Device and method for automatic adaptation and fusion of power plant data protocols based on machine learning

Abstract

The present invention relates to the field of power systems, and in particular to a device and method for automatic adaptation and fusion of power station data protocols based on machine learning, wherein the fusion device includes: an edge computing module: used for real-time analysis, learning and processing of data, network communication testing, parsing key parameter information of measurement point configuration files, automatically calling data protocol communication parameters, etc.; a storage module: used for storing real-time collected data and historical data, a power module, a data interface group, a liquid crystal display, an indicator light group and a box. The data protocol adaptation of this application is simple, the implementation cost is low, the debugging time is short, and the professional requirements are low. It significantly reduces the risk of inaccurate data caused by manual input requirements and human errors, saves a lot of human resource costs, and at the same time enhances data processing capabilities, improves data collection quality, integrates machine learning diagnostic functions, and quickly analyzes, processes, diagnoses and makes decisions on data, helping to discover and solve problems in a timely manner and avoid potential risks.

Description

Technical Field

[0001] The present invention relates to the field of power systems, and in particular to a device and method for automatic adaptation and fusion of power plant data protocols based on machine learning. Background Art

[0002] The new power system is the main carrier for building a new energy system. As the core link of the new power system's "source, grid, load and storage", the efficient interaction of power station data is the basic guarantee for the new power system to be safe, efficient, clean, low-carbon, flexible, and intelligently integrated.

[0003] The related patents in the prior art include the Chinese invention patent with the patent application number of 201410838561.9 and the name of "A Power Dispatching Automation Master Station System Test Platform and Its Test Method", which discloses: The present invention relates to a power dispatching automation master station system test platform and its test method, which organically integrates the support layer, power system simulation layer, dispatching automation system simulation layer, test verification layer, and interface adaptation layer into a test platform, provides a set of standard test environment configurations, and can provide source code testing, control software closed-loop testing, power system avalanche testing, protocol testing and other test items for the dispatching automation master station system. Although the above patent can effectively integrate various test verification resources to achieve standardized and automated testing of the object under test. However, in view of the complex and diverse data protocols of power stations, there are still problems such as complex data protocol adaptation, high implementation cost, long debugging time, and high professional requirements. Summary of the invention

[0004] The present invention provides a device and method for automatic adaptation and fusion of power plant data specifications based on machine learning, which automatically adapts and fuses power plant data specifications and uses artificial intelligence machine learning diagnostic algorithms to analyze and diagnose data, thereby enhancing the intelligent fusion capabilities of regional power generation bases.

[0005] To achieve the above technical effects, the present invention is implemented through the following technical solutions:

[0006] The power plant data protocol automatic adaptation and fusion device based on machine learning includes:

[0007] Edge computing module: used for real-time analysis, learning and data processing, network communication testing, parsing parameter information of measurement point configuration files, automatically calling data protocol communication parameters, data cleaning, data storage, machine learning diagnosis, reading database data in the storage module to obtain historical data of the measurement point to be diagnosed and related measurement points, calculating typical working condition benchmark values, comparing measurement point data with working condition benchmark values, and making abnormal judgments and data forwarding;

[0008] Storage module: used to store programs and data, including real-time collected data and historical data;

[0009] Power module: used to provide power to each part;

[0010] Data interface group: used to connect various data devices or debugging equipment;

[0011] LCD display: used for function settings;

[0012] Indicator light group: used to display the status of each module and data interface in the fusion device;

[0013] Box: used to place the edge computing module, storage module, power module, data interface, and LCD display.

[0014] Furthermore, the LCD display and indicator light group are arranged on the front side of the box; the edge computing module, storage module, and power supply module are arranged inside the box; and the data interface group is located on the back side of the box.

[0015] Furthermore, the LCD is a touch LCD, which performs parameter settings on the automatic adaptive fusion, data storage function, data forwarding function, and equipment configuration information function of the fusion device.

[0016] Furthermore, the data interface group includes an Ethernet interface, a serial interface, a switch input / output interface and an external display interface.

[0017] The automatic adaptation and fusion method of power plant data specification based on machine learning includes the following steps:

[0018] A. Use the LCD display to configure information functions, respectively configure the network attribute parameters of the fusion device and the external data source device, the network attribute parameters include: IP address, subnet mask, gateway and DNS server address;

[0019] B. The fusion device and the external data source device automatically perform a network communication test by sending a network connection program command, and the network connection program command includes an ICMP Echo request message command;

[0020] C. If the network communication is normal, the parameter information of the measurement point configuration file is automatically parsed, and the data protocol communication parameters are called from the parameter configuration database for automatic adaptation; if the network communication is abnormal, return to step A;

[0021] D. Clean and store the collected data;

[0022] E. Perform machine learning diagnosis on the selected data and forward the result data after diagnosis.

[0023] Furthermore, in step D, the data cleaning improves the quality of the collected data according to the steps of data review, missing value processing, outlier processing, data consistency check, and data verification; the data storage is to store the cleaned data into a database in the storage module.

[0024] Further,

[0025] The machine learning diagnostic steps in step E include:

[0026] EA. Determine the measuring point data to be diagnosed through the LCD display settings, and mark the associated measuring point data that affects the numerical value of the measuring point data to be diagnosed;

[0027] EB. Read the database data in the storage module to obtain the historical data of the measurement point to be diagnosed and the associated measurement points, and determine the reference value of the data of the measurement point to be diagnosed under the corresponding numerical conditions of each associated measurement point;

[0028] EC. Compare the current value of the data at the measurement point to be diagnosed with the reference value under the current working condition, determine whether the data value of the measurement point to be diagnosed is abnormal, and forward the diagnosis result data.

[0029] Furthermore, step E is specifically as follows:

[0030] Perform machine learning diagnosis on the measurement point A to be diagnosed. The associated data B and the associated data C are the associated data of the measurement point A to be diagnosed. The values ​​of the associated data B and the associated data C will affect the value of the measurement point A to be diagnosed. The data value (A, B, C) at this moment is regarded as a working condition.

[0031] The measurement point A to be diagnosed is determined by the LCD display setting, and the associated data B and associated data C associated with the measurement point A to be diagnosed are selected. The edge computing module reads the database data in the storage module to obtain the historical data of the measurement point A to be diagnosed and the associated measurement points B and C, and determines the reference value of the measurement point A to be diagnosed under the corresponding numerical conditions of each associated measurement point. The steps for determining the reference value are as follows:

[0032] Set the threshold ε, the values ​​of the diagnostic point A under three typical working conditions are A1, A2, and A3, and the corresponding values ​​of the associated data B and associated data C are (B1, C1), (B2, C2), and (B3, C3). Search the historical database for the values ​​of the diagnostic point A under the working conditions of the associated data B and associated data C in the range of (B1-ε≤B≤B1+ε, C1-ε≤C≤C1+ε), (B2-ε≤B≤B2+ε, C2-ε≤C≤C2+ε), and (B3-ε≤B≤B3+ε, C3-ε≤C≤C3+ε). Take 100 times, and we have (A1 1 , A1 2 , A1 3 , ..., A199 , A1 100 ), (A2 1 , A2 2 , A2 3 , ..., A2 99 , A2 100 ), (A3 1 , A3 2 , A3 3 , ..., A3 99 , A3 100 ), then its average The following are the benchmark values ​​of the test point A to be diagnosed under three typical working conditions:

[0033] ;

[0034] Compare the current value of the data at the point to be diagnosed with the reference value under the current working condition and set the threshold , that is, when the current associated data B and associated data C values ​​are within the range of (B1-ε≤B≤B1+ε, C1-ε≤C≤C1+ε), (B2-ε≤B≤B2+ε, C2-ε≤C≤C2+ε), (B3-ε≤B≤B3+ε, C3-ε≤C≤C3+ε), if the current value of the measurement point A to be diagnosed exceeds , , range, it means that under the three typical working conditions, the A value of the measurement point to be diagnosed is abnormal, and the current A value of the measurement point to be diagnosed and the abnormal alarm identification value of the diagnosis result data are forwarded.

[0035] Furthermore, the automatically parsed measuring point configuration file in step C includes measuring point name, measuring point type, measuring point coefficient, function code, register address, word length and offset parameters, and the above parameters are automatically configured.

[0036] Furthermore, the calling data protocol communication parameters in step C specifically automatically configure the port, baud rate, data bit, check mode, stop bit, interval time, byte stream mode, transmission delay, timeout, packet length, byte order and public address parameters.

[0037] The beneficial effects of the present invention are:

[0038] 1. The data protocol adaptation targeted by the device and method is simple, with low implementation cost, short debugging time and low professional requirements, which significantly reduces the risk of inaccurate data caused by manual input requirements and human errors, saves a lot of human resource costs, and at the same time enhances data processing capabilities, improves data collection quality, integrates machine learning diagnostic functions, and quickly analyzes, processes, diagnoses and makes decisions on key data, helping to promptly discover and solve problems and avoid potential risks.

[0039] 2. This application can be applied to thermal power plants, hydroelectric power plants, nuclear power plants, geothermal power plants, wind power plants, solar power plants, garbage power plants, ocean power plants, etc. And the data protocols applicable to this application include national standard protocols, power protocols, Internet of Things protocols, industrial protocols and other protocols. National standard protocols include Modbus RTU, Modbus TCP and Modbus ASCII, the power protocols include DLT645, IEC-61850, IEC-101, ICE103 and IEC104, the Internet of Things protocols include MQTT, the industrial protocols include CAN, and the other protocols include PLC protocols and smart meter related protocols. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic diagram of the structure of the fusion device of the present application.

[0041] Figure 2 It is a flowchart of the fusion method of the present application.

[0042] Among them, the figure markings are: 1-edge computing module, 2-storage module, 3-power module, 4-data interface group, 5-LCD display, 6-indicator light group, 7-box. DETAILED DESCRIPTION

[0043] The following is a detailed description of an embodiment of the present invention. This embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation method and a specific operation process are given, but the protection scope of the present invention is not limited to the following embodiment.

[0044] It should be pointed out that all directional indications in the embodiments of the present invention (such as both sides, edges, up, down, left, right, front, back, middle, top, bottom, tail, axial, radial...) are only used to explain the relative position relationship, movement state, etc. between the components in a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0045] Example 1

[0046] like Figure 1 As shown, the power plant data protocol automatic adaptation and fusion device based on machine learning includes:

[0047] Edge computing module: used for real-time analysis, learning and data processing, network communication testing, parsing parameter information of measurement point configuration files, automatically calling data protocol communication parameters, data cleaning, data storage, machine learning diagnosis, reading database data in the storage module to obtain historical data of the measurement point to be diagnosed and related measurement points, calculating typical working condition benchmark values, comparing measurement point data with working condition benchmark values, and making abnormal judgments and data forwarding;

[0048] Storage module: used to store programs and data, including real-time collected data and historical data, stored in time series databases or other similar databases;

[0049] Power module: used to provide power to each part;

[0050] Data interface group: used to connect various data devices or debugging equipment;

[0051] LCD display: used for function settings;

[0052] Indicator light group: used to display the status of each module and data interface in the fusion device;

[0053] Box: used to place the edge computing module, storage module, power module, data interface, and LCD display.

[0054] The liquid crystal display and indicator light group are arranged on the front side of the box; the edge computing module, storage module, and power supply module are arranged inside the box; and the data interface group is located on the back side of the box.

[0055] The power module is provided with a power switch button and an external power connection terminal. The power switch button is used to start or shut down the fusion device, and the external power connection terminal is connected to an external AC power source through a power connection line.

[0056] The LCD is a touch LCD, which is used to set parameters for the automatic adaptive fusion, data storage function, data forwarding function, and equipment configuration information function of the fusion device.

[0057] The data interface group includes an Ethernet interface, a serial interface, a switch input / output interface and an external display interface.

[0058] The data protocol adaptation of this device is simple, the implementation cost is low, the debugging time is short, and the professional requirements are low. It significantly reduces the need for manual input and the risk of inaccurate data caused by human errors, saving a lot of human resource costs. At the same time, it enhances data processing capabilities, improves data collection quality, and integrates machine learning diagnostic functions to quickly analyze, process, diagnose and make decisions on key data, helping to discover and solve problems in a timely manner and avoid potential risks.

[0059] Example 2

[0060] like Figure 2 As shown, the automatic adaptation and fusion method of power plant data specifications based on machine learning includes the following steps:

[0061] A. Use the LCD display to configure information functions, respectively configure the network attribute parameters of the fusion device and the external data source device, the network attribute parameters include: IP address, subnet mask, gateway and DNS server address, etc.;

[0062] B. The fusion device and the external data source device automatically perform a network communication test by sending a network connection program command, and the network connection program command includes an ICMP Echo request message command;

[0063] C. If the network communication is normal, the parameter information of the measurement point configuration file is automatically parsed, and the data protocol communication parameters are called from the parameter configuration database for automatic adaptation; if the network communication is abnormal, return to step A;

[0064] D. Clean and store the collected data;

[0065] E. Perform machine learning diagnosis on the selected data and forward the result data after diagnosis.

[0066] In step D, the data cleaning improves the quality of the collected data according to the steps of data review, missing value processing, outlier processing, data consistency check, and data verification; the data storage is to store the cleaned data into a database in the storage module.

[0067] The machine learning diagnostic steps in step E include:

[0068] EA. Determine the measuring point data to be diagnosed through the LCD display settings, and mark the associated measuring point data that affects the numerical value of the measuring point data to be diagnosed;

[0069] EB. Read the database data in the storage module to obtain the historical data of the measurement point to be diagnosed and the associated measurement points, and determine the reference value of the data of the measurement point to be diagnosed under the corresponding numerical conditions of each associated measurement point;

[0070] EC. Compare the current value of the data at the measurement point to be diagnosed with the reference value under the current working condition, determine whether the data value of the measurement point to be diagnosed is abnormal, and forward the diagnosis result data.

[0071] Step E is specifically as follows:

[0072] Perform machine learning diagnosis on the measurement point A to be diagnosed. The associated data B and the associated data C are the associated data of the measurement point A to be diagnosed. The values ​​of the associated data B and the associated data C will affect the value of the measurement point A to be diagnosed. The data value (A, B, C) at this moment is regarded as a working condition.

[0073] The measurement point A to be diagnosed is determined by the LCD display setting, and the associated data B and associated data C associated with the measurement point A to be diagnosed are selected. The edge computing module reads the database data in the storage module to obtain the historical data of the measurement point A to be diagnosed and the associated measurement points B and C, and determines the reference value of the measurement point A to be diagnosed under the corresponding numerical conditions of each associated measurement point. The steps for determining the reference value are as follows:

[0074] Set the threshold ε, the values ​​of the diagnostic point A under three typical working conditions are A1, A2, and A3, and the corresponding values ​​of the associated data B and associated data C are (B1, C1), (B2, C2), and (B3, C3). Search the historical database for the values ​​of the diagnostic point A under the working conditions of the associated data B and associated data C in the range of (B1-ε≤B≤B1+ε, C1-ε≤C≤C1+ε), (B2-ε≤B≤B2+ε, C2-ε≤C≤C2+ε), and (B3-ε≤B≤B3+ε, C3-ε≤C≤C3+ε). Take 100 times, and we have (A1 1 , A1 2 , A1 3 , ..., A1 99 , A1 100 ), (A2 1 , A2 2 , A2 3 , ..., A2 99 , A2 100 ), (A3 1 , A3 2 , A3 3 , ..., A3 99 , A3 100 ), then its average The following are the benchmark values ​​of the test point A to be diagnosed under three typical working conditions:

[0075] ;

[0076] Compare the current value of the data at the point to be diagnosed with the reference value under the current working condition and set the threshold , that is, when the current associated data B and associated data C values ​​are within the range of (B1-ε≤B≤B1+ε, C1-ε≤C≤C1+ε), (B2-ε≤B≤B2+ε, C2-ε≤C≤C2+ε), (B3-ε≤B≤B3+ε, C3-ε≤C≤C3+ε), if the current value of the measurement point A to be diagnosed exceeds , , range, it means that under the three typical working conditions, the A value of the measurement point to be diagnosed is abnormal, and the current A value of the measurement point to be diagnosed and the abnormal alarm identification value of the diagnosis result data are forwarded.

[0077] Automatically parsing the measuring point configuration file can eliminate the need to configure parameters on a liquid crystal display or on a Web page accessed through the debugging interface. The automatically parsing measuring point configuration file in step C includes measuring point name, measuring point type, measuring point coefficient, function code, register address, word length and offset parameters. The above parameters are automatically configured, and the configuration parameters of different data protocols may be different.

[0078] The calling data protocol communication parameters in step C specifically automatically configure parameters such as port, baud rate, data bit, check mode, stop bit, interval time, byte stream mode, transmission delay, timeout, packet length, byte order and public address. The communication parameters of different data protocols are different.

[0079] The data protocol for this method is simple to adapt, has low implementation cost, short debugging time, and low professional requirements. It significantly reduces the risk of inaccurate data caused by manual input requirements and human errors, saves a lot of human resource costs, and at the same time enhances data processing capabilities, improves data collection quality, integrates machine learning diagnostic functions, and quickly analyzes, processes, diagnoses and makes decisions on key data, helping to discover and solve problems in a timely manner and avoid potential risks. This application can be targeted at thermal power stations, hydropower stations, nuclear power stations, geothermal power stations, wind power stations, solar power stations, garbage power stations, ocean power stations, etc. And the data protocols applicable to this application include national standard protocols, power protocols, Internet of Things protocols, industrial protocols and other protocols. National standard protocols include Modbus RTU, Modbus TCP and Modbus ASCII, the power protocols include DLT645, IEC-61850, IEC-101, ICE103 and IEC104, the Internet of Things protocols include MQTT, the industrial protocols include CAN, and the other protocols include PLC protocols and smart meter related protocols.

[0080] Example 3

[0081] See attached Figure 1 ,The power station data protocol automatic adaptation and fusion device based on machine learning includes an edge computing module, a storage module, a power module, a data interface group, a liquid crystal display, an indicator group and a cabinet.

[0082] The power module is connected to the external power supply, and the power button on the power module is pressed to turn on the power module. The power indicator in the indicator group lights up; the RJ45 interface in the data interface group is directly connected to the external data source device through a network cable or indirectly connected to the external data source device through a networking device (such as a switch, firewall, etc.). The network IP address of the external data source device is configured through the LCD display. The edge computing module first performs a network communication test. If the network communication is normal, the multiple data protocols are automatically adapted, the key parameter information of the measurement point configuration file is automatically parsed, and the data protocol communication parameters are called from the parameter configuration database. For example, if it is a ModbusTCP data protocol, the data bit, check mode, stop bit, interval time, byte stream mode, transmission delay, timeout, packet length, and byte order parameters are automatically set. If it is an IEC104 data protocol, the time calibration mode, total call mode, cumulative collection mode, communication public address and other parameters are automatically set. Configure the data forwarding function parameters through the LCD display, automatically generate a forwarding data point table according to actual needs, and forward data to the external data receiving device according to the set forwarding protocol.

[0083] If a machine learning diagnosis is performed on a certain measurement point A to be diagnosed, the associated data B and the associated data C are the associated data of the measurement point A to be diagnosed, and the values ​​of the associated data B and the associated data C will affect the value of the measurement point A to be diagnosed, then the data value (A, B, C) at this moment is regarded as a working condition;

[0084] The measurement point A to be diagnosed is determined by the LCD display setting, and the associated data B and associated data C associated with the measurement point A to be diagnosed are selected. The edge computing module reads the database data in the storage module to obtain the historical data of the measurement point A to be diagnosed and the associated measurement points B and C, and determines the reference value of the measurement point A to be diagnosed under the corresponding numerical conditions of each associated measurement point. The steps for determining the reference value are as follows:

[0085] Set the threshold ε, the values ​​of the diagnostic point A under three typical working conditions are A1, A2, and A3, and the corresponding values ​​of the associated data B and associated data C are (B1, C1), (B2, C2), and (B3, C3). Search the historical database for the values ​​of the diagnostic point A under the working conditions of the associated data B and associated data C in the range of (B1-ε≤B≤B1+ε, C1-ε≤C≤C1+ε), (B2-ε≤B≤B2+ε, C2-ε≤C≤C2+ε), and (B3-ε≤B≤B3+ε, C3-ε≤C≤C3+ε). Take 100 times, and we have (A1 1 , A1 2 , A1 3 , ..., A1 99 , A1 100 ), (A2 1 , A22 , A2 3 , ..., A2 99 , A2 100 ), (A3 1 , A3 2 , A3 3 , ..., A3 99 , A3 100 ), then its average The following are the benchmark values ​​of the test point A to be diagnosed under three typical working conditions:

[0086] ;

[0087] Compare the current value of the data at the point to be diagnosed with the reference value under the current working condition and set the threshold , that is, when the current associated data B and associated data C values ​​are within the range of (B1-ε≤B≤B1+ε, C1-ε≤C≤C1+ε), (B2-ε≤B≤B2+ε, C2-ε≤C≤C2+ε), (B3-ε≤B≤B3+ε, C3-ε≤C≤C3+ε), if the current value of the measurement point A to be diagnosed exceeds , , range, it means that under the three typical working conditions, the A value of the measurement point to be diagnosed is abnormal, and the current A value of the measurement point to be diagnosed and the abnormal alarm identification value of the diagnosis result data are forwarded.

[0088] Those skilled in the art will understand that Figure 1 The structure shown in the figure is only a structural diagram of the part related to the fusion device disclosed in the present invention, and does not constitute a limitation on the fusion device of the present invention. The specific fusion device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0089] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for automatic adaptation and fusion of power plant data specifications based on machine learning, characterized by: The steps include: A. Use the LCD display to configure information functions, respectively configure the network attribute parameters of the fusion device and the external data source device, the network attribute parameters include: IP address, subnet mask, gateway and DNS server address; B. The fusion device and the external data source device automatically perform a network communication test by sending a network connection program command, and the network connection program command includes an ICMP Echo request message command; C. If the network communication is normal, the parameter information of the measurement point configuration file is automatically parsed, and the data protocol communication parameters are called from the parameter configuration database for automatic adaptation; if the network communication is abnormal, return to step A; D. Clean and store the collected data; E. Perform machine learning diagnosis on the selected data and forward the result data after diagnosis; The machine learning diagnostic steps in step E include: EA. Determine the measuring point data to be diagnosed through the LCD display settings, and mark the associated measuring point data that affects the numerical value of the measuring point data to be diagnosed; EB. Read the database data in the storage module to obtain the historical data of the measurement point to be diagnosed and the associated measurement points, and determine the reference value of the data of the measurement point to be diagnosed under the corresponding numerical conditions of each associated measurement point; EC. Compare the current value of the data at the measurement point to be diagnosed with the reference value under the current working condition, determine whether the data value at the measurement point to be diagnosed is abnormal, and forward the diagnosis result data; Step E is specifically as follows: Perform machine learning diagnosis on the measurement point A to be diagnosed. The associated data B and the associated data C are the associated data of the measurement point A to be diagnosed. The values ​​of the associated data B and the associated data C will affect the value of the measurement point A to be diagnosed. The data values ​​(A, B, C) at this moment are regarded as one working condition. The measurement point A to be diagnosed is determined by the LCD display setting, and the associated data B and associated data C associated with the measurement point A to be diagnosed are selected. The edge computing module reads the database data in the storage module to obtain the historical data of the measurement point A to be diagnosed and the associated measurement points B and C, and determines the reference value of the measurement point A to be diagnosed under the corresponding numerical conditions of each associated measurement point. The steps for determining the reference value are as follows: Setting the Threshold , the values ​​of the measurement point A to be diagnosed under three typical working conditions are A1, A2, and A3, and the corresponding values ​​of the associated data B and the associated data C are (B1, C1), (B2, C2), and (B3, C3), respectively. Search the historical database for the values ​​of the associated data B and the associated data C in the range of (B1-ε≤B≤B1+ε, C1-ε≤C≤C1+ε), (B2-ε≤B≤B2+ε, C2-ε≤C≤C2+ε), and (B3-ε≤B≤B3+ε, C3-ε≤C≤C3+ε) corresponding to the measurement point A to be diagnosed. Take 100 times, and we have (A1 1 , A1 2 , A1 3 , ..., A1 99 , A1 100 ), (A2 1 , A2 2 , A2 3 , ..., A2 99 , A2 100 ), (A3 1 , A3 2 , A3 3 , ..., A3 99 , A3 100 ), then its average The following are the benchmark values ​​of the test point A to be diagnosed under three typical working conditions: ； Compare the current value of the data at the point to be diagnosed with the reference value under the current working condition and set the threshold , that is, when the current associated data B and associated data C values ​​are within the range of (B1-ε≤B≤B1+ε, C1-ε≤C≤C1+ε), (B2-ε≤B≤B2+ε, C2-ε≤C≤C2+ε), (B3-ε≤B≤B3+ε, C3-ε≤C≤C3+ε), if the current value of the measurement point A to be diagnosed exceeds , , range, it means that under the three typical working conditions, the A value of the measurement point to be diagnosed is abnormal, and the current A value of the measurement point to be diagnosed and the abnormal alarm identification value of the diagnosis result data are forwarded.

2. The method for automatic adaptation and fusion of power plant data specifications based on machine learning according to claim 1 is characterized in that: In step D, the data cleaning improves the quality of the collected data according to the steps of data review, missing value processing, outlier processing, data consistency check, and data verification; the data storage is to store the cleaned data into a database in the storage module.

3. The method for automatic adaptation and fusion of power plant data specifications based on machine learning according to claim 1 is characterized in that: The automatically parsed measuring point configuration file in step C includes measuring point name, measuring point type, measuring point coefficient, function code, register address, word length and offset parameters, and the above parameters are automatically configured.

4. The method for automatic adaptation and fusion of power plant data specifications based on machine learning according to claim 1 is characterized in that: The calling data protocol communication parameters in step C are specifically to automatically configure the port, baud rate, data bit, check mode, stop bit, interval time, byte stream mode, transmission delay, timeout, packet length, byte order and public address.

5. A device for automatic adaptation and fusion of power plant data specifications based on machine learning, characterized in that: The method according to any one of claims 1 to 4 is performed, wherein the device comprises: Edge computing module: used for real-time analysis, learning and data processing, network communication testing, parsing parameter information of measurement point configuration files, automatically calling data protocol communication parameters, data cleaning, data storage, machine learning diagnosis, reading database data in the storage module to obtain historical data of the measurement point to be diagnosed and related measurement points, calculating typical working condition benchmark values, comparing measurement point data with working condition benchmark values, and making abnormal judgments and data forwarding; Storage module: used to store programs and data, including real-time collected data and historical data; Power module: used to provide power to each part; Data interface group: used to connect various data devices or debugging equipment; LCD display: used for function settings; Indicator light group: used to display the status of each module and data interface in the fusion device; Box: used to place the edge computing module, storage module, power module, data interface, and LCD display.

6. The automatic adaptation and fusion device for power plant data protocol based on machine learning according to claim 5 is characterized in that: The liquid crystal display and indicator light group are arranged on the front side of the box; the edge computing module, storage module, and power supply module are arranged inside the box; and the data interface group is located on the back side of the box.

7. The device for automatic adaptation and fusion of power plant data specifications based on machine learning according to claim 5 is characterized in that: The liquid crystal display is a touch liquid crystal display, which performs parameter settings on the automatic adaptive fusion, data storage function, data forwarding function, and equipment configuration information function of the fusion device.

8. The device for automatic adaptation and fusion of power plant data specifications based on machine learning according to claim 5 is characterized in that: The data interface group includes an Ethernet interface, a serial interface, a switch input / output interface and an external display interface.

Images