, which makes the model form existing defect.Up to now, there is no accurate leakage mathematical method internationally. Thispaper studies on how to obtain physical leakage and a dynamic model of leakage flowrates with a few flow meters and pressure gauges installation in the water distributionnetwork. The main research achievements of this paper include:1. Experimental platform of pipeline leakage is established, and onlinemeasurement equipment such as flowmeters, pressure gauges and water meters areinstallated, which are to simulate different users, different leakages, and different typesof pipeline network for the first time.2.2independence verification is applicated to verify the independence of waterconsumption and physical leakage in leakage experimental platform; kurtosis test is toverify the non-gaussian nature of water consumption, which lays the foundation ofusing blind source separation theory for leakage separation. Blind source separationalgorithm uses non gaussian as objective function, and fixed point iteration is acted asthe algorithm for optimum leakage separation. Fast Independent Component Analysismethod programs is compiled, which is to separate the physical leakage from node flow.Using relative error and correlation coefficient to evaluate the separation performance,the result indicates that the range of the relative error of simulation leakage ratefrom0.28%to7.55%. 3. The blind source separation theory of leakage and water balance equationsis areapplied to the actual water distribution network, which is to separate the observationsignal. Through the comparison of simulated leakage rate and production rate to judgepipe network leakage control level. Results show that different of the method of blindsource separation algorithm to calculate the leakage and nonrevenue water is1.2%, sothe method of blind source separation algorithm is reliable.The significances of this paper lie in: analysising the leakage causes of the waterdistribution network, which provides a basis for decision making and management inwater supply enterprises; calibrating the parameters for the model of pressure drivenleakage, which improves the precision of the model; providing reference for themanagement and technical level of water supply enterprises for national evaluation,which promotes respectively leakage rate management and technical standards.