DE102012025178A1 - Method for automatic characterization and monitoring of electrical system by comparing correlations between measured values, involves performing comparison through evaluation of mutual correlations of pairs of measurement variables - Google Patents

Abstract

The method involves performing comparison through evaluation of mutual correlations of pairs of measurement variables (1). Multiple currently measured pairs of measurement values of all evaluated pairs of measurement variables are checked to see whether they lie within the expected range limits (8'), which are obtained from multiple historical pairs of measurement values of all evaluated pairs of measurement variables. An independent claim is included for a device for automatic characterization and monitoring of an electric network or a power network section of an electric network or an electrical system by comparing correlations between measured values.

Description

The invention relates to a method and a device for automatically characterizing and monitoring an electrical network or a power network section of an electrical network or an electrical system by comparing relationships between measured values.

To a certain extent, this has already been practiced, for example, if the resistance, also referred to as impedance, of a subnetwork is determined from the ratio of voltage and current intensity. For example, in the DE1994869464 a method for generating at least one signal indicating a pendulum in an electric power grid described.

From the WO2005 / 088802A1 a device for the prevention of major disturbances in electrical supply networks by spectral analysis of the mains voltage is known. In this case, the spectrum determined by a voltage AD converter and a spectral analysis device is subjected to a classification according to frequency intervals. In a comparator, the values of the frequency interval classes are correlated with the frequency interval class reference patterns assigned to the specific disturbances, and an evaluation device outputs a signal, one or more relay messages or one or more messages via communication interfaces when a fault is detected. The frequency values and the number of frequency interval classes are adaptively adapted via a bifurcation analyzer to the local load and vibration dynamics of the network. The comparator is equipped with programmable limit values of the interval voltage for each frequency interval, whereby the limit values of the interval voltage automatically adapt to the load conditions of the network.

In the area of network analysis, the focus is on finding weak points in the supply of individual consumers, describing them and eliminating them with suitable measures. To recognize the causes, all data of the network parameters must be available for this purpose. In order to detect a malfunction due to events in the supply network, the Applicant has developed interference-resistant measuring systems which measure the network quality as well as the network and consumer behavior continuously and completely. The measured data are recorded in powerful databases. For example, is from the DE 10 2007 041 565 A1 a method and a measuring arrangement are known for transmitting time-stamped measurement data to a server if the measurement data is outside a standard or outside an individually specified tolerance range.

It is also known from the EP2355412A1 a method and an apparatus for evaluating an electrical system of an electrical energy system, wherein the electrical system comprises a communication network for transmitting data. The method comprises automatically acquiring real-time data transmissions in the communication network and automatically assessing the electrical equipment based on the acquired real-time data transmissions. For example, a scanning unit samples a reference signal comprising reference time information and transmits it as a real-time transmission over the communication network, wherein the electrical system is automatically evaluated based on a comparison of the reference time information with the time information of the real time transmission.

Electricity grids are dynamic structures in which it is normal for them to constantly change their structure by switching on and off operating equipment, and that the electrical load flows change in the short term and are also reversed by decentralized power generation. The current and power flows in given sections of the power grid can thus vary within wide limits, without violating technical and normative given limits. Behind these acute changes in network operating states, systematic shifts in dynamics often remain hidden.

The conventional measurement technology in the power grids is aimed in particular at ensuring the proper operation of the power grid and thus the network protection and tariffing. Power quality measurements, also referred to as power quality measurements, also measure whether, for example, the voltage curves present at the load comply with the normative specifications.

A disadvantage of all known measuring methods and evaluations, however, is that the evaluation time always lies in the past of a disturbing event, even if the measured data is transmitted in real time and the evaluation is immediate and almost real-time. Not every disturbing event can then be averted in time without any significant consequences for an electrical system or a consumer.

Another disadvantage of the known measuring methods and evaluations is the separate consideration of the measured values of individual measured variables.

The object of the present invention is therefore to provide a method and a device for the early detection of incidents in the electrical network and / or electrical installations, thus already in advance decisions on Prevention of incidents or required switching operations or other countermeasures can be performed automatically. The invention is also intended to help to optimize planning of current operations and investment decisions.

This object is achieved according to the present invention by a method according to claim 1, characterized in that not the current or historical measured values themselves, but the mutual relationships of their pairs of measured variables for characterizing an electrical network or power network section or an electrical system are used and by a device according to claim 7 for carrying out the method. The dependent claims define advantageous embodiments of the invention.

According to the method according to the invention, the comparison is carried out by the evaluation of mutual relationships of measured variable pairs by checking a multiplicity of currently measured measured value pairs of all measured variable pairs to see whether they are within expectation range limits previously determined from a plurality of historical values Measured value pairs of all investigated pairs of measured variables were obtained. For each pair of measured variables examined, one point cloud is formed from a plurality of historical measured value pairs. During a learning mode, an expectation range is then defined from the point cloud of many historical measured value pairs in each measured variable pair of a first matrix or a first half matrix, referred to below as an historical measured variable pair, and an expected range limit is stored. During a monitoring mode, currently measured pairs of measured values in measured-value pairs n of a second matrix or second half-matrix, referred to below as current measured-value pairs, are then defined with the expected ranges defined in the first matrix or first half-matrix and stored expected-range limits for each current measured value pair compared and evaluated. Those currently measured measured value pairs of the current measured variable pairs which are outside the expectation ranges and / or outside the expectation range limit are described in the measured value pairs of a third matrix or a third half matrix as abnormal measured value pairs. When an alarm is triggered by one or more abnormal pairs of measurements, a user-specified protection or switching action may be initiated to correct the problem before a negative event can occur.

The historically measured values stored in a data collection point can also be fed to the second matrix or second half-matrix according to the invention. Historical comparison measurements can then be carried out between the measured-value pairs of the second matrix or second half-matrix and those of the first matrix or first half-matrix, wherein the abnormal measured-value pairs of the historical comparison measurements match those in the first matrix or first half-matrix already formulated expected range limits of other historical pairs of measured variables are compared to carry out a trend analysis or to characterize the change in consumer behavior in the electrical network or for performance evaluation of an electrical system in continuous operation.

• • abgeleitete Messgrößen, wie Änderungsraten des Lastflusses und/oder
• • instationäre Messgrößen, für die über die Zeit nur eine Wahrscheinlichkeitsdichte angegeben werden kann und/oder
• • Messgrößen, die mittels Spektral- oder Autokorrelationsanalysen aus einer Basis-Messgröße zu berechnen sind,

zur Charakterisierung eines Netzabschnitts herangezogen werden. Dabei wird aus jeweils zwei unterschiedlichen Messgrößen jeweils ein Messgrößen-Paar erzeugt und die Messgrößen-Paare werden in Form eines Korrelations-Dreiecks dargestellt, wobei die aktuellen Messgrößen-Paare vorzugsweise die gleiche Anzahl an Messgrößen-Paaren wie die historischen Messgrößen-Paare aufweisen sollten.As measured variables, all typically used to describe the power quality measured variables such as current, voltage, power, harmonics of current and voltage, phase, interharmonic, flicker, etc., but also non-electrical factors such as day of the week, time of day, outside temperature, temperature of Transformers or precipitation and / or other measurands, such as

• • derived measures, such as rates of change of the load flow and / or
• • transient measurements for which only one probability density can be given over time and / or
• • measured quantities which are to be calculated from a basic measured variable by means of spectral or autocorrelation analyzes,

used to characterize a network section. In each case, a pair of measured variables is generated from two different measured variables in each case, and the measured variable pairs are represented in the form of a correlation triangle, wherein the current pairs of measured variables should preferably have the same number of measured variable pairs as the historical measured variable pairs.

• • einen ersten Datenspeicher zur Programmierung und Pflege einer ersten Matrix oder ersten Halb-Matrix aus historischen Messgrößen-Paaren zur Erfassung von Punktwolken aus historischen Messwert-Paaren und Speicherung der zugehörigen Erwartungsbereichsgrenzen,
• • einen zweiten Datenspeicher für eine zweite Matrix oder zweite Halb-Matrix aus aktuellen Messgrößen-Paaren zur Erfassung von aktuell gemessenen Messwert-Paaren,
• • ein Programmteil, der prüft, ob aktuell gemessene Messwerte der aktuellen Messgrößen-Paare innerhalb von Erwartungsbereichen und den definierten Erwartungsbereichsgrenzen der historischen Messgrößen-Paare liegen,
• • einen dritten Datenspeicher für eine dritte Matrix oder dritte Halb-Matrix zur Beschreibung von anormalen Messwert-Paaren sowie
• • eine Ausgabeeinrichtung.

The device according to the invention has at least

• A first data memory for programming and maintaining a first matrix or first half-matrix from historical measured-value pairs for recording point clouds from historical measured-value pairs and storing the associated expected-range limits,
• A second data memory for a second matrix or second half-matrix of current measured-value pairs for the acquisition of currently measured measured-value pairs,
• A program part which checks whether currently measured values of the current measured-value pairs lie within expectation ranges and the defined expected-range limits of the historical measured-value pairs,
• A third data memory for a third matrix or third half-matrix for describing abnormal measured value pairs as well as
• • an output device.

The device according to the invention optionally has a database server with a data bus and at least one data collection point for the acquisition and storage of historically measured values. A first data line connects the data collection point to the first data memory via a first connection and transmits the historical measured values from the data collection point via the first data line to the first matrix or a first half-matrix of the first data memory. The data bus has connections to the electrical network as well as equipment data sources, time data sources, calendar data sources, weather data sources and / or other data sources of systems or devices to be monitored or protected. In addition, the data bus is connected via a second data line and a second terminal to the second matrix or second half-matrix of the second data memory. The second data line transmits currently measured values of the data sources to the second matrix or second half-matrix of the second data memory and records them as currently measured value pairs for each current pair of measured quantities. The data collection point of the database server is preferably connected to a measuring device known per se for power quality measurement. Alternatively, the device according to the invention can be designed as an additional module for the measuring device for power quality measurement, or the device according to the invention is integrated directly into the measuring device for power quality measurement, the data collecting point permanently supplying the first data memory with historical measured value pairs during the current measuring operation and thus the expected range limits in a learning mode can update continuously.

According to the invention, therefore, not individual measured values of different measured variables are considered and compared with one another, but all possible pairs from the measured values of the different measured measured variables.

Instead of measured value pairs of the different pairs of measured variables, measured value n-tuples from measured variable n-tuples could also be considered according to the invention, but this is less practicable because the number of combinations in question increases extremely rapidly with the number of measured variables.

In the following the invention will be explained on the basis of exemplary embodiments and illustrated with reference to the attached drawings. Show it:

1 A correlation triangle according to the invention of a first half-matrix, shown here schematically in a possible first embodiment with expectation range limits defined in each measured variable pair, based here on 21 measured quantities, resulting in 20 measured values per column S and row R and 210 different pairs of measured variables.

2 A point cloud according to the invention belonging to a pair of measured variables, comprising a multiplicity of measured value pairs as a basis for calculating an expected range boundary, here represented as an ellipse.

3 A schematic diagram of the basic structure of the invention according to a further embodiment.

For the characterization of a network section, the measured values of all measured variables are suitable, which are typically used to describe the network quality:
Current, voltage, power, harmonics of current and voltage, phase angles, interharmonics, flicker etc.

• • Werte und/oder Messwerte von nicht-elektrischen Größen, wenn diese Einfluss auf die Auslastung des untersuchten Netzabschnittes haben. Beispiele für solche nicht-elektrischen Einflussgrößen sind Wochentag, Tageszeit, Außentemperatur, Temperatur von Trafos oder Niederschlag.
• • Abgeleitete Messgrößen Ein Beispiel wären Änderungsraten des Lastflusses, weil Leistungsänderungen die Entstehung von Flicker und von Interharmonischen bewirken können und weil die Antwort auf die Frage, ob und inwieweit dies der Fall ist, ein Charakteristikum eines elektrischen Netzes 36 bzw. Teilnetzes ist.
• • Messgrößen, die an sich instationär sind und für die daher über die Zeit nur eine Wahrscheinlichkeitsdichte angegeben werden kann. So sind in normalen Netzen Über- und Unterspannungs-Transienten, Brownouts und Blackouts seltene Ereignisse. Trotzdem ist es wichtig zu wissen, ob und wie diese Ereignisse mit anderen charakteristischen Netzgrößen korrelieren.
• • Messgrößen, die mittels Autokorrelationsanalysen aus einer Basis-Messgröße zu berechnen sind.

Also to be considered are:

• • Values and / or measured values of non-electrical quantities, if these have an influence on the utilization of the examined network section. Examples of such non-electrical influencing variables are day of the week, time of day, outside temperature, temperature of transformers or precipitation.
• • Derived metrics An example would be rates of change in load flow because power changes can cause the formation of flicker and interharmonics and because the answer to the question of whether and to what extent this is the case is a characteristic of an electrical network 36 or subnet is.
• • Measurands that are transient in nature and therefore can only be given a probability density over time. Thus, in normal networks, over and under voltage transients, brownouts and blackouts are rare events. Nevertheless, it is important to know if and how these events correlate with other characteristic network sizes.
• • Measured variables which are to be calculated by means of autocorrelation analyzes from a basic measured variable.

For some pairs of measured variables, it is advisable to use one or both of the measured variables as an argument only after filtering via a low-pass filter, a drag-pointer function or the like.

For example, an operating temperature of a transformer of 70 ° can be considered separately be completely uncritical if the transformer is designed for a maximum full load operating temperature of 90 °, while the same operating temperature of 70 ° is highly likely indicator of an error, if it is reached even at low utilization of the transformer. The combination of the measured variables operating temperature and load factor provides more information about the condition of the transformer than in the two individual measured quantities alone.

All measurands used for the characterization and / or monitoring of an electrical network 36 , a power line section of an electrical network 36 or an electrical system of interest are, according to the inventive method in a correlation triangle 7 Are defined. Such a correlation triangle according to the invention 7 is in 1 schematically illustrated in a possible first embodiment in the form of a first half-matrix 4 , In the embodiment according to 1 At least twenty-one metrics in a variety of dependencies on each other were numerically measured as pairs of measures 1 defined and detected in columns S1 to S20 and rows R1 to R20. Two different measured variables each form a pair of measured variables 1 , From historically measured values 18 ' , which were previously or in the past, among other things determined by means of power quality measurements, can now expect areas 14 with expectation range limits 8th' for each pair of measured variables to be observed 1 define. The expectation range limits 8th' are according to the first embodiment of 1 and 2 as ellipses 8th shown. However, it is within the scope of the invention that the expectation range limits 8th' may also look polygonal or hyperbolic, or have another closed curve.

Basically, between pairs of measured variables 1 and metric pairs 2 to distinguish. Many reading pairs 2 a pair of measured variables 1 each form a point cloud 3 within a measurand pair 1 , In 2 is a pair of measured variables 1 with such a point cloud 3 shown.

The point cloud according to the invention 3 in the measured quantity pair 1 consists of a large number of measured value pairs 2 within the defined expectation range limit 8th' that also in 2 as an ellipse 8th is shown. The expectation range limit 8th' but may also be rectangular or in another closed form, if it is known that one or more measured value pairs 2 while still in the margins, but still within an ellipse 8th or massive problems or other problems in the electrical network 36 or at an electrical system to be monitored. Such expectation ranges 14 must be outside the expected range limit 8th' be settled, an ellipse 8th is then no longer possible. For the sake of simplicity, however, the emergence of an expected range limit according to the invention 8th' based on an ellipse 8th explained.

As a description of the point clouds 3 for the different measured variable pairs 1 those mathematical methods are suitable, with which from the analysis of the point cloud 3 existing measured value pairs 2 a pair of measured variables 1 expectancy ranges 14 let determine that for new reading pairs 2 indicate with what probability they are in the calculated expected range 14 lie. These expectation areas 14 You can think of yourself as boundaries at different distances around the point cloud 3 the measured value pairs 2 in normal operation. Readings pairs 2 outside of this expectation range 14 are in monitoring mode as abnormal pairs of measured values 2 ''' and considered as potential indicators of critical or erroneous operating conditions.

• • die lineare Regression mit ihren Vertrauensbereichen oder
• • die Ellipse 8 um die Punktwolke 3 mit einer kurzen Achse 10 und einer langen Achse 9, die den minimalen mittleren Abstand zu den Messwert-Paaren 2 der Punktwolke 3 hat, wie in 2 dargestellt.

Examples of calculation methods that have such expectation ranges 14 can deliver

• • the linear regression with its confidence intervals or
• • the ellipse 8th around the point cloud 3 with a short axis 10 and a long axis 9 that the minimum mean distance to the measured value pairs 2 the point cloud 3 has, as in 2 shown.

• • haben keine Bedeutung,
• • sind trivial,
• • sind redundant oder
• • enthalten zu wenige Daten für eine signifikante Aussage,

können somit vernachlässigt werden.Many of the relationships that are captured

• • have no meaning
• • are trivial,
• • are redundant or
• • contain too little data for a significant statement,

can thus be neglected.

According to the embodiment according to the invention 2 is the point cloud 3 through an ellipse 8th described, whose long axis 9 the smallest mean distance to the measured value pairs 2 has that in 2 are shown punctiform. The long axis 9 and the short axis 10 the ellipse 8th intersect at the center of gravity of the point cloud 3 as ellipse center 11 , Four bold points 12 on the axes 9 . 10 give the mean distance of the measured value pairs 2 to each perpendicular to axes. The length of these half-axes is determined by the distances of the points 12 to the ellipse center 11 plus 2.5 times σ, the scattering of mean distances. The in 2 arbitrarily chosen factor 2.5 is a tolerance parameter. Other values of the tolerance parameter would result in larger or smaller Lengths of elliptical axes 9 . 10 and thus different sized ellipses 8th but not smaller than a cross 13 within the points 12 can be.

Each field of a half-matrix 4 thus contains descriptive information about the point cloud 3 every measurand pair 1 ,

• • einen ersten Datenspeicher 16 zur Programmierung und Pflege der ersten Halb-Matrix 4 zur Berechnung und Erfassung der Erwartungsbereichsgrenzen 8' in historischen Messgrößen-Paaren 1'' auf der Basis von Referenzdaten historischer Messwert-Paare 2''
• • einen ersten Anschluss 17 für eine erste Datenleitung 18, die historische Messwerte von einer Datensammelstelle 19 eines Datenbankservers 31 über die erste Datenleitung 18 an den ersten Datenspeicher 16 überträgt,
• • einen zweiten Datenspeicher 20 für die zweite Halb-Matrix 5 zur Erfassung und Zwischenspeicherung von aktuell gemessenen Messwert-Paaren 2' innerhalb entsprechend definierter aktueller Messgrößen-Paare 1',
• • einen zweiten Anschluss 21 für eine zweite Datenleitung 22, die aktuelle Messwerte von Datenquellen an den zweiten Datenspeicher 20 überträgt,
• • ein Programmteil 23 für die Durchführung des Vergleichs, auch als Vergleicher bezeichnet, der alle aktuell gemessenen Messwert-Paare 2' der aktuellen Messgrößen-Paare 1' daraufhin überprüft, ob sie innerhalb der für die historischen Messgrößen-Paaren 1'' in der ersten Halb-Matrix 4 hinterlegten Erwartungsbereichsgrenzen 8' liegen,
• • einen dritten Datenspeicher 24 für eine dritte Halb-Matrix 6 zur Erfassung und Bewertung von anormalen Messwerten 2'''. Hier wird abgelegt, für welche Messgrößen-Paare 1 die aktuell gemessenen Messwert-Paare 2' außerhalb der für die historischen Messgrößen-Paare 1'' festgelegten Erwartungsbereichsgrenzen 8' liegen sowie
• • eine Ausgabeeinrichtung 25, die ein zuvor definiertes Signal abgibt und/oder versendet und die auch als Steuergerät arbeiten kann.

The basic structure of the invention and the device according to the invention 15 is in 3 shown. The device 15 has in this embodiment:

• • a first data store 16 for programming and maintaining the first half-matrix 4 for calculating and recording the expected range limits 8th' in historical quantities pairs 1'' on the basis of reference data of historical measured value pairs 2 ''
• • a first connection 17 for a first data line 18 , the historical readings from a data collection point 19 a database server 31 over the first data line 18 to the first data store 16 transmits,
• • a second data store 20 for the second half matrix 5 for the acquisition and temporary storage of currently measured value pairs 2 ' within correspondingly defined current measured value pairs 1' .
• • a second connection 21 for a second data line 22 , the current readings from data sources to the second data store 20 transmits,
• • a program part 23 for the implementation of the comparison, also referred to as comparator, all currently measured pairs measured 2 ' the current measured variable pairs 1' then check to see if they're within the range for the historical metric pairs 1'' in the first half matrix 4 stored expectation range limits 8th' lie,
• • a third data store 24 for a third half-matrix 6 to record and evaluate abnormal readings 2 ''' , Here is filed, for which pairs of measured variables 1 the currently measured pairs of measurements 2 ' outside the for the historical metric pairs 1'' defined expectation range limits 8th' lie as well
• • an output device 25 which emits and / or transmits a previously defined signal and which can also operate as a control unit.

The second data line 22 is on a data bus 26 tethered. In addition, the data bus points 26 in this embodiment, connections to a time data source 27 , a calendar data source 28 , a weather data source 29 , a resource data source 30 and possibly to other data sources of installations or equipment to be monitored or protected.

The device according to the invention 15 to 3 has a third data line in this embodiment 35 on. This allows those in the data collection point 19 stored, historically measured values 18 ' the second half-matrix 5 be fed and that between the pairs of measured variables 1 the second half-matrix 5 and those of the first half-matrix 4 historical comparison measurements are performed, with the abnormal measured value pairs 2 ''' the historical comparison measurements with those in the first half-matrix 4 already formulated expectation range limits 8th' that come from other historical metric pairs 2 '' are compared to carry out a trend analysis or to characterize the change in consumer behavior in the electrical network or for performance evaluation of an electrical system in continuous operation.

Basically, the current measured value pairs 1' the same number of measured variable pairs 1 on like the historical metric pairs 1'' , There are differences if, for example, a data source is switched on or off during ongoing measuring operation. The comparison of current measured value pairs 1 with historical measured value pairs 1'' It is only possible to do so if there are already expectation limits 8th' from historical metric pairs 2 '' are defined. However, since all currently measured values 22 ' via the data busbar 26 both the second data store 20 with the second half-matrix 5 as well as the data collection point 19 can be fed via a later historical comparison measurement in learning mode 32 z. B. the missing expectation areas 14 and expectation range limits 8th' in historical metric pairs 1'' newly defined and to the current measured value pairs 1' be adjusted.

It is particularly advantageous if the data collection point 19 with a measuring device, for example with a measuring device known per se for power quality measurement 34 connected is. All in the meter for power quality measurement 34 currently measured values for an electrical network or a power supply section or a resource to be monitored 30 , z. As transformers in a substation, then both the data collection point 19 as well as the second data store 20 fed and in the device 15 be processed directly by the novel process.

It is within the scope of the invention that the data collection point 19 directly in the meter for power quality measurement 34 is arranged. Also, the device can 15 as additional module for a measuring device for power quality measurement 34 or another suitable measuring device. But according to the invention can also the entire device 15 into the meter for power quality measurement 34 be integrated. This is particularly advantageous if it is provided that the data collection point 19 the meter for power quality measurement 34 the first data store 16 permanently supplied with measured values in the course of measuring operation and thus the learning capability of the device 15 guaranteed. Measuring devices for power quality measurement 34 , as they are already used today for the power quality measurement at the entrance or at the entrances to a network section under investigation, provide a stream of data of different parameters to the network state. Same data sources or calendar data sources 28 and time data sources 27 provide time information, weather data sources 29 provide weather information. Optionally, further data sources provide further data on the state of resources, such. B. the transformer internal temperature of the resource data source 30 , The at the data bus 26 incoming data go to the data collection point 19 , Their content becomes areas of expectation 14 determined and in the first half-matrix 4 Are defined. The over the second data line 22 and the second port 21 in the second data store 20 incoming data will be measured as currently measured pairs 2 ' captured and as current pairs of measurements 1' in the second half-matrix 5 processed. In the program part 23 then becomes a comparison of the current measured value pairs 1 with the historical measured value pairs 1'' performed and those measured pairs 2 in which the current measurement pairs 1' outside the expected ranges 14 lie, as abnormal measured value pairs 2 ''' in the third half matrix 6 displayed. In the case of the abnormal measured value pairs 2 ''' this can be live data or real-time data or even test data.

• • Zu Beginn des Lernmodus selektiert der Anwender diejenigen Messgrößen-Paare 1, welche in die Analyse eingehen sollen.
• • Dann bestimmt der Anwender den Zeitraum, dessen Daten als Referenzdaten dienen sollen.
• • Optional bestimmt der Anwender noch, ob alle verfügbaren Messwerte oder ob beispielsweise nur zeitlich gerasterte Stichproben der Daten als Referenzdaten dienen sollen.
• • Optional können auch Daten aus einem Simulationsprogramm, welches das Verhalten des zu überprüfenden Netzes synthetisiert, als Referenzdaten dienen.
• • Dann werden die Zusammenhänge der vom Anwender vorher selektierten Messgrößen-Paare 1 bestimmt.

The method and the device according to the invention 15 work in at least two modes: a learning mode 32 and a monitoring mode 33 , The learning mode 32 works with reference data.

• • At the beginning of the learning mode, the user selects the pairs of measured variables 1 which should be included in the analysis.
• • Then the user determines the period of time whose data should serve as reference data.
• • Optionally, the user also determines whether all available measured values or, for example, only time-sampled samples of the data should serve as reference data.
• • Optionally, data from a simulation program that synthesizes the behavior of the network to be checked can also serve as reference data.
• • Then the relationships of the pairs of measured variables previously selected by the user become 1 certainly.

As a result of the latter operation, each position in the first half-matrix becomes 4 the description of a historical pair of measured variables 1'' z. B. a correlation analysis, an ellipse 8th or the like and thus the definition of the expectation ranges 14 for data in monitoring mode 33 assigned. The calculations can be made from those in the data collection point 19 archived data of historically measured values 18 ' respectively.

It is important that the reference data comes from a sufficiently long and typical period, which covers all operating conditions relevant in normal operation. With the aid of a tolerance parameter, the user can define the width of the actual expected ranges of expectation 14 individually or together for all areas of expectation 14 pretend.

• • berechnet das System in regelmäßigen Abständen die aktuell gemessenen Messwert-Paare 2' und
• • prüft das System, ob die aktuell gemessenen Messwert-Paare 2' innerhalb der Erwartungsbereiche 14 der im Lernmodus 32 selektierten historischen Messgrößen-Paare 1'' liegen.
• • Liegt ein aktuell gemessenes Messwert-Paar 2' außerhalb des Erwartungsbereichs 14 des entsprechenden historischen Messgrößen-Paars 1'', wird nach dem Vergleich im Programmteil (23) durch anormale Messwert-Paare 2''' ein Alarm ausgelöst und in der dritten Halb-Matrix 6 angezeigt.
• • Jeder Alarm enthält die Information, welches Messgrößen-Paar 1 mit welchem anormalen Messwert-Paar 2''' ihn ausgelöst hat. Diese Information kann dem Anwender auf Wunsch im Klartext angezeigt werden.

In monitoring mode 33

• • the system calculates the currently measured measured value pairs at regular intervals 2 ' and
• • the system checks whether the currently measured measured value pairs 2 ' within the expected ranges 14 the one in learning mode 32 selected historical variables pairs 1'' lie.
• • Is a currently measured pair of measured values 2 ' outside the expected range 14 of the corresponding historical parameter pair 1'' , after the comparison in the program part ( 23 ) by abnormal measured value pairs 2 ''' an alarm is triggered and in the third half-matrix 6 displayed.
• • Each alarm contains the information, which pair of measured variables 1 with which abnormal pair of measured values 2 ''' has triggered him. This information can be displayed to the user on request in plain text.

The alarm messages, which contain all the information of the underlying measured variable pairs 1 can also be used for a controlled extension of the expected ranges 14 be used. If archived raw data of the sensors are available to the system according to the invention, the suitability of the expected ranges can be determined 14 for detecting known changes in the electrical network 36 or an electrical system.

• • Anzahl und Auswahl der Messgrößen-Paare 1, welche in die Analyse eingehen sollen,
• • Die Auswahl der Referenzdaten,
• • Die Auswahl des Zeitrasters der Referenzdaten sowie der Stichproben im Überwachungsmodus 33
• • Die Wahl der Toleranz-Parameter für die Messgrößen-Paare 1
• • Ein automatisierbarer Austausch der Referenzdaten, weil zum Beispiel bekannt ist, dass das untersuchte Netzsegment sich an Werk- und Feiertagen sehr verschieden verhält.

In the configuration of the method and the device according to the invention 15 the following degrees of freedom exist:

• • Number and selection of the measured variable pairs 1 which should be included in the analysis,
• • the selection of reference data,
• • The selection of the time frame of the reference data as well as the samples in monitoring mode 33
• • The choice of tolerance parameters for the measured quantity pairs 1
• • An automatable exchange of the reference data, because it is known, for example, that the examined network segment behaves very differently on workdays and public holidays.

Advantages of the invention over the usual network analysis systems are that the inventive method is "self-learning", ie it "learns" the limits of "normal operation" of the network section independently and automatically. In addition, the inventive method works independently of limits, standards and physical limitations. However, existing limitations can be within the bounds of expectation 14 be incorporated and act as alarm triggering thresholds.

Also, the inventive method "lazy user-suitable". This means that the manufacturer can have a "over-cautious" configuration with an excessive number of pairs of measurements 1 and deliver too tight tolerance parameters. Then the user gets too many and too narrow expectation ranges 14 generated. In monitoring mode, this leads to an excessive number of alarms. During a preliminary phase of the monitoring operation, these alarms can then be used to extend the tolerance parameters to the expected ranges 14 gradually adapt automatically to the normal operation of the target network segment. Also, the new reading pairs can 2 in order to keep the expected range limits in such operation 8th' to recalculate at constant tolerance parameters to adapt the system to the existing measurement situations.

Another advantage is that configurations can be pre-tested. For this purpose, the method according to the invention can be tested, for example, offline with archive data from times before and after a known change in the network segment to be examined. First of all, expectation ranges from reference data of the time before the change in the network segment are expected 14 and then tested with data from the time after the change in the network segment to see if the change with the selected configuration would have been detected as desired. The user can - provided appropriate software tools - the expected ranges 14 all measured variable pairs 1 query at any time, represent (comparable to 2 ) and for compatibility with technical and economic requirements.

A further advantage of the invention is that systematic shifts in the dynamics of network operating states can be detected in good time. Such changes arise, for example, from creeping degradation of equipment, from changes in the type and number of connected consumers and generators, etc .. These hidden trends in systematic shifts in dynamics are of particular interest to network operators to the planning of ongoing operations and investment decisions optimize.

The method according to the invention obtains despite the diversity of the pairs of measured variables used 1 the traceability of all alarms up to the raw data. This distinguishes the invention of network monitoring solutions with neural networks. The number of measured variable pairs 1 and the associated expectation ranges 14 increases with the square of the number of measured variables: N _{expectancy ranges} - ≈0,5 · N _metrics

If the amplitudes and phase angles of the first 50 harmonics are included as measured variables in a 3-phase network, a total of about 1000 measured variables, ie 500,000 pairs of measured variables, can be achieved 1 of which a large part is superfluous.

In summary, it can be stated that from the historical measured value pairs 2 '' and the currently measured measured value pairs 2 ' Characteristics can be extracted automatically, with which a characterization of the electrical network 36 or the examined network section is possible and with which after a learning mode 32 Automatically detect significant changes and trends in the characteristics of these network sections online or offline.

With the invention it is achieved that the normal operating states of a given electrical network 36 or a power supply section can be automatically characterized and acute and creeping deviations from a previously learned characteristic can be detected and averted in advance of an event.

LIST OF REFERENCE NUMBERS

1

Metrics pair

1'

current pair of measured variables

1''

historical measure pair

2

Reading pair

2 '

currently measured value pair

2 ''

historical reading pair

2 ''

abnormal reading pair

3

point cloud

4

first half-matrix

5

second half-matrix

6

third half-matrix

7

Correlation triangle

8th

ellipse

8th

Expected range limit

9

long axis of the ellipse 8th

10

short axis of the ellipse 8th

11

Ellipse center

12

Points

13

cross

14

Expected range

15

contraption

16

first data store

17

first connection

18

first data line

18

historically measured values

19

data collection

20

second data store

21

second connection

22

second data line

22 '

currently measured values

23

program part

24

third data memory

25

output device

26

Data bus

27

Time data sources

28

Calendar data sources

29

Weather data sources

30

Resource data sources

31

Database server

32

learning mode

33

monitoring mode

34

Meter for power quality measurement

35

third data line

36

Electrical network

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• DE 1994869464 [0002]
• WO 2005/088802 A1 [0003]
• DE 102007041565 A1 [0004]
• EP 2355412 A1 [0005]

Claims (10)

Method for automatically characterizing and monitoring an electrical network or a power supply section of an electrical network or an electrical system by comparing relationships between measured values, characterized in that the comparison is carried out by the evaluation of mutual relationships of measured variable pairs ( 1 ) he follows. Method according to claim 1, characterized in that a plurality of - currently measured pairs of measured values ( 2 ' ) of all measured variable pairs ( 1 ) are checked to see if they are within expectation 8th' ), which previously consisted of a large number of historical measured value pairs ( 2 '' ) of all measured variable pairs ( 1 ) were won. Method according to claim 1 or 2, characterized in that - for each pair of measured variables examined ( 1 ) from a plurality of measured value pairs ( 2 ) one point cloud each ( 3 ) and - that during a learning mode ( 32 ) from the point cloud ( 3 ) of many historical measured value pairs ( 2 '' ) in each historical measure pair ( 1'' ) a first matrix or a first half-matrix ( 4 ) each one expectation area ( 14 ) and an expectation range limit ( 8th' ) and that during a monitoring mode ( 33 ) in a second matrix or second half matrix ( 5 ) currently measured pairs ( 2 ' ) with those in the first matrix or a first half-matrix ( 4 ) defined areas of expectation ( 14 ) and stored expectation range limits ( 8th' ) for each current pair of measured variables ( 1' ) are compared and evaluated. Method according to one of the preceding claims, characterized in that the currently measured measured value pairs ( 2 ' ) of the measured variable pairs ( 1 ) outside the expected ranges ( 14 ) and / or outside the expectation range limit ( 8th' ), in a third matrix or a third half-matrix ( 6 ) as abnormal measured value pairs ( 2 ''' ) and that when an alarm is triggered by one or more abnormal measured value pairs ( 2 ''' ) a user-defined protection or switching operation can be triggered to correct the problem before a negative event can occur. Method according to one of the preceding claims, characterized in that in a data collection point ( 19 stored, historically measured values ( 18 ' ) of the second matrix or second half-matrix ( 5 ) and that between the pairs of measured variables ( 1 ) of the second matrix or second half-matrix ( 5 ) and those of the first matrix or first half-matrix ( 4 ), the abnormal measured value pairs ( 2 ''' ) of the historical comparison measurements with those in the first matrix or first half matrix ( 4 ) already on the basis of other historical measured value pairs ( 2 '' ) formulated expectation range limits ( 8th' ) are compared to carry out a trend analysis or to characterize the change in consumer behavior in the electrical network or to evaluate the performance of an electrical system in continuous operation. Method according to one of the preceding claims, characterized in that the measured variables used are all parameters typically used to describe the network quality, such as current, voltage, power, harmonics of current and voltage, phase angles, interharmonics, flicker etc., but also non-electrical parameters, such as day of the week, time of day, outside temperature, temperature of transformers or precipitation and / or other measured variables, such as • derived measured variables, such as rates of change of the load flow and / or • transient measured variables for which only a probability density can be specified over time and / or • Measured variables which can be calculated by means of spectral or autocorrelation analyzes from a basic measured variable, used to characterize a network section, and that in each case a pair of measured variables from two different measured variables each ( 1 ) and the measured variable pairs ( 1 ) in the form of a correlation triangle ( 7 ), where the current measured variable pairs ( 1' ) preferably the same number of pairs of measured variables ( 1 ) like the historical measures pairs ( 1'' ) should have. Device for automatically characterizing and monitoring an electrical network or a power supply section of an electrical network or an electrical system by comparing relationships between measured values, characterized in that the device ( 15 ) at least a) a first data memory ( 16 ) for programming and maintaining a first matrix or first half-matrix ( 4 ) from historical measured value pairs ( 1'' ) for capturing point clouds ( 3 ) from historical measured value pairs ( 2 '' ) and storage of the associated expectation range limits ( 8th' ), b) a second data memory ( 20 ) for a second matrix or second half-matrix ( 5 ) from current measured value pairs ( 1' ) for the acquisition of currently measured value pairs ( 2 ' ) c) a program part ( 23 ), which checks whether currently measured values ( 22 ' ) of the current measured variable pairs ( 1' ) within expected ranges ( 14 ) and the defined expectation range limits ( 8th' ) of historical measured value pairs ( 1'' ), d) a third data memory ( 24 ) for a third matrix or third half-matrix ( 6 ) for the description of abnormal measured value pairs ( 2 ''' ) and e) an output device ( 25 ). Device according to claim 7, characterized in that the device ( 15 ) a database server ( 31 ) with a data bus ( 26 ) and at least one data collection point ( 19 ) for the acquisition and storage of historically measured values ( 18 ' ) and that a first data line ( 18 ) the data collection point ( 19 ) via a first connection ( 17 ) with the first data memory ( 16 ) and the historical measured values ( 18 ' ) from the data collection point ( 19 ) over the first data line ( 18 ) to the first matrix or a first half-matrix ( 4 ) of the first data memory ( 16 ) and that the data busbar ( 26 ) Connections to the electrical network ( 36 ) and to a) resource data sources ( 30 ) and / or b) time data sources ( 27 ) and / or c) calendar data sources ( 28 ) and / or d) weather data sources ( 29 ) and / or e) other data sources of equipment or devices to be monitored or protected, and that the data bus ( 26 ) via a second data line ( 22 ) and a second port ( 21 ) with the second matrix or second half-matrix ( 5 ) of the second data memory ( 20 ) and that the second data line ( 22 ) currently measured values ( 22 ' ) of the data sources to the second matrix or second half-matrix ( 5 ) of the second data memory ( 20 ) transmits and as currently measured measured value pairs ( 2 ' ) for each current pair of measured variables ( 1' ) and that between the first data memory ( 16 ), the second data store ( 20 ) and the third data memory ( 24 ) program part ( 23 ) is a comparator. Apparatus according to claim 7 and / or 8, characterized in that the data collection point ( 19 ) of the database server ( 31 ) with a per se known measuring device for power quality measurement ( 34 ) or that the device ( 15 ) as an additional module for the measuring device for power quality measurement ( 34 ) or that the device ( 15 ) into the power quality meter ( 34 ) and that the data collection point ( 19 ) the first data memory ( 16 ) in ongoing measuring operation permanently with historical measured value pairs ( 2 '' ) and in a learning mode ( 32 ) the expectation range limits ( 8th' ) continuously updated. Device according to one of claims 7 to 9, characterized in that a method according to one of claims 1 to 6 is used.

Images