RO128440A0 - Process for optimizing the configuration of water distribution networks - Google Patents

Abstract

The invention relates to a process for increasing the efficiency of water supply systems by the optimal configuration of water distribution networks, by using the genetic algorithms. According to the invention, the process consists in representing the admissible solutions, named chromosomes, corresponding to the unlooping of a network by disconnecting some sections (4, 7, 15 and 8, 13, 19, and 3, 9, 18, respectively,), evaluating the solutions by means of an objective function (f), reproducing the solutions according to the formula: wherein frepresents the function for adaptation of the chromosome i, after which there is carried out the crossover recombination and then the solving of mutations by means of a specific mutation operator, choosing at random one of the chromosomes, thereafter there is generated a random number ranging between 1 and the length (L) of the chromosome chain and there is carried out the alternation of the element of said order number within the said chain between 0 and 1.

Description

PROCEDURE FOR OPTIMIZING THE CONFIGURATION „__________ '' jOFÎC'Ut. OF STATE FOR INVENTIONS AND TRADEMARKS

WATER DISTRIBUTION NETWORKS Patent Application I

I Nr. Λ.

Yes ^r a deposit ..1. ^ .: 02: .211¾ ....

The present invention relates to a way to increase the efficiency of water supply systems, by optimally configuring water distribution networks, using the method of genetic algorithms. The field of use of the invention is obviously very wide and encompasses everything involved in water distribution networks, be it drinking water, domestic or industrial.

It is known that AG genetic algorithms are adaptive algorithms, which determine the optimal solution of an optimization problem based on specific mechanisms of genetics and natural selection. For this purpose, AGs use a set of solutions, often called chromosomes that, during the evolution from one generation to another, change their structure, orienting towards an acceptable solution (optimal or suboptimal). The initial population has a randomly established composition, and its evolution takes place according to the principle "the most adapted survives".

The application of traditional optimization methods, which ensures the minimization of an objective function, taking into account all three mentioned aspects, is laborious and involves the adoption of simplifying assumptions, which greatly influence the selected solution.

The parameters used in a genetic algorithm determine the efficiency of the algorithm and the quality of the solutions provided

Genetic algorithms (GAs) have been used so far mainly to determine the optimal configuration of urban electricity distribution networks.

Solutions for the application of genetic algorithms in the field of water networks are also known, such as:

Patent document number US 7,593,839 Bl, entitled “Method for optimizing the design and rehabilitation of water distribution systems”, in which the criteria chosen for optimization are: minimum cost, maximum quality and minimum cost / quality ratio

This method has the disadvantage that it does not provide clear details on how to implement these algorithms, such as recombination, mutation and ranking of candidate solutions. Also, the solution does not provide details about the structure of chromosomes and does not present concrete examples of application.

Another method of the patent document with known humus automatic calibration using genetic algorithms is presented in US 2003/0093236 A1, entitled "Method and system - ete7-r ^ water distribution model", which allows the calibration of a ^ rele de

IONESCU GHEORGHE - CONSTANTIN

CP ck-2 0 1 2 - 0 0 6 5 8 -1 4 -09- 2012 water based on data entered by the user. The process uses genetic algorithms to perform this calibration, but has the disadvantage that it does not give details about their parameters.

Another U.S. Patent No. 4,562,552, entitled "Method and Apparatus for Controlling Flow and Pressure in Water Distribution Networks," uses genetic algorithms to optimize flow and pressure in a network and does not refer to configuration optimization. water distribution network. The disadvantage of this solution is that it does not detail how to implement algorithms.

The technical problem that the claimed invention aims to solve is to achieve a concrete process, perfectly applicable in practice, which allows the optimization of the configuration of water distribution networks, including how to implement the algorithms.

The invention solves this technical problem by using the method of genetic algorithms already established for electrical circuits, to identify the optimal operating configurations of annular water distribution networks with a complex looping structure.

The method of optimizing the configuration of water distribution networks, according to the claimed invention, claims the following advantages:

Establishes a rational direction of water circulation, so that water flows on the shortest path to the consumer;

Sizes the network by establishing the number of nodes, the number of sections and the diameters related to the sections;

Determines the most advantageous sectioning by establishing the optimal sectioning points that ensure the operation in a branched scheme;

Optimizes the efficiency of water supply systems and facilitates their management;

Reduces pregnancy losses as the number of generations taken into account increases;

Ensures the water supply of a maximum number of consumers in case of a failure in the ring network.

The following is an example of a practical embodiment of the method according to the invention for optimizing the configuration of an annular water distribution network, with a complex looping structure, in connection with:

Fig. 1: example of complex loop configuration for water distribution;

Fig. 2: Chromosomal chains corresponding to the network debugging in Fig. 1.

Fig. 3 the chain of descending chromosomes, derived from the crossing of the parent chromosomes in Son 2.

IONESCU GHEORGHE - CONSTANTIN

G_-l 012-00658-1 4 -09- 2012

The process of optimizing the configuration of water distribution networks using genetic algorithms requires five steps, namely

Representation of admissible solutions

Evaluation of solutions

Reproduction

recombination

Induction of mutations

Of the five steps listed above, the last four are repeated iteratively until a stopping condition is met.

For the design of a water distribution network in a complex loop configuration, of the type in fig. 1, it is necessary to solve the following problems:

Establishing a rational flow of water so that water flows on the shortest path to the consumer;

Network sizing (establishing the number of nodes, the number of sections and the diameters related to the sections);

Determining the most advantageous sectioning (establishing the sectioning points that ensure the operation in a branched scheme).

In order to select the optimal solution according to a given criterion, it is necessary to analyze all the mentioned problems simultaneously. The number of possible combinations of route choices is usually very large, and the sizing of the network and its sectioning is quite laborious.

Representation of admissible solutions

An admissible solution, which in the context of AG is called a chromosome, is represented by a chromosomal chain of finite length. In most studies in the literature, the values of the elements in the chromosomal chain have a binary coding.

For the water distribution network in fig. 1, the chromosomal chains of FIG. 2 represent admissible solutions, corresponding to the network deployment by disconnecting sections 4, 7, 15 and 8, 13, 19, respectively 3, 9, 18.

Evaluation of solutions

The admissible solutions will be evaluated using the objective function, which describes the analyzed optimization problem. Based on this assessment, the degree to which the solution is adapted to the problem is established, hence the national name for the respective function.

IONESCU GHEORGHE - CONSTANTIN f λ (Χ-2 0 1 2 - 0 0 6 5 0 -1 4 «09-2012“ adaptation function. ”It should be noted that, by their very specificity, the AG evolves in the sense of maximizing adaptation function f. Therefore, if the considered problem aims at minimizing the objective function F (as is the case with many energy problems), the adaptation function is defined as its inverse f = F ¹ .

During this stage, the information entered in each chromosome is decoded and, based on it, the adaptation function for each chromosome is evaluated.

Reproduction in this stage, takes place the creation of a new population based on the degree of adaptation of each chromosome, established in the previous stage. The new population is generated based on exact copies of existing chromosomes: the chromosome with the highest adaptation function value will produce the most copies, and the chromosome with the lowest adaptation function will produce the fewest copies.

The most common application of the breeding operator is the so-called proportional breeding. In this case, each chromosome produces a number of copies proportional to the ratio.

(1) where fj represents the adaptation function of chromosome i.

Recombination (crossing)

The optimal solution of the analyzed problem is disseminated in the information contained in all the chromosomes of the population. In order to aggregate the different segments of the information corresponding to the optimal solution, the AG uses the recombination stage. The most commonly used recombination technique is "crossing". In this case, two “parent” chromosomes are chosen from the current population, through which two other “offspring” chromosomes will result. For a length L of chromosomes, the crossing is performed at a point M randomly generated in the interval (1, L). For example, for the first two "parent" chromosomes in FIG. 2, if the crossing occurs in point 6, the “descendant” chromosomes represented in fig. 3. It is found that both "offspring" chromosomes are admissible solutions to the problem, leading to new outflows 4,8,13 and 7,8,15, which did not exist in the previous population.

IONESCU GHEORGHE - CONSTANTIN

Λ-2 0 1 2 - 0 0 6 5 8 - 1 4 -09- 2012

mutations

Although the recombination step ensures the enrichment of the essential information stored in the chromosomal chains, it is still possible that, starting from an initial generation with a relatively deficient structure, any recombination between chromosomes may not lead to the optimal solution.

For example: if for the network in Figure 1 the optimal unbundling corresponds to a triplet that also contains section 11, the overall solution of the optimization problem cannot be reached if the initial generation contains only the three chromosomes in Figure 2, because none of them has bit 11 set to 1. Elimination of this shortcoming is possible if a specific mutation operator is applied to some of the chromosomes with a certain probability, which randomly chooses one of the chromosomes, after which it generates a random number between 1 and L (length of the chromosomal chain) and tilts between 0 and 1 of the element with the respective sequence number in the chain.

Claims (1)

CLAIM Process for optimizing the configuration of water distribution networks characterized by the use of genetic algorithms and consists of 5 stages, namely: representation of admissible solutions for network debugging by disconnecting sections (4, 7, 15) and (8, 13, 19) , respectively (3, 9, 18), evaluation of solutions using the objective function (f), reproduction of solutions according to formula (1), recombination by cross-linking and then solving mutations using a specific, mutation operator who randomly chooses one of the chromosomes, after which it generates a random number between 1 and the length (L) of the chromosomal chain and swings between 0 and 1 of the element with the respective serial number in the chain.