WO2023200361A1

WO2023200361A1 - Data storage method and data extraction method

Info

Publication number: WO2023200361A1
Application number: PCT/RU2023/000095
Authority: WO
Inventors: Алексей Александрович БРЫЗГАЛОВ
Original assignee: Общество С Ограниченной Ответственностью "Авадс Софт"
Priority date: 2022-04-11
Filing date: 2023-03-30
Publication date: 2023-10-19

Abstract

The group of inventions relates to methods for storing data received from different data sources. The technical result is an increase in the speed of recording data without losses, and a reduction in the time taken to extract necessary data. A data storage method includes: receiving data from a data source; saving the data in the form of data records in the memory of a computer, wherein each data record contains a time stamp related to the moment of creation of the data, and a parameter value; generating, from the data records, identically-sized record blocks in the computer memory, wherein a record block can contain data records for only one parameter; assigning to a record block a title including an identifier of the parameter and an address; saving the generated record block in a database on a machine-readable medium; generating and saving in the database a list of addresses of the record blocks for each parameter.

Description

DATA STORAGE AND DATA RETRIEVAL METHOD

Field of technology

The present invention generally relates to methods for storing data and, in particular, to collecting and retrieving data obtained from various data sources.

The manufacturing process is increasingly dependent on data collection and use systems to improve process efficiency and reduce maintenance costs.

Most technological processes are automated and the operation of the equipment is monitored by sensors, the information from which enters the data storage and processing system.

An important aspect of collecting data from sensors is the speed of data recording, without loss, and data archiving in such a way as to obtain data from the archive also at high speed to develop an operational solution.

State of the art

A person skilled in the art is aware of such data archiving systems as SQL Server (Microsoft), archiving systems developed by Wonderware (https://www.wonderware.com, SIEMENS

(https://new.siemens.com/ru/ru/produkty/avtomatizacia/raspredelennye-sistemy-uprayleniya/simatic-7/arhivaciya-dannyh-processa-i-otchetnost.html), etc.

However, these systems (archives) are complex and resource-intensive, have a certain speed of retrieving data from the archive, which does not always meet the objectives of the enterprise and, in addition, they are expensive products and not every enterprise can afford to install these data storage systems.

The closest analogue of the method for collecting and retrieving data is the method described in patent US6651030, cl. G06F15/177, published 18.1 1.2003. The proprietary data collection method involves receiving data from data sources, storing the data as records in a data file with a timestamp when the data was created, and writing a header file that associates the time interval with those records in the data files whose timestamps fall within the time interval. A method for retrieving data records involves selecting a time interval of interest, accessing a header file that associates the time interval of interest with the data records, whose timestamps fall within the interval of interest, retrieving data records based on the timestamps of the data records.

The disadvantage of these methods is that the collected data is written to a data file without separating the data by parameters, which increases the time of data recording and the time of data retrieval for a certain parameter.

The essence of the invention

The claimed technical solution eliminates these disadvantages of the analogue.

The technical result that can be obtained by implementing the proposed solution is to increase the speed of data recording without loss and reduce the time for retrieving the necessary data.

The specified technical result is obtained by creating a method for storing data, including receiving data from a data source, storing the data in the form of data records in computer memory, each data record containing a time stamp associated with the moment of data creation, and a parameter value, formation in memory computer from data records of record blocks of the same size, while one record block can contain data records of only one parameter, assigning a header to the record block, including the parameter identifier and address, and saving the generated record block to the database on a machine-readable wearable gel, formation and saving in the database a list of record block addresses for each parameter.

Moreover, the header of the record block includes the address of the subsequent block of records or the header of the record block includes the address of the previous block of records; in another special case of execution, the header of the record block includes the address of the previous and subsequent block of records.

The record block header may include a timestamp of the first data record in the record block and/or a timestamp of the last data record in the record block and other service information.

Store data records in a record block in ascending order of timestamps. The data record contains additional information.

The list of record block addresses contains timestamps. In this case, for each address from the list, the timestamp of the first data record and/or the time stamp of the last data record of the block of records at this address is stored. Computer memory can be used as a machine-readable medium.

The specified technical result is also obtained through a method of data extraction, including entering a search query for a given parameter and a given time interval, searching in the list of addresses of record blocks and addresses from the headers of record blocks for the address of a record block containing the first time record of parameter data from a given time interval interval, reading data records from the found record block, starting with the first data record whose timestamp is greater than the start time of the specified time interval, moving to subsequent record blocks at addresses from the record block headers, up to the record block that contains the data record, label whose time exceeds the end time of the specified interval, retrieving data records with timestamps that fell within the specified time interval.

In a particular case of performing a method for retrieving data from a list of addresses and addresses from the headers of record blocks, a block of records is found containing the last data record, the timestamp of which does not exceed the end time of the interval, and the data records are read with a transition to the previous blocks of records at the addresses from the headers of the record blocks , to a block of records that contains the first data record whose timestamp is less than the start time of the specified time interval, retrieving data records with timestamps that fall within the specified time interval.

The specified forms of presenting data records, forming blocks of records from them, saving blocks of records into the database, allows you to quickly and reliably find all records by the desired parameter for a given time interval, without going through all blocks of data records recorded in the database, but according to a search query identify the required block of records and from it obtain a record about the requested parameter. This significantly increases the speed of saving records into the database and reduces the search time for the desired parameter and the reliability of the search without missing the required data.

It should be noted that the term “parameter” here means the value of any physical quantity (for example, pressure or temperature in the reactor, liquid level in a container, CO2 concentration, etc.), measured and presented in digital form or obtained based on the values of other parameters .

The term “data record” here means the final form of representing information about the value of a parameter with a timestamp of its receipt. The term "block of records" here means a set of records of the same parameter, sorted by time in their timestamps, and the set of records has a fixed size, the same for all parameters.

The term “database” here means a collection of record blocks of different parameters and lists of record block addresses of each parameter, which may contain timestamps of the first and/or last record of the record block stored on a computer-readable medium and/or computer memory.

Brief description of drawings

The claimed technical solution can be illustrated by the following figures of the drawings, where in FIG. 1 schematically shows the form of representation of the stored information. So record block 1 includes data records 2 and record block header 3. Data record 2 contains a timestamp 4, a parameter value 5, and may optionally contain additional data record data 6, such as, for example, an OPC Foundation data quality attribute. An important component of the record block 1 is its header 3, which in turn includes the parameter identifier 7, the address 8 of the previous or subsequent record block of this parameter and may contain additional information 9. Such additional information may be, for example, timestamps 4 of the first and/or the last data record 2 in record block 1, the address of the previous or subsequent record block, complementing address 8 to store in header 3 the addresses of both the previous and subsequent record block. This form of presentation of the stored information allows you to quickly, without searching through all the saved blocks of records, select the desired block by the header of the record block and find the data record of the desired parameter in it.

In fig. Figure 2 shows a special case of a system for storing blocks of records of one parameter, which displays the placement in the address space of the database sin of successive blocks of records 1. In FIG. 2 shows that each block of records 1, in addition to data records 2, contains a header 3, which stores the address of the subsequent block of records. Thus, the first block of records 1 is located at address 10, the second block of records is at address 11, and the third is at address 12. From FIG. 2 it can be seen that header 3 of the first block of records 1 stores the address 11 of the second (subsequent in time) block of records 1, and the second block of records 1 in its header 3 stores the address 12 of the third block of records 1. In fig. Figure 3 shows another special case of the system for storing record blocks 1 of one parameter, when the header 3 of record block 1 contains the address of the previous record block in time. Here, as in FIG. 2, three consecutive blocks of records 1 are presented. From FIG. 3 you can see that the title 3 of the last block! records 1 stores the address 11 of the second block of records 1 (previous in time), which in its header 3 stores the address 10 of the first in time from those presented in FIG. 3 blocks of records 1. This structure of interconnection of blocks of records allows you to quickly retrieve all data on a parameter for any time interval.

In fig. 4 schematically shows a special case of executing a list 13 that contains structures 14, where each structure 14 consists of the address 8 of a record block 1 and the timestamps 4 of the first and last data record 2 in a given record block 1. The presence of such a list 13 allows for very fast find record block 1 storing data records 2 at a given point in time without scanning all database record blocks.

In fig. Figure 5 schematically shows the process control system. It includes three subsystems: subsystem 16 for collecting and processing data, subsystem HMI (human-machine interface) 18 and archiving subsystem 17. Software components 21, 23 and 26 of these subsystems operate on one computer 15. All of the following examples of the storage method are and data extraction are considered for the system shown in FIG. 5.

Detailed Disclosure of the Invention

Information about the operating status of various production devices represents a huge amount of data. Such large amounts of data must be stored in chronological order and problems arise in storing real-time data. Temporary failure of individual devices brings its own inconveniences, because... Some information may be missed in such a data storage flow. There is a need for a reliable and high-speed method of archiving and retrieving data, which will allow you to quickly make the right decision and, thus, increase the productivity of enterprise equipment or prepare a report with up-to-date information.

The claimed technical solution can be used in industrial automation systems (process control systems, automated control systems, etc.), commercial and technical accounting systems (ASKUE, ASTUE, etc.), as well as in dispatch systems for industrial facilities, housing and communal services facilities and in building automation.

The history of changes over time of each parameter is stored in the form of data records 2 (Fig. 1), containing the value of parameter 5, a time stamp 4 (real time) associated with the moment the value of parameter 5 was received and may contain a set of additional (service) information 6, if one is required. For example, a record may contain an 8-byte timestamp (64-bit Unix time), the value of parameter 5 in floating-point format is 4 bytes, and as additional (service) information 6, for example, a quality attribute in in accordance with the Fondation OPC standard. This construction of data record 2 allows you to arrange data records in sequence according to their time stamps 4. Additional data 6 allows you to take into account their reliability in the future when analyzing the data.

Data records 2 about the history of changes in one parameter over time are stored in the form of blocks of records 1, which have the same size. The number of data records 2 in one record block 1 is limited by its size, all of which refer to only one parameter. Each block of records 1 has a header 3, which in turn includes a parameter identifier 7. This allows for a reduction in record size, since the parameter identifier 7 is present only in header 3 of record block 1 and is distributed to all data records 2 in it, which also leads to reduction time of saving and reading data by reducing the volume of stored and read information, as well as through the use of block operations.

The specified structure of record block 1 leads to the fact that when searching for the necessary information about a parameter, blocks with the identifier of only this parameter are viewed. This results in reduced search time.

Header 3 of record block 1, in addition to the identifier of parameter 7, contains the address 8 of another record block 1 storing data records 2 of the same parameter. This may be the address of the subsequent block of records 1 or the previous one. In fig. Figure 2 shows a special case of implementing a data storage system for one parameter, when each block of records 1 in header 3 stores the address 8 of the subsequent block of records 1 of this parameter. Moreover, record blocks 1 with addresses 10, 11 and 12 are written sequentially in time: block of records 1 with address 10 are the first records in time, and with address 12 the last ones. In fig. Figure 3 shows another special case of implementing a single parameter data storage system, where the same record blocks 1 are shown, but in their headers 3 the address 8 is stored not of the subsequent, but of the previous block of records 1. The presence of addresses 8 of the subsequent or previous block of records 1 in header 3 allows you to quickly find the next block of records 1 to read if the request for data and time exceeds the time range of the stored data records 2 in this block records 1 . This allows you to significantly reduce data retrieval time.

Within each block 1, all data records 2 about the parameter are sorted by time.

The header 3 of record block 1 contains the timestamp 4 of the first record in the record block and/or the timestamp of the last record in this record block. This allows you to quickly search for the desired record block 1 when retrieving data for a given time interval, since to analyze the compliance of a given record block 1 with the request requirements, it is enough to read only its header 3, and not the entire record block 1. If the timestamps of 4 data records 2 of this record block 1 do not fall within the specified time interval, then a transition is made to another block of records 1 of the desired parameter, the address 8 of which is written in header 3 of this block 1.

A database is formed from blocks of records 1. At the same time, I enter the database!' record blocks 1 containing the history of changes in various parameters. They are sequentially written to free addresses on machine-readable media (in the database), regardless of the parameter. In this case, a list of 13 addresses of 8 record blocks 1 is formed for each parameter, which is also saved to the database. The list 13 of addresses 8 of record blocks 1 is shown schematically in FIG. 4. In FIG. 4 shows that information about the addresses of 8 blocks of records 1, containing the history of changes in one parameter, is stored in the form of structures 14 containing the addresses of 8 blocks of records 1. At the hour of execution, these structures 14, in addition to addresses 8, contain timestamps 4 of the first and last data records 2 in a given block of records 1. When each new block of records 1 is saved to the database, the following structure 14 is added to the list 13. In the special case of executing structure 14, it can contain only one timestamp 4 - the first or last data record 2 in the block records 1.

In another special case of execution of structure 14, it may not contain a timestamp 4. In this case, adding a new structure 14 to list 13 is not performed when saving block of records 1 to the database, but at a specified time interval (for example, once a day). Blocks of records 1 are formed with different time intervals, therefore, during the formation of one of such blocks of records 1, several structures 14 (at fixed time intervals) with address 8 of the generated block of records 1 can be stored in list 13. In another special case, for the specified time interval, several record blocks 1 can be saved. In this case, the number of structures 14 with addresses of 8 record blocks 1 in the list 13 will not correspond to the number of record blocks 1 storing the history of this parameter.

In another special case, the number of structures 14 may be less than the number of blocks of records 1 storing the history of one parameter, when, in order to reduce the size of the list 13, structures 14 are stored in it not for each block of records 1, but, for example, every third or every tenth, or another principle for reducing list size is used 13.

The database is stored in the form of one or several files on a machine-readable storage medium (HDD, SSD, SD, etc.) and/or in the computer's RAM.

The proposed method of saving data allows you to obtain information about a parameter from the database as quickly as possible, without loss, i.e. extract all data on a parameter for any time interval, due to the fact that it is not necessary to scan all record blocks 1 stored in the database, and to search for record block 1 of the desired parameter, select from a list of 13 record block addresses 1 address 8 of the block containing the first by time, a record whose timestamp 4 exceeds the start time of the specified time interval. If there is no structure 14 in list 14 with address 8 of such record block 1 (as a result of reducing the size of list 13 or writing structures 14 at fixed time intervals), then a search is made for the last block of records 1 whose timestamp 4 of the last data record 2 is less than the start time a given time interval and then a search is performed with transition to addresses 8 in the headers 3 of blocks of records 1 with analysis of labels 4 of the first and last data record until a block of records 1 is identified containing the first data record 2, the timestamp of which 4 is greater than the start time of the specified time interval. Records are read from the found record block 1, starting from the first data record 2, the timestamp of which is greater than the start time of the specified time interval, to the record block 1, which contains the first data record 2, the timestamp of which is greater than the end time of the specified time interval. , retrieving 2 data records with 4 timestamps that fell within a given time interval. All read records from the first one falling within a given time interval to the last one are the result of data retrieval.

In a particular case of performing a data retrieval method, when in the headers 3 of record blocks 1 the address 8 is stored not of the subsequent record block, but of the previous one, a search is performed using a list of 13 addresses 8 of record blocks 1 and addresses 8 from the headers 3 of record blocks 1 of that record block 1, which contains the first entry data 2, whose timestamp 4 exceeds the end time of the specified time interval. Records are read from the found record block 1, starting from the first data record 2, the timestamp of which 4 is less than the end time of a specified time interval, in the reverse time direction to the record block 1, which contains the last data record 2, the timestamp of which 4 is greater than the start time of a given interval, retrieving records with timestamps that fell within a given time interval. All read records from the first one falling within a given time interval to the last one are the result of data retrieval.

Particular cases of the method of storing and retrieving data can be illustrated by the given non-exhaustive examples, which are illustrated in Fig.5.

In fig. Figure 5 schematically shows the process control system. It includes three subsystems: a data collection and processing subsystem 16, an HMI (human-machine interface) subsystem 18 and an archiving subsystem 17. Software components 21, 23 and 26 of these subsystems operate on one computer 15.

The data collection and processing subsystem 16 includes input/output modules 19, to which sensors are connected that monitor process parameters, and which are polled via the RS-485 interface connected to the USB port 22 of the computer 15 through an RS-485/USB interface converter 20 , as well as a software component 21, which, using the MODBUS protocol with a given frequency, polls the input/output modules 19, processes the received data, converting them into values of physical quantities in float format, assigns them time stamps corresponding to the moment of receiving this data via the protocol, and a sign of quality in accordance with the OPO Foundation standard. Next, the data collection and processing subsystem 16 transmits the values of the monitored process parameters along with time stamps and quality attributes to the HMI subsystem 18 and the archiving subsystem 17. Values are transferred only for those parameters whose values have changed compared to the previous transmitted value by an amount greater than 0, 2% of their possible range of change.

The HMI subsystem 18 consists of means of interaction with the operator (monitor 27, keyboard 28 and mouse-type manipulator 29) and a software component 26, which processes operator commands submitted through the keyboard 28 and mouse-type manipulator 29, sends requests and receives data from the collection subsystem 16 and processing data and archiving subsystem 17, forms a visual image based on the received data and displays it on the monitor 27.

The archiving subsystem 17 consists of two mirrored hard drives 25, a fixed area of computer RAM 24 and an archiving software component 23, which receives data from the data collection and processing subsystem 16, forms data records 2 from the received data, from which it forms blocks in RAM 24 records 1, for example, 200 data records 2 of one parameter in each block of records 1 and saves blocks of records 1 as they are filled with data records 2 in the database on disks 25. Moreover, for each parameter in RAM 24 space is allocated to form one block records 1. After filling this block of records 1 with data records 2 and saving it to the database on disks 25, the space it occupies in memory 24 is released and used to form the next block of records 1 of this parameter. Moreover, data record 2 consists of a time stamp 4 in 64-bit Unix time format (8 bytes), parameter value 5 in float format (4 bytes) and additional information 6 in the form of a data quality indicator in accordance with the OPC Foundation standard (1 byte ). In addition, the archiving subsystem 17 generates in RAM 24 for each parameter lists 13 of addresses for the placement of record blocks 1 storing data records 2 with the values of this parameter. These lists 13, as new information is added to them, are also saved to the database on disks 25. In addition, the archiving subsystem 17 receives requests for data searches from the HMI subsystem 18, retrieves the requested data and transmits them to the HMI subsystem 18.

Example 1.

The time stamp 4 of the last data record 2 and the address 8 of the subsequent record block 1 are stored in the header 3 of the record block, and in the list 13 for each parameter, structures 14 are stored containing the address 8 and the time stamp 4 of the last data record 2, for every tenth block of records 1 of this parameter in the database.

In this case, when each block of records 1 is saved to the database on disks 25, in its header 3 the time stamp 4 of its last data record 2 is entered as additional information 9, and in the header 3 of the previous block of records 1 the address 8 is entered, at which this block of records 1 is written. The block of records 1 is saved to the database on disks 25 if it is filled with data records 2 to the full specified volume. In addition, when saving every tenth block of records 1 of this parameter to the database on disks 25, structure 14 is added to list 13, containing the address 8 of the record block 1 being stored and the timestamp 4 of the last data record 2 of this record block 1.

To view archived data in the HMI subsystem 18, for example, to display changes in a parameter on a trend for the previous week, the software component 26 of the HMI 18 subsystem generates a request to the archiving subsystem 17, which specifies the parameter identifier 7, as well as the start time and end time of the time interval . In this case, the start of the time interval will be 00:00:00.000 of the previous Monday, and the end will be 23:59:59.999 of the previous Sunday. To prepare a sample of records in this time interval, first search the database for record block 1, for which the start time of the specified time range (00:00:00.000 Monday) is between timestamps 4 of the first and last data record 2. To do this, access the list 13 for a given parameter. In this list, the address 8 of the last block of records 1 is found, the time of the last record of which is less than the specified start time of the time interval. Read header 3 of the found record block 1 and move to the next record block 1 at address 8, written in the found record block 1. Read its header 3. If the timestamp 4 of the last data record 2 stored in it is less than the start time of the specified time interval, then address 8 from header 3 of this block of records 1 goes to the next block of records 1. And so on until the time stamp 4 of the last data record 2 in header 3 of the next block of records 1 does not exceed the start time of the specified time interval. This data record 2 is considered to be the first one found in the required interval. Next, all data records 2 of this block of records 1 are read and the first data record 2 is found among them, the timestamp 4 of which will be greater than the start time of the specified time interval. All data records 2, starting from the first found and up to the last of this block of records 1, are saved in the sample. Next, at address 8 from header 3 of this record block 1, they move to the next record block 1. If timestamp 4 from header 3 of this record block 1 is less than the end time of the specified time interval (23:59:59.999 Sunday), then all data records 2 from of this block of records 1 is extracted and added to the selection. If timestamp 4 from header 3 of this record block 1 exceeds the end time of the specified time interval, then all data records 2 of this record block 1 are read, but put! The sample includes only those data records 2 whose timestamps 4 are less than the end of the specified time interval. At this point, the search for records ends and all data records placed in the sample are transferred to the software component 26 of the I1M1 1 subsystem, which, based on the resulting sample, builds an image of the trend of changes in the value of a given parameter, which is then displayed on monitor 1. Moreover, when forming the image, the value of the quality attribute is analyzed, which is an additional value 6 of each data record 2. Where the value of this attribute is not equal to 192 (the data has GOOD quality) the trend is shown as a dotted line.

Example 2.

The 1 timestamp 4 of the first data record 2 of this record block 1 and the address 8 of the previous record block 1 are stored in the header 3 of the record block 1, and a structure 14 is added to the list 13 containing only the address 8 of the current record block 1 to be filled, once a day at the moment the onset of a new day.

In this case, when each record block 1 is saved to the database on disks 25, its address 8 is written to header 3 of the newly created record block 1. When the first data record 2 is added to the newly created record block 1, its timestamp 4 is written to header 3 of this block 1 entries as additional information 9.

Just as in Example 1, to view in the HMI subsystem 18 changes in the parameter on the trend for the previous week, the software component 26 of the subsystem 1 GM1 18 generates a request to the archiving subsystem 17, which specifies the identifier of parameter 7, as well as the start time and time the end of the time interval: 00:00:00.000 of the previous Monday and 23:59:59.999 of the previous Sunday. In this case, to prepare a sample of data records 2, first find in the database a block of records 1 for which the end time of the specified time range (23:59:59.999 Sunday) is located between timestamps 4 of the first and last data record 2. For this, use list of 13 addresses 8 for a given parameter. Since entries in this list 13 are entered once a day, the serial number of each entry determines the time accurate to the day, provided that the date of the first entry is known. For example, by the beginning of the specified search time interval, the system had already been operating for 1 year and 2 weeks. Then address 8 of record block 1, which was filled at the beginning of a given time interval, is in the list of 13 addresses in 379th place (365 - one year and 14 - two weeks), and address 8 of record block 1, which was filled at the end of the specified time interval interval is in 386th place (a week later). Therefore, the 386th address in the list of 13 addresses is read and goes to record block 1 at this address 8. All data records 2 from this record block 1 are read. All data records 2 whose timestamp 4 is less than the end time of the specified time interval are retrieved and placed in sample. Next, they go to address 8 from header 3 of this record block 1 to the previous record block 1. Read data records 2 from the previous record block 1. If time stamp 4 in header 3 of this record block 1 is greater than the start time of the specified time interval, then all data records 2 of a given block of records 1 are extracted and placed in the sample. Otherwise, only data records 2 whose timestamp 4 is greater than the start time of the specified time interval are placed in the sample. This completes the sampling, and it is transferred to the software component 26 of the HMI 18 subsystem.

Example 3.

The address 8 of the subsequent record block 1 is stored in the header 3 of the record block 1, as well as additional information 9, including the timestamps 4 of the first and last data records 2 of this record block 1 and the address 8 of the previous record block 1. In this case, the list 13 stores structures 14, including the address 8 of record block 1 and the timestamps 4 of its first and last data records 2.

In this case, when saving a record block 1 to the database on disks 25, the address 8 of the newly created (subsequent to the saved) record block 1 and the timestamp 4 of the last record of the saved record block 1 are written into its header 3. In this case, the address 8 of the saved record block and the timestamp 4 of the first data record 2 when it is added to the new record block 1 are written into the header 3 of the newly created record block 1.

The HMI 18 subsystem requests parameter data for the previous week. The HMI subsystem 18 was given the extracted sample of data records 2. However, after receiving the sample of data records 2 to display the trend, the operator changed the scale of the trend time axis, which required adding data three more days before the start time of the previous request time interval and two days after end of the previous request interval. In this case, you need to add data to the already obtained selection. Moreover, both after and before a given time interval. To do this, two additional samples are performed. The first is to obtain data records from the end time of the initial time interval (23:59:59.999 of the previous Sunday) to 23:59:59.999 of the Tuesday of the current week. To do this, use the mechanism described in Example 1. First, using list 13, find the address of record block 1 corresponding to the start time of the time interval of this sample (23:59:59.999 of the previous Sunday). In this case, the timestamps 4 of the first and last data records 2 from the structures 14 of the list 13 are used. Read, according to the one found in this way address 8, record block 1. If the time stamp 4 of the last data record 2 in header 3 is less than the end time of the time interval (23:59:59.999 Tuesday of the current week), then select from the read record block 1 all data records 2, the timestamps of which 4 greater than the start time of the time interval, extract them into the sample and then read record block 1, located at address 8 of the subsequent record block 1 from header 3. If time stamp 4 of the last data record 2 in header 3 of this (subsequent) record block 1 is less than the end time time interval, then all data records 2 of this block of records 1 are retrieved and stored in the sample. The specified operations are repeated until the time stamp 4 of the last record 2 of the next block of records 1 is greater than the end time of the specified interval. In this case, only those data records 2 from the last read block of records 1 are saved in the sample, the timestamp 4 of which is less than the end time of the specified interval. This completes the filling of the first additional sample. If the value of the time stamp 4 of the last record 2 of the first read block of records 1 exceeds the end time of the specified time interval, then the data records 2 of this block of records 1 are placed in the sample, the timestamps 4 of which fall within the specified time interval and there is no transition to other blocks of records 1 perform.

To form a second additional sample, use the sequence of actions described in Example 2, when you go to address 8 of the previous block of records 1 from header 3 of the current block of records 1, and compare the value of timestamp 4 of the first data record 2 with the start time of the specified time interval to determine last read block of records 1 to form a sample. In this case, a selection is made of data records 2, the timestamps 4 of which are in the range from 00:00:00.000 Friday of the previous week to 00:00:00.000 of the previous Monday. Using list 13, the address of record block 1 is determined, corresponding to the end time of the time interval of the second sample (00:00:00.000 of the previous Monday). For this, time stamps 4 of the first and last data records 2 from structures 14 of list 13 are used. Address 8 of record block 1 is read from the address 8 found in this way. If the value of timestamp 4 of the first data record of this record block 1 is less than the start time of the specified time interval, then select from record block 1 data records 2 whose timestamps 4 fall within a given time interval, extract and place them in the sample, and complete the search. If the value of timestamp 4 of the first data record 2 of this block of records 1 is greater than the start time of the specified time interval 4, then it is placed in fetch all data records 2 whose timestamps 4 are less than the end time of the specified time interval and read the previous record block 1 at address 8 from header 3 of this record block 1. If the timestamp 4 of the first data record 2 is greater than the start time of the specified time interval, then all data records 2 of this block of records 1 are extracted and placed in the sample and the transition to reading the previous block of records 1 is performed. These operations are performed until the time stamp 4 of the next read block of records 1 is less than the start time of the specified time interval. In this case, only those data records 2 are saved in the sample, the timestamps 4 of which are greater than the start time of the specified time interval, and the formation of the sample is completed.

The resulting two additional samples of data records 2 are transferred to the HMI 18 subsystem to complete the missing trend fragments.

Due to the presence in the header 3 of the block of records 1 of addresses 8 of both the previous, the same and the subsequent block of records 1, as well as time stamps 4 of both the first and last data record 2, it is possible to make samples of data records 2 both forward, backward and backward in time, depending on what is more convenient in a particular case.

Claims

CLAIM

1. A method of storing data, including receiving data from a data source, storing the data in the form of data records in computer memory, with each data record containing a time stamp associated with the moment of data creation and a parameter value, forming blocks from data records in computer memory records of the same size, while one block of records can contain data records of only one parameter, assigning a header to the block of records, including the parameter identifier and address, and saving the generated block of records to the database on a machine-readable medium, generating and saving a list of block addresses in the database entries for each parameter.

2. The method of claim 1, wherein the record block header includes the address of a subsequent record block.

3. The method of claim 1, wherein the record block header includes the address of the previous record block.

4. The method of claim 1, wherein the record block header includes the address of the previous and subsequent record block.

5. The method of claim 1, wherein the record block header includes a timestamp of the first data record in the record block and/or the last data record in the record block.

6. The method of claim 1, wherein data records are stored in the record block in ascending order of timestamps.

7. The method of claim 1, wherein the list of record block addresses contains timestamps.

8. The method according to claim 7, wherein the list contains, for each address, a timestamp of the first data record and/or a timestamp of the last data record of a block of records at a given address.

9. The method according to claim 1, in which computer RAM can be used as a machine-readable medium.

10. A method for retrieving data saved by the method according to claim 1, including entering a search query for a given parameter and a given time interval, searching in the list of addresses of record blocks and addresses from the headers of record blocks for the address of a record block containing the first time record of parameter data from a given time interval, reading data records from the found record block, starting with the first data record whose timestamp is greater than the start time of the specified time interval, moving to subsequent record blocks at the addresses from the record block headers, up to the record block that contains the data record, timestamp which exceeds the end time of a given interval, retrieving data records with timestamps that fell within a given time interval.

11. The method for retrieving data according to claim 10, in which, using the list of addresses and addresses from the headers of record blocks, they find a block of records containing the last data record, the time stamp of which does not exceed the end of the interval, and read the data records with a transition to the previous blocks of records according to addresses from the record block headers, up to the record block containing the first data record whose timestamp is less than the start time of the specified time interval, retrieving data records with timestamps that fall within the specified time interval.