AT516425A2 - Method for storing and selecting data - Google Patents

Abstract

The invention relates to a method for storing and selecting data (5) from a data memory, in which data (5) from a plurality of data sources in a decentralized system architecture, for example in a low voltage network, which at least in a data set (1) according to time sequence (3) are stored and queried by means of a user request. In order to ensure fast and reliable database access, the records (1) are stored in rows (2, 9, 10, 11, 12), which are each associated with a data source, according to the time sequence (3) and successive line segments (2a, 2b , 2c, 9a, 9b, 9c, 10a, 11a, 11b, 12b) covering predetermined periods of time are combined into data blocks (6, 7, 8), the method of selecting data (5) in a user request comprising subsequent steps : Selection of those data blocks (7, 8) covering the searched area (4); Analysis of data (5) in the line segments (2b, 2c, 9b, 9c, 11b, 12b) in the data blocks (7, 8); Selection of those line segments (2b, 2c, 9b, 9c, 11b, 12b) containing data (5).

Description

Method for storing and selecting data

FIELD OF THE INVENTION

The present invention relates to a method for storing and selecting data from a data memory, in which data from multiple data sources in a decentralized system architecture, for example in a low-voltage network, which are stored at least in a record according to a time sequence and by means of a user request be queried.

Such data sources, which may also be referred to as network nodes or data points, may be: Generators (eg photovoltaic systems, small wind turbines) or their actuators (eg converters), flexible consumers (eg electric storage heaters, buildings with and without building automation system) and corresponding actuators, transformers , Measuring devices such as smart meters and in particular so-called intelligent electronic devices (intelligent electronic devices), such as processor-controlled regulators of circuit breakers or transformers.

Data within the meaning of the invention can be: measured values for electrical current, measured values for electrical voltage, measured values for electrical power, time of the measurement, as well as metadata relating to the transmission, e.g. Number of nodes passed, latency (delay time).

STATE OF THE ART

Local intelligent control schemes distributed over a low-voltage grid are gaining more and more importance in the architecture of intelligent networks. A basic requirement for automatic monitoring and control is a solid base of measuring equipment, for example in the form of "smart meters". (so-called smart meters), intelligent inverters or intelligent building services. While immediate values or information are crucial for short-term control measures, a stable regime would also consider historical values and incorporate them into medium- or long-term planning. Historical data can be used in a central database not only for current and voltage regulation, but also for short-term correction in case of data loss, and facilitate various planning.

Such a central database is created by means of an analysis method and a data management system, such approach having a decentralized system architecture, such as a low voltage network, several disadvantages: The data from a low level of the system architecture (eg from a smart electronic System (IED)) must be transported with the highest possible data rate to the top level of the system architecture (eg the central database), where

High-speed connections, for example in the form of broadband connections, must be used. However, these high-speed connections can only be used with a small number of node connections, such as those found in intelligent home automation. With a large number of node connections, a high latency and a low bandwidth occur, whereby a low power in the data transmission via the power network, the so-called "Distribution Line Communication". (DLC), the result is. - The data center or the central control and control logic is a danger point for potential errors, with the default probability and the associated costs represent a significant point. Data centers are usually redundant several times, so that the probability of failure is very low, but not negligible. However, the cost of a total outage is high, which is why a decentralized and cost-effective backup solution should be used. - The actuators in low-voltage networks are usually arranged in the network itself, that is, either on the secondary side, for example in the form of a transformer, or in the network itself, e.g. as a programmable inverter, distributed. This means that in a central system, decisions must be made regarding a network level at the automation level. This results in an error point in the data connection between these layers, even if the data center itself is multiply redundant. - The amount of data of a central entity, which regulates a plurality of low-voltage network nodes, is enormous compared to medium or high voltage networks. It is expected that in normal operating condition suitable

Algorithms can control such a flood of data, with unpredictable conditions or events should be handled by a human for legal or security reasons.

Due to these problems, it is recommended to install at least one basic backup solution for local remote operation (in the respective low-voltage grid). Or to provide the local operators with a local database based on historical measurement data so that such a problem can be inexpensively overcome by the operator.

There is currently no generic commercially available solution for controlling and monitoring local low voltage networks. An uncomplicated technical solution would be to implement either a volatile memory or a database, both of which have disadvantages.

The use of a volatile memory, such as a RAM memory, to secure measurement data in a device has the advantage that very fast access times compared to a permanent memory are possible, but the risk of data loss is enormous. As is well known, a planned or unplanned restart of the device can result in complete data loss, for example, a short-term power loss. In addition, the maximum possible data volume of a volatile memory is usually much lower than that of a so-called read-only memory (ROM, ie hard drives, flash drives, EEPROMs, etc.), which due to the high cost of these preferably in embedded or Compact devices are applied.

A database typically avoids the problem of data inconsistency and can create a variety of request mechanisms, but databases pose a risk to many IT applications due to technical requirements such as transmission security. In this regard, high performance databases are a known solution, but at a high cost.

Although databases are typically two-dimensional, usually only the vertical dimension (columns) is dynamic, while the horizontal dimension (rows) is static. Of course, the rows and columns may also be constructed in a reverse dynamic and static manner. In general, however, this division into a dynamic and a static area presents a significant space problem since databases are best suited for storing dense data and a two-dimensional table is usually too small for large amounts of data.

PRESENTATION OF THE INVENTION

It is therefore an object of the invention to provide a method by which specific data entries can be read out in a simple manner and with low access and transmission times. In other words, it is an object of the invention to provide a database construction and a data storage architecture and an associated method, whereby a reliable rapid data access is ensured. This in particular for data from low-voltage networks, which arrive from different Datenguellen with different time intervals in the data memory.

This object is achieved by a method according to claim 1, by storing the data records in rows, which are each assigned to a data source, according to the time sequence and successive

Line segments that cover predetermined periods of time are combined into data blocks, wherein the method for selecting data in a user request comprises the following steps: selecting those data blocks that cover the searched region; - analysis of data lying in the row sections in the data blocks; - Selection of those line segments containing data.

An inventive data set thus consists of several lines and contains for each data source at least one line. In one line, all data are entered after the time of their measurement, whereby the position of the respective data in the line corresponds to a certain point in time. Thus, for example, the line could be subdivided into many time units, in particular the same length, the data being allocated to a specific time unit. Time units of a line where there is no data of the corresponding data source remain free.

The lines are now subdivided into line sections that correspond to a given period of time (that is, contain a certain number of time units), which is the same for all lines. All line segments of a given time period yield a block of data, also called "time strip". can be designated. A data block thus contains the data of all data sources in a given period of time.

If a user request is now made to read the data for a certain period of time for certain data sources, first those data blocks ("time strips") are selected which cover the desired period of time. It then analyzes which lines of data there are, and then selects those lines.

The data structure thus consists of rows and columns, the columns correspond to specific time periods, while the rows correspond to different data sources, such as network nodes. The length of the rows will increase over time and is not limited, the number of rows and thus the length of the columns are fixed.

In other words, the user queries a specific area within the time sequence, whereby those line sections are detected which lie within the searched area. Subsequently, the detected line segments are analyzed in more detail, whereby it is determined which line segments actually contain data. Thereafter, the data information of the data-containing line segments are provided to the user.

This row and column structure is present both in the lower level of the data memory or in a RAM where the most recent data is stored, as well as in the persistent memory or upper level of the data memory.

Moreover, the present application discloses a two-stage data memory for a decentralized system architecture, for example in the form of a low-voltage network, for selecting data arranged at least in a data record. In detail, the data sets in rows, which are each assigned to a data source, stored in accordance with the time sequence and successive line segments that cover predetermined periods summarized into data blocks, the data structure is at least two stages formed by a first stage as a volatile memory younger Contains data, while a second level as persistent storage contains older data than the first level.

In other words, this is a two-tiered data storage architecture, where the first (lower) tier provides general mechanisms for storing arbitrary data and unifies the benefits of an internal memory ("cache") and a persistent database solution. The data storage architecture takes into account any division of physical data storage between different types of storage. These are: volatile memory (eg RAM memory etc.),

Permanent memory (eg, ROM, HDD, flash memory, etc.) and network nodes with local memories, which are, for example, local computer architectures with memory drives.

The second layer of the storage architecture is a project-specific data specialization that allows retrieval of data in the database.

The combination of the method according to the invention with the two-stage data memory affords the advantage that the otherwise usually high storage duration and the processing time are carried out much more efficiently by the system architecture according to the invention and the associated method, so that efficient data storage is ensured. Depending on the particular application, a kind of permanent data storage is used, as otherwise historical values would be lost and only the latest values could be delivered, which is inadequate for a modern, service-controlled voltage network with stable regulation.

It is possible that the sequence for storing in constant steps in the form of a timeline is performed. This has the advantage that each data entry or all data are assigned to a particular point in time or to a specific step, as a result of which a rapid query can be carried out in accordance with the method described above.

In a further embodiment according to the invention, a determination of the number of data present in the line sections is carried out during the analysis. This is essentially to be able to quickly determine in a user request, how large the amount of data that the user has selected and wants to download.

In another embodiment of the invention, the download of the data from the selected line sections is limited to a maximum number. This is to prevent user requests that cause excessive traffic and other important processes in the data store or communication network of the distributed system architecture, such as collecting the data in the data store.

A query of the data of the line sections can also be carried out, with which query the data is transmitted to another location in the decentralized system architecture.

The method can also be designed such that latencies (when the data is transferred to the data memory) or also data failures based on statistics are analyzed during a query of the line sections, whereby statements about the behavior of the communication network can be made.

The described architecture is thus suitable for storing measured values in a chronological order, this also because the method behaves flexibly with respect to an actual data implementation. The two-stage concept allows the architecture to be used quickly in a variety of applications.

The availability of historical data at a secondary level enables novel applications, such as network-based historical graph monitoring and monitoring systems, allowing network operators a precise sense of the current network status, such as a low voltage network.

To that end, algorithms that are provided with historical data could immediately enable network operation without having to gather information at the beginning for some time to get an idea of the current network condition.

In doing so, statistical analysis of the data may provide insight into the network architecture and system architecture, e.g. a correlation analysis of the voltage levels gives a rough impression of the network layout or the topology. Complex algorithms such as neural networks can use historical data for training.

In an embodiment according to the invention, it is provided that a frequency distribution and / or an arrival time of the data record is stored for each of the data blocks. This means that it is determined in which time intervals the data arrive at the data memory, from which then, as already mentioned above, even faster latency times or data failures of the communication network can be determined.

In summary, the processing of user requests is significantly accelerated by the method according to the invention, since the area to be examined is considerably restricted within the given sequence.

It should be particularly emphasized that the inventive two-stage data storage and the associated method for reading the data not only for power grids, but for any decentralized system architectures, since the respective specific requests are flexible, without causing large data transfers are caused.

BRIEF DESCRIPTION OF THE FIGURES

To further explain the invention, reference is made in the following part of the description to the figure, from the further advantageous embodiments, details and further developments of the invention can be found. Showing:

Fig. 1 is a schematic representation of the basic two-dimensional database structure according to the invention.

WAYS FOR CARRYING OUT THE INVENTION

1 shows a schematic representation of a basic two-dimensional data record 1 according to the invention, wherein a time sequence 3 from t = 0 to t = 20 is indicated continuously in the upper region. Each of the values from 0 to 20 corresponds to a time unit, the time units are the same size. The time units from 0 to 6 are summarized at a predetermined period corresponding to the data block 6, the time units from 7 to 12 are combined at a predetermined period corresponding to the data block 7, and the time units from 13 to 20 are combined at a predetermined time which corresponds to the data block 8. The periods comprising data blocks 6-8 are the same length. Several exemplary data sources (data points) A, Bx, By, Bz, C are given in the form of lines 2,9,10,11,12 with each other. Thus, the time sequence 3 represents a kind of vertical column division, while horizontally extending lines 2, 9, 10, 11, 12 represent a type of line division, the column division being subdivided into data blocks 6, 7, 8 for further subdivision Lines 2,9,10,11,12 in the range of line sections 2a, 2b, 2c, 9a, 9b, 9c, 10a, 1la, 1lb, 12b cut.

A request made by a user for a "data point B *" in this database, which is split into data blocks 6, 7, 8. (ie, all data points beginning with "B") in a range 4 from t = 10 to t = 17 provides only the two lines 9 and 11, ie the data points Bx and Bz, within the data blocks 7 and 8, since the data point By or the line segments 10b and 10c contain no data 5. It should be noted that "B * " includes all "B" data points, say Bx, By and Bz.

In this case, for each line 2, 9, 10, 11, 12 or for each line section 2a, 2b, 2c, 9a, 9b, 9c, 10a, 11a, 11b, 12b, a frequency distribution of the requests and / or an associated time of arrival can be stored, whereby the duration of the data transmission is also logged, which serves two purposes: 1. For certain data 5, the frequency measurements can be exported to provide useful statistics on the behavior of the underlying data transmission network, for example with regard to latencies, power failures, etc. 2. When a request accesses the data record 1, the frequencies of the requested data blocks 7, 8 can be compared with respect to the data 5 concerned and its lines 2, 9, 11, 12 and line sections 2 a, 2 b, 2 c, 9 a, 9 b, 9 c, 10 a, 1la, 1lb, 12b can be used to quickly generate an estimate of the data sets. This could, for example, also serve to suppress very large inquiries by which the

Network connection would be unnecessarily overloaded.

List of Reference Signs: 1 record; 2, 9 ... 12 lines; 2a ... 2c line sections of line 2; 3 temporal sequence; 4 area to be examined; 5 requested data (e.g., from a data point B *); 6 ... 8 data blocks; 9a ... 9c line segments of line 9; 10a line portion of the line 10; 11a ... 11b line segments of line 11; 12b line portion of the line 12; A, Bx, By, Bz, C Data sources (data points)).

Claims (10)

1. A method for storing and selecting data (5) from a data memory in which data (5) from a plurality of data sources in a decentralized system architecture, for example in a low-voltage network, present at least in a record (1) according to a temporal Sequence (3) are stored and queried by means of a user request, characterized in that the records (1) in rows (2,9,10,11,12), which are each assigned to a data source, according to the time sequence (3) and successive line segments (2a, 2b, 2c, 9a, 9b, 9c, 10a, 11a, 11b, 12b) covering predetermined periods of time are combined to form data blocks (6,7,8), the method of selecting Data (5) in a user request comprises the following steps: - selecting those data blocks (7, 8) covering the searched area (4); - analysis of data (5) in the line sections (2b, 2c, 9b, 9c, 11b, 12b) in the data blocks (7, 8); - Selection of those line segments (2b, 2c, 9b, 9c, 11b, 12b) containing data (5). 2. The method according to claim 1, characterized in that the time sequence (3) for storing in constant steps in the form of a timeline is performed. 3. The method according to claim 1 or 2, characterized in that in the analysis, a determination of the number of in the line sections (2b, 2c, 9b, 9c, 11b, 12b) existing data (5) is performed. 4. The method according to any one of the preceding claims, characterized in that the downloading of the data (5) from the selected line sections (2b, 2c, 9b, 9c, 11b, 12b) is limited to a maximum number. 5. The method according to any one of claims 1 to 4, characterized in that latencies are analyzed based on statistics in a query of the line sections (2b, 2c, 9b, 9c, 11b, 12b). 6. The method according to any one of claims 1 to 4, characterized in that data losses based on a statistic in a query of the line sections (2b, 2c, 9b, 9c, 11b, 12b) are analyzed. 7. The method according to any one of the preceding claims, characterized in that a frequency distribution of the data record is stored for each of the data blocks (6,7,8). 8. The method according to any one of the preceding claims, characterized in that an arrival time of the data record is stored for each of the data blocks (6,7,8). 9. The method according to any one of the preceding claims, characterized in that a query of the data (5) of the line sections (2b, 2c, 9b, 9c, 11b, 12b) is performed, in which query the data (5) to another location be transmitted in the decentralized system architecture. 10. Two-stage data memory for a decentralized system architecture, for example in the form of a low-voltage network, for selecting at least one dataset (1) arranged data (5) according to a method according to one of claims 1 to 9, characterized in that the records (1) in rows (2, 9, 10, 11, 12), which are each assigned to a data source, are stored according to the time sequence (3) and successive line segments (2a, 2b, 2c, 9a, 9b, 9c, 10a, 11a, 11b, 12b), which cover predetermined periods of time, are combined into data blocks (6, 7, 8), wherein the data structure is formed at least in two stages, in which a first stage as volatile memory contains younger data (5), while a second stage as permanent Memory contains older data (5) than the first level.