DE102008048747B3 - Method for leakage recognition in area of supply network, and for infrastructure network for transporting fluid or gas or long-distance heating network, involves determining computer forecast value of total consumption quantity - Google Patents

Abstract

The method involves determining a computer forecast value of total consumption quantity from individual consumers of the area during a measuring period by statistic projection. An actual value of the total consumption quantity is determined by measuring the supply flow in the area and discharge flow from the area during the measuring period. The computer forecast value is subtracted from the actual value, where a positive difference is a noted for a leakage.

Description

The The present invention relates to a method and a use for leak detection and leak detection in the area of a supply network.

Especially Drinking water belongs with the most important commodity in the twenty-first century. In Water distribution networks occur significant water losses. These amount in developed countries to 25%, whereby for Germany 8%, and in emerging and developing countries more than 50% (Source: [1]).

The Preservation of this property poses a major challenge to leak detection and location in water networks. The detection of water loss through a mass balance, in which all feed-in quantities and in particular measured the water consumption of all individual water consumers be exceeded the effort limits the feasibility. Furthermore, require Legislation precise consumption measurements and consumption records all customer data.

conventional Way is used to detect leaks in water networks only the total amount of water that flows into the network at a certain time, like for example the period at night between 2 and 4 o'clock is, measured. These values create a time series that points to it was analyzed, whether a sudden Increase in consumption and thus a possible leakage exists. A Accounting for the acceptance of the system is conventionally not possible. It Rule of thumb for the typical water consumption of certain consumers in front. But these are very general and can, in particular time-limited special effects should not be considered with sufficient accuracy.

The usually very big Water supply networks are divided into water supply zones. These Zones are in turn subdivided into subzones that, as this is due British engineers is shaped be referred to as district meter area (DMA). The DMAs will be designed so that they have only one inflow, its flow is measured. From the observation of this flow measurement is on irregularities closed in water consumption and thus leakages. Concrete becomes more conventional Way a so-called "not flow analysis "performed. in this connection is characterized by detailed recording of nocturnal inflows into a DMA (eg every 5 seconds between 2: 00-4: 00 o'clock) a minimum inflow value determined, here also called background consumption, which the normal nightly (Minimum) consumption and existing (especially small) leaks includes.

On Based on these minimal inflow into a DMA during low-consumption Night time, for example, between 2 and 4 clock, then only One value per night is provided over days and weeks created a time series. The, in particular sudden, increase of these minimum Consumption values, for example, by a threshold value exceeded can be recognized, may have as a cause a new leak.

to Localization or leakage location is usually performed a step test. For this At times of low consumption, one successively separates small areas from the DMA and observed the change in consumption. areas to a strong inexplicable Waste of consumption then examine further for leaks.

alternative can with noise meters spot the water system to be leaked on leakages and by viewing the noise correlation the Leakage point to be calculated.

Both called conventional Procedures are not suitable for a permanent monitoring. Step Tests are associated with a lot of effort since the affected Households need to be informed of the shutdown and a replacement supply must be ensured.

noise measurement needed a high effort, since the measurements only by specialists on site carried out can be. Furthermore these investigations are always only possible locally. In addition, both conventional Procedure can only be applied in times of low consumption, so that the measurements are not disturbed too much by consumption fluctuations.

The usual Basically leak detection procedures only allow the detection of new leaks. About the retrospective creation a water balance over for example, the period of one year is additionally attempted Areas with big ones Identify water losses. Such a water balance will for example, based on a difference in the inflow values that the Feed quantity is in an area and the billed consumption quantities determined in the year. However, such an approach does only rough clues ready.

According to the state of the art, new, cheaper measuring devices enable the monitoring of physically relevant values such as pressure, flow and volume in numerous places. This generates a large amount of data. An efficient and reliable evaluation with the aim of a reliable Le Recognition is the subject of this patent application.

The DE 28 41 674 discloses a method and apparatus for controlling losses from public water pipe networks. In particular, a method is disclosed for monitoring pipeline systems, such as public drinking water networks, for losses and for locating leaks, in which flow characteristics are detected and evaluated in relation to the fluid supplied to and discharged from the pipeline system and in which the leaks are then carried out targeted measurement and positioning measures are located. The local method is characterized in that a plurality of control points is set up in the line system, that the flow characteristics detected there, especially during the time of low flow, the flow rate, flow direction, flow noise, fluid pressure, are evaluated for determining measuring points in the line system the latter measuring points are then continuously and continuously monitored and the measured values obtained therein, in particular changes in the flow behavior, are evaluated, and that regularly measured values are used by the control points in addition to the detection and localization of the fluid loss causing leaks at fluid losses.

It The object of the present invention is a method and a use for the detection of leaks in an area of a supply network to provide such that already existing or emerging Leaks, small leaks with respect dynamic, also not periodic special effects are detected. Furthermore, should a further location of a detected leak in the area will be made possible.

special effects or special events are, for example, the use of lawn sprinklers, Holidays or sporting events, ie one off events, the one related cause special consumption.

The Task is by a method according to the main claim and a Use according to the additional claim solved.

A Leak detection in supply networks is done by balancing the inflow and the total consumption determined by statistical extrapolation from individual consumers. The present invention is based on a complete Mass balance.

According to one The first aspect is a method for leak detection in an area a supply network provided with the following steps. Using statistical extrapolation, a determination of a Extrapolation value of a total consumption amount of individual Consumers of the area during at least one measurement period. A measurement period is especially within someday. There is a further measurement of the inflows into the Area and drains out of the area during the measurement duration, where an actual Value of the total consumption quantity is determined. Finally done subtracting the extrapolation value from the actual one Value, where a positive difference value is an indication of one already existing leakage in the area. In principle, the measuring periods can be arbitrary chosen become.

It is by extrapolation, with known electoral projections comparable to the measured consumption of some Consumer, which determines total dynamic usage in the system. With such an extrapolation value, leakage detection, based on a mass balance and thus leakages are detected. The claimed method can be carried out particularly advantageous automated.

It will now be for any area that has a DMA and / or virtual zone / DMA (for explanation of terms see also further back), a mass balance, as well as Consumption balance can be called created.

The supply-side inflows into any zone or area and, where appropriate, outflows from this zone to other zones or areas will be over the defined time periods, such as between 2 and 4 o'clock at night and for 24 h, measured and summed up. From this sum value then the for this Zone or for This area calculated extrapolation values for all consumers across all Subtracted from classes. Ideally, that is, with a correct estimate and no If there are leaks, this value is 0.

One significantly positive value is an indication of an already existing one Leakage in this zone or in this area.

One Leak detection system based on the claimed embodiment, allows the detection of leaks already before installing the Leak detection system in the supply network.

By the technique of extrapolating the current total consumption for a certain period of time from the measurements of a few consumers, the mass balance can be set up with a very good accuracy, so too relatively small leaks can be identified. The values obtained in this way are much more accurate than the estimates hitherto obtained by rule of thumb. This makes it possible for a network operator to make economic sense with further measures for eliminating the leak, as is the case, for example, of so-called noise logging. The new method differs from the existing one particularly in that dynamic, non-periodic special effects or special events are detected. This leads to greater reliability, a lower error alarm rate, since, for example, special event consumptions such as sports broadcasts on television, lawn sprinklers in fair weather and the like are automatically detected. This information may also be useful for adjacent non-consumable zones, as it can explain correlating effects.

Further advantageous embodiments are claimed in conjunction with the subclaims.

According to one advantageous embodiment is for each example Every day or weekly Repeating measurement in particular one day each time series of the difference values, whereby statistically a newly created leakage or a special effect detected in the area becomes. With these values, that is, for example, every 2 o'clock until 4 o'clock at night and every 24 hours a value he testifies, can last for several days or weeks away time series are formed by means of statistical Methods can be analyzed (see [2]). Simplified, it can be said that a clearer Increase in difference values then an indication of a newly created leakage is. The measurement duration can be, for example, daily or weekly to repeat. In this way a measuring cycle is determined. Other Measuring cycles are also possible.

According to one Further advantageous embodiment, a division of consumers takes place of the area in consumption classes; randomly selecting consumers from each Consumption class; measuring an average consumption amount of chosen Consumers per consumption class for the duration of the measurement and a statistical Extrapolate the average consumption quantities of all consumption classes using linear regression or as a stratified sample for Determine the extrapolation value.

It be first Consumers based on their consumption amount, for example the annual statements are known, of their kind, for example, a Detached house with garden, an apartment building, a small business, a large company etc. and, where appropriate, their consumption profile. A consumption profile shows the course of a consumption, for example, over a day. Then each class becomes random selected some consumers. This selection is made once. The selected consumers and with their approval, equipped with flow sensors. Through an appropriate installation of data transmission technology can now timely flow readings are sent to a central office, so that the statistical extrapolation to determine an extrapolation value can be done. The quantities withdrawn from the selected consumers are different Time periods, for example between 2 and 4 o'clock in the morning and for 24 hours measured and transmitted. This can be a number basis for the extrapolation of consumption.

Out These measurements are extrapolated for the total consumption during the performed periods. The measurement is statistically a stratified random sample. From this is concluded on the total consumption. They are the following two approaches to disposal.

Linear regression (see [2]):

der Vertreter der Verbrauchsklassen (i = 1, 2, 3, ..., N) gemessen. Dann werden mittels einer Regressionsgeraden Werte b₀, b₁, b₂, ..., b_N bestimmt, derart, dass eine Schätzfunktion der Bauart

aufgestellt werden kann. Mittels eines Korrelationskoeffizienten kann die Güte der Schätzfunktion gemessen werden. Der Wert b₀ ist ein Schätzwert für den Background Leckage, wohingegen sich nach

der aktuelle Verbrauch schätzen lässt. Der Koeffizient zu den separat gemessenen Großverbrauchern sollte nahe 1 liegen. Er kann auch vorab mit 1 gewählt werden und aus der Ausgleichsrechnung entfernt werden. Ferner sollten alle Koeffizienten positiv sein. Bei der Zunahme von weiteren Verbrauchsklassen oder Verbrauchern muss die Ausgleichsfunktion nicht notwendig neu aufgestellt werden. Sie kann unter Umständen auch durch manuelles Anpassen der Koeffizienten und Ergänzen von weiteren Summanden korrigiert werden. Bei der Initialisierungsphase wird von der Annahme ausgegangen, dass sich in diesem Zeitintervall keine neuen Leckagen bilden.It is about a certain lead time at times t = 1, 2, 3, ..., T belonging to an initialization phase, both the total consumption V ^t , as well as the average consumption

the representative of the consumption classes (i = 1, 2, 3, ..., N) measured. Then values b ₀ , b ₁ , b ₂ ,..., B _{N are} determined by means of a regression line such that an estimation function of the type

can be set up. By means of a correlation coefficient, the quality of the estimation function can be measured. The value b ₀ is an estimate for the background leakage, whereas after

the current consumption estimate. The coefficient for separately measured large consumers should be close to 1. He can also be chosen in advance with 1 and be removed from the compensation calculation. Furthermore, all coefficients should be positive. With the increase of other consumption classes or consumers, the compensation function does not necessarily have to be re-established. It may also be corrected by manually adjusting the coefficients and adding more summands. At the initialization phase it is assumed that no new leaks occur during this time interval.

hochrechnen, der im Folgenden mit dem gemessenen Verbrauch verglichen wird. Je nach Zusammensetzung des zu untersuchenden Gebietes muss/kann die Hochrechnungsfunktion für jeden Wochentag separat aufgestellt werden.From measurements of the current average consumption v _i , the representative of the consumption classes i = 1 2, 3, ..., N is a current total consumption after

which will be compared below with the measured consumption. Depending on the composition of the area to be examined, the extrapolation function must / can be set up separately for each day of the week.

• 1. Zunächst ist die Grundgesamtheit vollständig und überlappungsfrei in Teilpopulationen zu unterteilen.
• 2. In jeder Schicht wird unabhängig voneinander eine einfache Zufallsauswahl getroffen.
• 3. Aus den schichtspezifischen Stichproben wird pro Schicht ein Mittelwert separat geschätzt. Diese Mittelwerte werden dann zu einer Schätzung des Gesamtwertes gemittelt.
• 4. Analog müssen die Varianzen erst separat und dann im Gesamten geschätzt werden.

Stratified sample (stratified sample): In a stratified sample, the following steps are taken: (see [2])

• 1. First, the population is to be subdivided completely and without overlap into subpopulations.
• 2. In each layer, a simple random selection is made independently.
• 3. An average is estimated separately for each layer from the layer-specific samples. These averages are then averaged to give an estimate of the total score.
• 4. Similarly, the variances have to be estimated separately and then as a whole.

wobei folgende Größen verwendet wurden:

V:

Gesamtverbrauch

N:

Anzahl der Verbrauchsprofile

v _i:

gemessener Durchschnittsverbrauch der Verbrauchsklasse i.

n₁:

Anzahl der Elemente der Verbrauchsklasse i.

In the case under consideration, the estimate of total consumption is as follows.

the following sizes were used:

V:

total consumption

N:

Number of consumption profiles

v _i :

measured average consumption of the consumption class i.

n ₁ :

Number of elements of the consumption class i.

Separately measured consumers become like a separate class with only one Element treated.

Here returns the difference between the extrapolated consumption and the measured one Inflow an estimator for the Background usage.

According to one Further advantageous embodiment, a division of consumers takes place of the area in consumer classes on the basis of the annual consumption, the type and consumption profile. A classification takes place over the Usage quantity known from the annual accounts of the kind the consumer, such as a detached house with a garden, a multiple dwelling, a small business, a large enterprise, etc. are and if necessary on the basis of the consumption profile of the consumer.

According to one Another advantageous embodiment, individual unclassifiable Consumer or bulk consumer measured directly. Individual consumers whose consumption is can not be properly classified, or bulk consumers become direct measured. This can for example hospitals or large companies be that have an internal water tank, the corresponding is filled irregularly.

According to one Another advantageous embodiment, the at least one measurement period one day the time from 2 to 4 o'clock, from 0-24 o'clock and / or from 6-18 o'clock be (exemplified periods of time). With the repetitive Measuring times can be over several Days or weeks away time series are formed statistically can be evaluated. This means, instead of an investigation at night, which is due to the better estimate of the consumption behavior, one can beside a 24 hours Consider also other time intervals, for example 8 hours, or several measurement periods while consider one day. In the case of the regression model, the Coefficients for each measurement period will be calculated separately, as is the composition Consumption from consumption classes changes over time.

A statistical method for the evaluation of time series, is the means Use of a threshold. According to an advantageous embodiment is a passing a difference value of the time series over a threshold Hint for a newly created leak or special effect in the area. If the difference value exceeds certain thresholds, it will become automatic generates an alarm. The thresholds can be set for each area either for all times fixed or dynamic become. For example, dynamic means that the threshold equals the mean of all previous ones Measurements and three standard deviations is.

According to a further advantageous embodiment, it is decided on the basis of all previously recorded difference values and a current difference value of the time series by means of a Sequential Propability Ratio Test (SPRT) whether there is a newly created leakage or a special effect in the area. Instead of carrying out a statistical test with a fixed sample size, the Sequential Probability Ratio Test tests after every observation made on the basis of all previously collected data, whether a decision for a null hypothesis or an alternative hypothesis can be made. If this is not the case, the observation is continued until this decision can be made. Since, in particular, water consumption is dependent on the day of the week, the measured values of the time series, in each case in addition to the weekdays, can be correspondingly divided into seven time series in each case. The average value τ _middle in the time series is determined. The sequential propability ratio test (SPRT) decides on the basis of the current measured values between the null hypothesis H ₀ ("OK case") and the alternative hypothesis H ₁ ("Leak case"). Mathematically, these hypotheses correspond to the following statements:
H _0: The values of τ (t) is normally distributed with the mean value and variance σ τ _medium. ²
H1: The values τ (t) are normally distributed with the average mean + ΔQ and variance σ ² .

ΔQ is the estimated size of a leak. The variance σ ² reflects the measurement inaccuracy in the flow measurement or in the total flow determination. The test is done separately for each of the time series. The number of past values should be limited. A network-configurable number should be provided.

According to one Another advantageous embodiment is one within a measuring cycle recorded larger increase the difference value of the time series as an average change of the difference value within the measuring cycle an indication of a leakage or a special event.

According to one Another advantageous embodiment is one within a day recorded larger increase the difference value of the time series as an average change of the difference value within one day an indication of a large leakage or a special effect. If the difference value increases, this is on sign for a leak. A sudden Rise provides a candidate for a leak. A sudden Rise is evidenced by the fact that the difference between two consecutive Difference values with the average change of the difference value is compared. This average is obtained by taking the slope a balancing line for the time series is considered by the previous difference values. Lies this comparison over a certain barrier, a leak alarm is triggered.

According to one Another advantageous embodiment is one within a week recorded larger increase the difference value of the time series as an average change of the difference value within a week an indication of a constructive leakage. If the average change in the difference value is above one positive barrier, an alarm is triggered that the difference value increases, which may indicate creeping leaks. A slow increase thus provides an indicator for creeping leaks.

According to one Another advantageous embodiment, instead of the measured Values of inflows and drains, moving averages used. In this way it is considered that the individual measured variables with Measuring errors are affected. The moving averages have one Smoothing property.

According to one Another advantageous embodiment, a further step the further location of a detected leak in the area. It takes place splitting the supply network into more as well as virtual ones Zones or virtual district meter areas (DMA) designated areas, their inflows and outflows Measured by flow meter, the areas overlap and adjacent areas each have at least one common flow meter exhibit. There is an evaluation of the time series for all areas and detecting leaks in the areas. Finally done delimiting the location of the leak based on the leakage information for the individual areas. Leakage information is information about that, whether leakage has been detected in the area or not. This is the basis for a sectional area algorithm for virtual Zones / DMAs that can be used here to address the local Limiting a leakage to allow.

According to one Further advantageous embodiment, the location of the leakage is in the intersection of areas, for a leak has been detected.

According to one Another advantageous embodiment, a special effect was detected if in several have no common flow meter Areas a leak has been detected.

According to one Another advantageous embodiment is when the leakage in only an area is detected, the location of the leakage thereby limited, that this is in the area in the area that no other Area covered becomes.

According to a further advantageous embodiment, an algorithm is claimed which allows a further location of a detected leak in the area:
There is an initial allocation of the number 1 to each section of a line of the supply network, wherein each line section is limited by a branch line or a flow meter. Each line section is between two Limits. A boundary of a line section may be a line branch, or a flow meter. For example, a conduit section may be bounded by a conduit branch at one end and a flowmeter at the other end. A line section can be between two branch lines or between two flow meters. There is a further step of selecting first areas for which no leakage has been detected and decreasing their numbers 1 to 0 for all line sections of these first areas. There is a sequential selection of all other areas for which a leak has been detected and a respective increase in the numbers each by 1, which are not equal to 0. After selecting all areas, the resulting number of a line section is a measure of the probability of the location of the leak there.

According to one Another advantageous embodiment is an event or special event or special effect detected when the pipe sections with the largest number in different areas.

According to one Another advantageous embodiment, the inventive method for an infrastructure network used to transport a fluid.

According to one Another advantageous embodiment, the infrastructure network is a Water supply, or a gas supply or a district heating network.

The The invention is applicable to all infrastructure networks in which fluids transported and consumed. Examples of such infrastructure networks are gas supply and district heating networks.

The The present invention will be described with reference to an embodiment closer with the figures described. Show it:

1 Exemplary embodiments of areas according to the invention;

2 the second embodiment of the invention according to areas 1 ;

3a -D an embodiment of an algorithm according to the invention;

4 an embodiment of a method according to the invention.

1 Figure 2 shows two embodiments of areas according to the invention called virtual DMAs or virtual zones. Virtual district meter areas (DMAs) differ from traditional DMAs as follows. Conventionally, when dividing zones into DMAs, it has always been attempted to form them in such a way that only one inflow pipe and / or one inflow pipe has been produced, and this can be monitored by a single sensor. According to the present application, this restriction is dropped. In the supply zones, additional flow sensors are installed at selected points in such a way that sections of the network can be measured whose inflows and outflows can be measured.

These sections allowed to not be disjunctive. This means, these cuts should have a common element. The cuts should to overlap and have common flow meters. Such cuts are the territories, virtual zones or virtual DMAs.

2 shows the second embodiment according to 1 , areas or virtual zones, or virtual DMAs labeled A, B, and C. These areas or sections are referred to below as virtual zones, or virtual DMAs. The number of inflows and outflows should not be too large, since sensors and data transmission technology are required per measuring point, whereby, despite greatly improved measuring technology, the measuring errors of the individual measuring instruments add up and thus the errors for determining the total fluxes increase.

With the approach of creating virtual zones / DMAs becomes presented a universal procedure to subdivide supply networks that Subareas, such as one or more line sections is / are monitored several times for leakage detection can be. The supervision Each virtual zone works on the same principle and can be in a technical solution be applied several times accordingly.

The Subdivision of a network into virtual zones / DMAs delivers the Advantage, except for the installation of flow meters, no change must be made on the existing network. Another advantage is that without the operation of the utility network too to disturb, or awkward, perform costly field measurements, leak detection can run automatically.

Furthermore, the use of the non-disjunctive, that is to say common flow meters, virtual zones / DMAs makes it possible to automatically limit the leakage. The described simultaneous analysis of multiple virtual zones / DMAs leads to greater reliability and a lower error rate because special effects or special events, which are also called events that result in a one-time high water consumption, can be automatically detected by the new method. If multiple tests indicate leakage during a statistical analysis of the data, a leak alarm will be triggered. It is now investigated in which virtual zones / DMAs the alarm is triggered. A first location is achieved by having the leakage in the intersection of the virtual zones / DMAs for which an error is generated. If the alarm is triggered simultaneously in several disjoint virtual zones / DMAs, it can be assumed that a special effect or a special event or special event, not multiple leaks, will result in this increased consumption. Nevertheless, an indication should be generated in an automated technical application to indicate this fact. Optionally, the values from different virtual zones / DMAs can also be used to form the correlation coefficients known from statistics. If such a significant correlation is found, this indicates a nationwide special event, such as an increased nighttime water consumption due to a sports broadcast on television.

Becomes on the other hand detects a leak in just one virtual zone / DMA, then the location can be limited by the location of the leak in this zone / DMA is exactly in the range, that of no other virtual DMA covered becomes.

An algorithm is proposed that works with the 3a to 3d is described in more detail. The mathematical algorithm is shown in steps. 3a to 3d are based on the classification into virtual zones / DMAs according to the 2 , In this case, the considered section of the network is divided by means of flow meters into three non-disjoint sections. On the right side in a virtual zone B, on the left in a virtual zone A and on the bottom half in a virtual zone C. It is assumed that an alarm was triggered for the virtual zones / DMAs B and C. Then the mathematical algorithm is applied as it is 3a to 3d represent. 3a shows all line sections, like this 2 represents. Each line section is now an edge section. An initialization is carried out by initializing each edge section with the value 1. This poses 3a Thereafter, evaluations are performed for all virtual zones / DMAs A, B and C, in the form of a loop over all virtual zones / DMAs. According to 3b the first selected zone is the virtual zone A. The score of the edge sections belonging to the virtual zone / DMA A is set to 0 because virtual zone A does not indicate a leakage error. This poses 3b Furthermore, the virtual zone B is selected. The score of the edge sections belonging to the virtual zone B, which are not equal to 0, is incremented by one. This poses 3c There are edge sections with the rating number 2 before. According to another step, the virtual zone C is selected. The score of the edge sections belonging to the virtual zone C and rated unequal to 0 is increased by 1. The poses 3d Now edge sections have been evaluated with the rating number 3. The edge sections with the rating number 3 are located in two virtual zones B and C, for which an alarm was triggered. The alarm is therefore likely to be triggered by events in these virtual zones B and C. The corresponding edge sections lie in a common area of the virtual zones B and C. However, other conspicuous edges were rated with the evaluation number 2. There is a high probability that the location with the leakage will be in edge sections with the rating number 3. However, since the edge segments evaluated with rating number 2 are equally relevant, a special event based on an event can not be excluded.

In summary The mathematical algorithm works according to the following steps. In an initialization step, each line section is the Number 1 assigned. After that it will loop over all virtual zones / DMAs executed. And although if no leak alarm for a virtual zone / DMA is present, the evaluation of all edge sections, those belonging to this virtual zone / DMA are set equal to 0. In the other Case, the evaluation of each edge portion leading to a heard virtual zone, increased by 1, unless he is 0. At the end of the procedure, each section has which is not in a virtual zone for which an alarm was triggered, the value 0. If the value is greater than 0, then the number is exactly one greater than the number of errors for the section was reported. The sections that have the largest rating numbers are the most likely for the Alarm responsible. Are they in different virtual zones, so the alarm may be just as by an event or a special effect or a special event triggered have been.

4 shows an embodiment of a method according to the invention for leakage detection in an area of a supply network with the steps S1 to S3. A step S1 is used to determine, by means of statistical extrapolation, an extrapolation value of a total consumption quantity of individual consumers of the area during at least one measurement period, in particular during one day. A step S2 is followed by determining an actual who the total consumption by measuring inflows into the area and outflows from the area during the measurement period. In a step S3, the extrapolation value is subtracted from the actual value, a positive difference value being an indication of an already existing leakage in the area.

bibliography

• [1] Forum Water Management 10.12.07
• [2] Voss, Werner: "Paperback Statistics "Leipzig (2004)

Claims (19)

Leak detection method in a field a supply network with the steps: - by means of statistical extrapolation determining an extrapolation value of a total consumption quantity of individual consumers of the area during at least one measurement period, by means of Division of consumers of the area into consumption classes; random selection of Consumers from every consumption class; Measuring an average consumption amount the selected one Consumers per consumption class for the duration of the measurement; statistical Extrapolate the average consumption quantities of all consumption classes using linear regression or as a stratified sample for Determining the extrapolation value; By measuring the inflows into the Area and drains out of the area during the measuring period, determining an actual value of the total consumption amount; - Subtract of the extrapolation value from the actual value, where a positive Difference value a hint for is already existing leakage in the area. Method according to claim 1, characterized in that that for each repetitive determination each time series of the difference values statistically a newly created leakage or a special event is detected in the area. Method according to claim 1, characterized by division the consumer of the area in consumer classes by means of a Annual consumption, one type and consumption profile. Method according to claim 1 or 3, characterized that individual non-classifiable consumers or bulk consumers be measured directly. Method according to one of claims 1 to 4, characterized that the at least one measuring time from 2.00 to 4.00 o'clock, from 0.00 until midnight and / or from 6:00 to 18:00. Method according to one of claims 2 to 5, characterized that is a passing a difference value of the time series over a threshold Note for a newly created leak or special event in the area is. Method according to one of claims 2 to 6, characterized that by means of all previously recorded difference values and a current Difference value of the time series based on a sequential probability Ratio Testes is decided whether a newly developed leakage or there is a special event in the area. Method according to one of claims 2 to 7, characterized that a larger increase in the difference value detected within a measuring cycle the time series as an average change of the difference value an indication of a leak or a special event within the measuring cycle is. Method according to claim 8, characterized in that that a larger increase in the difference value recorded within one day the time series as an average change of the difference value within one day an indication of a major leak or a special event is. Method according to claim 8, characterized in that that a larger increase in the difference value recorded within a week the time series as an average change of the difference value Within one week, an indication of an accumulating leakage is. Method according to one of the preceding claims, characterized characterized in that instead of the measured values of the inflows and Outflows, sliding Mean values are used. Method for further locating one of one of claims 1 to 11 detected leakage in the area, with the steps divide of the supply network in more as well as virtual zones or virtual district meter areas (DMA) designated areas whose inflows and drains Measured by flow meter, the areas overlap and adjacent areas each have at least one common flow meter exhibit; Evaluation of the time series for all areas and recognition leaks in the areas; Limiting the location of the leak based on the leakage information for the individual areas. A method according to claim 12, characterized in that the location of the leakage in the Intersection of the areas for which a leak has been detected. Method according to claim 12, characterized in that that a special event was detected, if in several no common Flowmeter having detected leaks. Method according to claim 12, characterized in that that if the leakage is detected in only one area, the location of the leakage is limited by the fact that this is in the Area in the area that is not covered by any other area becomes. Method according to one of claims 12 to 15, characterized by initial Assign the number 1 to each section of a line of the supply network, wherein each line section through a branch line or a flow meter is limited; Select first areas for which none Leakage was detected and decreasing their numbers from 1 to 0 for all line sections of these first areas; successively selecting all other areas, for a leakage has been detected and respective enlargement of the Numbers each 1, which are not equal to 0; being after Choose all areas the resulting number of a line section a measure of the probability the local presence of the location of the leakage is. Method according to claim 16, characterized in that that a special event is detected when the line sections with the biggest ones number in different areas. Use of a method according to one of claims 1 to 17, for an infrastructure network for transporting a fluid. Use according to claim 18, characterized that the infrastructure network is a water supply or a gas supply or a district heating network is.