Coal-fired power unit process data cleaning method, system, medium, equipment and terminal

Abstract

The present invention belongs to the field of data processing technology, and discloses a method, system, medium, equipment and terminal for cleaning process data of coal-fired power units. The method uses the SIS historical database interface to obtain historical data; performs singular value processing based on a segmented box plot; performs stuck value processing based on an empirical cumulative distribution function; and performs missing value processing based on spline interpolation. The present invention obtains effective process data of target parameters by obtaining historical data, performing singular value processing, stuck value processing and missing value processing, thereby achieving cleaning of the process data of coal-fired power units. The present invention targets the three error conditions of singular values, stuck values and missing values that occur when obtaining the operation process data of coal-fired power units from the SIS system of a power station, and designs a set of error data cleaning processes based on mathematical statistics methods and spline interpolation methods. The effective process data of the unit operation is obtained, the accuracy and reliability of the coal-fired power unit data are improved, and the problem of error values in the operation process data of the coal-fired power units is solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and in particular relates to a method, system, medium, equipment and terminal for cleaning process data of a coal-fired power unit. Background Art

[0002] Currently, in the actual operation of coal-fired power plants, tens of thousands of measurement sensors are deployed throughout various subsystems and equipment to capture operational status monitoring indicators such as vibration, temperature, pressure, and flow. These measurement devices operate under complex interference environments, such as high temperatures, high pressures, and electromagnetic interference. This can lead to hardware and software failures during data acquisition, such as communication and sensor failures, seriously impacting measurement accuracy and reliability. Errors in coal-fired power plant operational data often fail to reflect the health of the equipment and may even lead to misjudgment of status. The flexibility upgrades required for coal-fired power plants require the units to operate under more complex, variable operating conditions. Improving the quality management of operational data can help address the challenges posed by this more in-depth flexibility adjustment. Currently, the main methods used for industrial process data cleaning include those based on statistical theory, those based on fitting residuals, and those based on machine learning. Coal-fired power plant process data is characterized by high dimensionality, large data volumes, real-time generation, and variable operating conditions. In actual production, unreasonable singular values, stuck values with unchanging continuous values, and missing values with signal loss or storage errors may occur. Traditional global statistical theory methods can miss local erroneous data, while methods based on fitting residuals have difficulty detecting data stuckness. Machine learning-based methods require complex modulation for model training and updating, and lack unified guidance for different operating parameters. Therefore, there is an urgent need to design a new coal-fired power plant process data cleaning method and system.

[0003] Through the above analysis, the problems and defects of the existing technology are as follows:

[0004] (1) The long-term operation of existing coal-fired power unit measurement devices in a complex interference environment may cause software and hardware failures during the data acquisition process, seriously affecting the accuracy and reliability of the measurement.

[0005] (2) The error values in the existing coal-fired power unit operation process data often cannot represent the healthy status of the equipment operation and may even lead to misjudgment of the status.

[0006] (3) There is currently no complete process-based processing method for the characteristics of coal-fired power unit operating data and the type of error data. Summary of the Invention

[0007] In response to the problems existing in the prior art, the present invention provides a method, system, medium, equipment and terminal for cleaning process data of a coal-fired power unit.

[0008] The present invention is implemented as follows: a method for cleaning process data of a coal-fired power unit, the method comprising:

[0009] By acquiring historical data, processing singular values, stuck values and missing values in turn, the effective process data of the target parameters are obtained, and the process data of the coal-fired power unit is cleaned.

[0010] Furthermore, the coal-fired power unit process data cleaning method includes the following steps:

[0011] Step 1: Use the SIS historical database interface to obtain historical data;

[0012] Step 2: singular value processing based on segmented box plot;

[0013] Step 3: Perform hysteresis value processing based on the empirical cumulative distribution function;

[0014] Step 4: Process missing values based on spline interpolation.

[0015] Furthermore, the historical data acquisition in step 1 includes:

[0016] The SIS historical database interface is used to obtain the original data of the historical operation process of the target monitoring parameters. Based on the operation records, invalid data when the equipment is not running is eliminated to obtain the operation process data reflecting the health status of the equipment, which is expressed as several continuous time series and recorded as "operation data 1". The operation records of the equipment are obtained based on the operation switch quantity, 0 indicates shutdown and 1 indicates operation. For equipment without an operation switch quantity, the operation records are obtained when the monitored parameters reach a certain value. When the motor current is greater than 20A, it means that the equipment is running.

[0017] Furthermore, the singular value processing based on the segmented box plot in step 2 includes:

[0018] The target monitoring parameters are segmented in two ways, and the box plot is used to judge the singular values in different segmentation segments.

[0019] The singular value processing based on the segmented box plot specifically includes:

[0020] Based on "operation process data 1", the segmented box plot method is used to judge the singular values. The previous and next means are used to fill in the single singular values, and the continuous singular values are replaced with NaN, which is recorded as "operation process data 2". Among them, the data segmentation methods include: direct segmentation based on operation process data 1; segmentation of target monitoring parameters based on the distribution of operating condition parameters.

[0021] Furthermore, the hysteresis value processing based on the empirical cumulative distribution function in step 3 includes:

[0022] Based on "operation process data 2", the continuous constant value of the values in each subsequence is characterized to obtain the empirical distribution of the length of the continuous constant value of the target parameter in the historical process; according to the actual amount of data obtained, the empirical cumulative distribution threshold is specified to obtain the threshold of the length of the continuous constant value of the target parameter; the continuous segments exceeding the threshold are judged as stuck values and replaced with NaN. The result after the stuck value processing is recorded as "operation process data 3".

[0023] The hysteresis value processing based on the empirical cumulative distribution function specifically includes:

[0024] Characterize the continuous constant value situation in the historical process data of the target monitoring parameter; traverse the original data sequence in the order of sampling time, and the number of consecutive occurrences of the current value is the characterization value of the current continuous segment; traverse all subsequences to calculate the empirical cumulative distribution function of the continuous segment in the process data of the target monitoring parameter;

[0025]

[0026] in, is the empirical cumulative distribution function; I(·) is the indicator function; n is the number of samples; x represents any possible value, an integer greater than 0; a threshold α is set for the empirical cumulative distribution function to obtain a threshold L corresponding to the length of continuous constant values; when the representation value of the continuous segment exceeds the threshold L, it is considered that a stuck value has occurred.

[0027] Furthermore, the missing value processing based on spline interpolation in step 4 includes:

[0028] For single missing values, the mean of the previous and next values is used to fill the missing value; for continuous missing values, if the number of continuous missing values is less than the threshold N, cubic spline interpolation is used, otherwise it is directly deleted.

[0029] The missing value processing based on spline interpolation specifically includes:

[0030] When acquiring data, missing values are represented as NaN values. The NaN values at a single moment are filled with the mean of the previous and next moments. Continuous NaN values within the maximum allowable range are filled with spline interpolation, otherwise they are directly deleted. The effective process data of the target parameters are obtained, which are represented as several continuous time series.

[0031] Another object of the present invention is to provide a coal-fired power unit process data cleaning system using the coal-fired power unit process data cleaning method, the coal-fired power unit process data cleaning system comprising:

[0032] A historical data acquisition module is used to acquire historical data using the SIS historical database interface;

[0033] Singular value processing module, used for singular value processing based on segmented box plot;

[0034] A hysteresis value processing module is used to perform hysteresis value processing based on the empirical cumulative distribution function;

[0035] Missing value processing module, used to process missing values based on spline interpolation.

[0036] Another object of the present invention is to provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the coal-fired power unit process data cleaning method.

[0037] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to perform the steps of the coal-fired power unit process data cleaning method.

[0038] Another object of the present invention is to provide an information data processing terminal, which is used to implement the coal-fired power unit process data cleaning system.

[0039] In combination with the above technical solutions and the technical problems solved, please analyze the advantages and positive effects of the technical solutions to be protected by the present invention from the following aspects:

[0040] The coal-fired power unit process data cleaning method provided by the present invention divides the target monitoring parameters into segments in two ways, and uses box plots to judge singular values in different segmented segments. The present invention first characterizes the continuous constant value situation in the historical process data of the target monitoring parameter; traverses the original data sequence in sequence according to the sampling time, and the number of consecutive occurrences of the current value is the characterization value of the current continuous segment; after traversing all subsequences, the empirical cumulative distribution function of the continuous segment in the target monitoring parameter process data can be calculated. For a single missing value, the present invention uses the mean of the previous and next values to fill in; for continuous missing values, if the number of continuous missing values is less than the threshold N, cubic spline interpolation is used, otherwise it is directly deleted.

[0041] Aiming at the three error situations of singular values, stuck values and missing values that occur when obtaining the operation process data of coal-fired power units from the SIS system of a power plant, the present invention designs a set of error data cleaning processes based on mathematical statistics methods and spline interpolation methods to obtain effective process data of the unit operation.

[0042] The expected benefits and commercial value of the technical solution of the present invention after transformation are as follows: the present invention helps to improve the quality of power plant data, and high-quality data can generate valuable information, creating new added value for power plant management and operation production.

[0043] Does the technical solution of this invention solve a long-cherished but unsuccessful technical problem? Industrial processes often involve various uncertainties, which lead to more or less problematic data in industrial processes. The type of erroneous data varies in different application scenarios. This invention designs a data cleaning process for coal-fired power unit process data, focusing on cleaning singular values, stuck values, and missing values in the data, ultimately obtaining valid process data reflecting unit operation, which is of great significance for both unit operation monitoring and operation control. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0045] Figure 1 This is a flow chart of a method for cleaning process data of a coal-fired power plant provided by an embodiment of the present invention;

[0046] Figure 2 This is a schematic diagram of a method for cleaning process data of a coal-fired power plant provided by an embodiment of the present invention;

[0047] Figure 3 middle Figure 3 (a) is a schematic diagram of the flow rate change trend within the data acquisition time range. Figure 3 (b) is a schematic diagram of the bearing temperature variation trend within the data acquisition time range;

[0048] Figure 4 1 is a schematic diagram of the result after singular value cleaning provided by an embodiment of the present invention;

[0049] Figure 5 This is a schematic diagram of continuous fixed value representation provided by an embodiment of the present invention;

[0050] Figure 6 1 is a schematic diagram of the result of continuous constant value characterization of bearing temperature process data and its corresponding empirical cumulative distribution function provided by an embodiment of the present invention;

[0051] Figure 7 This is a data effect diagram after the hysteresis value is cleaned according to an embodiment of the present invention;

[0052] Figure 8This is a diagram showing the good accuracy of the cubic spline interpolation provided by the embodiment of the present invention;

[0053] Figure 9 This is a schematic diagram of the creative added value generated by power plant management and operation provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0055] In response to the problems existing in the prior art, the present invention provides a method, system, medium, equipment and terminal for cleaning process data of a coal-fired power unit. The present invention is described in detail below with reference to the accompanying drawings.

[0056] Based on a set of data cleaning processes, the present invention focuses on proposing a solution for the detection of stuck values.

[0057] like Figure 1 As shown, the coal-fired power unit process data cleaning method provided by the embodiment of the present invention includes the following steps:

[0058] S101, using the SIS historical database interface to obtain historical data;

[0059] S102, performing singular value processing based on the segmented box plot;

[0060] S103, performing hysteresis value processing based on the empirical cumulative distribution function;

[0061] S104: Process missing values based on spline interpolation.

[0062] As a preferred embodiment, Figure 2 As shown, the coal-fired power unit process data cleaning method provided by the embodiment of the present invention specifically includes the following steps:

[0063] (1) Obtaining historical data: Use the SIS historical database interface to obtain the original data of the target monitoring parameter during its historical operation. Based on the operation records, invalid data from equipment not in operation is eliminated to obtain the operation process data reflecting the health status of the equipment. This data is usually presented as several continuous time series, recorded as "operation data 1". The operation records of the equipment are generally obtained based on the operation switch value, 0 indicates shutdown and 1 indicates operation. For equipment without an operation switch value, the operation record can be obtained when the monitored parameter reaches a certain value. For example, a motor current greater than 20A indicates that the equipment is running.

[0064] (2) Singular value processing: Based on "Operation Process Data 1", this work uses the segmented box plot method to judge singular values. For single singular values, the previous and next means are used to fill them, and for continuous singular values, NaN is replaced, which is recorded as "Operation Process Data 2". There are two ways to segment the data: one is to directly segment based on "Operation Process Data 1"; the other is to segment the target monitoring parameters based on the distribution of the operating parameters.

[0065] (3) Stuck value processing: Based on the “operation process data 2”, this work characterizes the continuous constant value of the values in each subsequence and obtains the empirical distribution of the length of the continuous constant value of the target parameter in the historical process; according to the actual amount of data obtained, the corresponding empirical cumulative distribution threshold is specified to obtain the threshold of the length of the continuous constant value of the target parameter; the continuous segment exceeding the threshold is judged as a stuck value and replaced with NaN. The result after the stuck value processing is recorded as “operation process data 3”.

[0066] (4) Missing value processing: When acquiring data, missing values are generally also expressed as NaN values. For NaN values at a single moment, the mean of the previous and next moments is used to fill the gap to ensure the continuity of the sequence. For continuous NaN values within the maximum allowable range, spline interpolation is used to fill them. Otherwise, they are directly deleted. Finally, the effective process data of the target parameter is obtained, which is usually expressed as several continuous time series.

[0067] The coal-fired power unit process data cleaning system provided by the embodiment of the present invention includes:

[0068] A historical data acquisition module is used to acquire historical data using the SIS historical database interface;

[0069] Singular value processing module, used for singular value processing based on segmented box plot;

[0070] A hysteresis value processing module is used to perform hysteresis value processing based on the empirical cumulative distribution function;

[0071] Missing value processing module, used to process missing values based on spline interpolation.

[0072] This example uses the operating data of an induced draft fan at a power plant as an example. Bearing temperature is used as the target parameter, and fan flow rate is used as the operating parameter to clean the bearing temperature process data during operation. Historical data for bearing temperature and fan flow rate are obtained from the SIS system, with a 1-minute interval and a data collection period from January 1, 2021, to July 1, 2022.

[0073] The first step is to filter and obtain the bearing temperature and fan flow data during the operation of the induced draft fan based on the operating switch value of the induced draft fan, that is, when the equipment operating status is 1, such as Figure 3 shown. Figure 3 (a) is the changing trend of the flow rate within the data acquisition time range, Figure 3 (b) is the changing trend of the bearing temperature within the data acquisition time range. It can be clearly seen that the bearing temperature is lower when the equipment is just started than when the equipment is operating normally. At the same time, there is a clear hysteresis value in the historical process data of the bearing temperature.

[0074] In the second step, the operating parameters (flow in this embodiment) are divided into four intervals according to the quartiles, and the target parameters (bearing temperature in this embodiment) are processed for singular values using box plots in the corresponding four intervals. In this embodiment, the operating parameters are divided into four intervals: (, 2.66), (2.66, 3.21), (3.21, 4.17) and (4.17,). For a single singular value, the mean before and after are used to fill it, and for continuous singular values, NaN is used to replace it. The statistics of the singular values judged in the historical process data of the bearing temperature are shown in Table 1; the results after the singular value cleaning are shown in Table 1. Figure 4 As shown in the figure, the sample points identified as singular values in this example are primarily concentrated at the time the equipment is initially started, when the bearing temperature rises from ambient temperature to the normal operating temperature. This process does not reflect the true level of the bearing temperature during operation. Furthermore, equipment operating conditions vary widely, and the bearing temperature distribution varies at different output levels. Therefore, it is more reasonable to divide the historical process data of the target parameter and perform singular value processing based on the equipment operating conditions.

[0075] Table 1

[0076] Start time End Time Number of singular values 2021-03-02 09:36:00 2021-03-02 10:50:00 75 2021-03-02 10:53:00 2021-03-02 11:25:00 33 2021-03-13 17:33:00 2021-03-13 17:34:00 2 2021-03-20 15:40:00 2021-03-20 17:04:00 85 2021-03-21 07:43:00 2021-03-21 08:59:00 77 2021-05-05 12:44:00 2021-05-05 13:25:00 42 2021-06-30 10:47:00 2021-06-30 11:05:00 19

[0077] The third step is to further characterize the continuous fixed value situation in the cleaned data obtained in step 2. Traverse the original data sequence in the order of sampling time, and the number of times the current value appears continuously is the characterization value of the current continuous segment. The continuous fixed value characterization diagram is as follows Figure 5 As shown; after traversing all subsequences, the empirical cumulative distribution function of the continuous segment in the target monitoring parameter process data can be calculated. In this embodiment, the result of the continuous constant value representation of the bearing temperature process data and its corresponding empirical cumulative distribution function are as follows: Figure 6 As shown. Figure 6 It can be seen that the bearing temperature has obvious continuous constant values during operation, and the maximum continuous length even exceeds 25,000. In this embodiment, the threshold of the cumulative experience distribution is 0.9999, which corresponds to a maximum allowable continuous constant value length of 92. In other words, the continuous constant value length exceeding 92 in the historical process data of the bearing temperature is considered to be a stuck value and is replaced with NaN. The statistics of the bearing temperature identified as a stuck value are shown in Table 2. The data after the stuck value cleaning is shown in Table 2. Figure 7 As shown in the figure, due to the large amount of data, some hysteresis values are invisible to the naked eye.

[0078] Table 2

[0079] Start time End Time Number of stuck values 2021-01-26 21:28:00 2021-01-26 23:04:00 97 2021-01-29 09:36:00 2021-01-29 21:22:00 707 2021-04-20 13:24:00 2021-04-21 09:50:00 1227 2021-04-25 09:35:00 2021-04-25 17:54:00 500 2021-05-14 16:49:00 2021-05-14 18:24:00 96 2021-05-24 15:45:00 2021-05-24 18:50:00 186 2021-08-10 02:37:00 2021-08-28 23:59:00 27203 2021-09-15 20:50:00 2021-09-15 22:24:00 95 2022-01-24 02:19:00 2022-01-24 03:59:00 101 2022-03-04 10:12:00 2022-03-04 14:54:00 283 2022-03-23 16:44:00 2022-03-24 09:39:00 1016 2022-05-18 18:33:00 2022-05-19 21:37:00 1625 2022-05-22 02:00:00 2022-05-22 03:35:00 96 2022-05-25 15:04:00 2022-05-25 16:59:00 116 2022-06-08 00:00:00 2022-06-08 22:40:00 1361 2022-06-15 16:36:00 2022-06-15 18:11:00 96 2022-06-19 14:34:00 2022-06-19 16:07:00 94 2022-06-22 05:07:00 2022-06-22 06:44:00 98

[0080] The fourth step is to handle missing values. In step one, missing data that occurs when obtaining historical data from the SIS database is generally expressed as NaN values. In steps two and three, continuous singular values and stuck values are also replaced with NaN values. Therefore, in this step, missing values are determined by screening NaN values. In this embodiment, the cubic spline interpolation method is used to repair missing values to maintain the continuity of historical process data as much as possible. In this embodiment, the maximum threshold of the repair length is set to 5, that is, when the continuous missing values are greater than 5, the missing segment is directly deleted; when it is less than or equal to 5, it is repaired, and 5 points before and after the missing segment are used for cubic spline interpolation. Figure 8 As shown in the figure, a continuous original sequence is randomly selected from the historical process data of bearing temperature, and 5 consecutive points are replaced with missing values. The missing values are repaired by cubic spline interpolation. The mean square error between the repaired value and the true value is 0.0025. It can be seen that cubic spline interpolation has good accuracy.

[0081] The present invention aims to process singular values, stuck values and missing values that appear during the cleaning process of coal-fired power unit process data, and to ensure the continuity of data as much as possible. Among them, there are many mature technologies for processing singular values and missing values, such as methods based on mathematical statistics and methods based on machine learning; however, in industrial data processing, there is not much research on the processing of stuck values, and it is mainly judged based on experience. In an embodiment of the present invention, the bearing temperature is affected by the equipment operating conditions and lubricating oil temperature in the actual industrial process, and its change trend has the characteristics of slow change; with respect to the process data of the bearing temperature obtained in the SIS database, its historical process data shows the frequent occurrence of continuous constant values. This feature makes it difficult to judge how long the continuous constant value is stuck by human experience. Therefore, the present invention proposes a general method based on empirical cumulative distribution, which detects the stuck values in the operating process data of thousands of parameters of coal-fired power units, and combines it with singular value processing and missing value processing to propose a coal-fired power unit process data cleaning method, the purpose of which is to obtain effective and continuous process data of coal-fired power units as much as possible, and to generate creative added value for power plant management and operation. Figure 9 shown.

[0082] 1. Singular value processing based on segmented box plot

[0083] The current operation of coal-fired power units presents a frequently switched power generation mode under multiple load levels, which causes the process data trend of the unit to show the characteristics of alternating steady and unsteady states under multiple modes. The distribution of equipment target monitoring parameters may change with changes in unit load or equipment operating conditions; for example, when the induced draft fan flow rate of a certain unit is less than 2.6t / h, the temperature range of the fan front bearing is 38-62°C, and when the fan flow rate is greater than 7.3t / h, the temperature range of the fan front bearing is 51-64°C; therefore, it is inaccurate to simply treat the massive historical data of the target monitoring parameters as a sequence for singular value judgment. The present invention divides the target monitoring parameters into segments in two ways, and uses box plots to judge singular values in different divided segments.

[0084] 2. Hysteresis value processing based on empirical cumulative distribution function

[0085] The state monitoring parameters of coal-fired power units during operation generally change continuously, that is, the two values at adjacent moments are not equal; however, sometimes, due to the sensor sampling frequency being less than the data acquisition frequency or under accidental circumstances, the values at multiple consecutive moments may be equal. The present invention first characterizes the continuous constant value situation in the historical process data of the target monitoring parameter, as shown in Table 1; then traverses the original data sequence in order of sampling time, and the number of consecutive occurrences of the current value is the characterization value of the current continuous segment; after traversing all subsequences, the empirical cumulative distribution function of the continuous segment in the target monitoring parameter process data can be calculated, as shown in Formula (1).

[0086]

[0087] in, is the empirical cumulative distribution function; I(·) is the indicator function; n is the number of samples; and x represents any possible value, an integer greater than 0. By setting a threshold α on the empirical cumulative distribution function, we can obtain a threshold L corresponding to the length of continuous fixed values. When the representation value of a continuous segment exceeds threshold L, a stuck value is considered to have occurred.

[0088] Table 1 Characterization of continuous constant value in time series

[0089]

[0090] 3. Missing value processing based on spline interpolation

[0091] For single missing values, the mean of the previous and next values is used to fill the missing value; for continuous missing values, if the number of continuous missing values is less than the threshold N, cubic spline interpolation is used, otherwise it is directly deleted.

[0092] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware portion can be implemented using dedicated logic; the software portion can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. Those skilled in the art will understand that the above-mentioned devices and methods can be implemented using computer-executable instructions and / or contained in processor control code, for example, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, or can be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0093] The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with this technical field within the technical scope disclosed by the present invention and within the spirit and principles of the present invention should be covered by the scope of protection of the present invention.

Claims

1. A method for cleaning process data of a coal-fired power unit, characterized in that: The coal-fired power unit process data cleaning method includes: By acquiring historical data, processing singular values, stuck values, and missing values, the effective process data of the target parameters is obtained, and the process data of the coal-fired power unit is cleaned. Singular value processing based on segmented box plots includes: The target monitoring parameters are segmented in two ways, and the box plot is used to judge the singular values in different segmented segments; The singular value processing based on the segmented box plot specifically includes: Based on "Operation Data 1," a segmented box plot method was used to identify singular values. Single singular values were filled with the mean of the preceding and following values, while consecutive singular values were replaced with NaN. This data was recorded as "Operation Process Data 2." Data segmentation methods included: direct segmentation based on "Operation Process Data 1"; and segmentation of target monitoring parameters based on the distribution of operating condition parameters. Hysteresis value processing based on the empirical cumulative distribution function includes: Based on "Operational Process Data 2," characterize the continuous constant value of each subsequence and obtain the empirical distribution of the target parameter's historical continuous constant value length. Based on the actual amount of data obtained, specify the empirical cumulative distribution threshold to obtain the threshold for the target parameter's continuous constant value length. Continuous segments exceeding the threshold are identified as stuck values and replaced with NaN. The result after stuck value processing is recorded as "Operational Process Data 3." The hysteresis value processing based on the empirical cumulative distribution function specifically includes: Characterize the continuous constant value situation in the historical process data of the target monitoring parameter; traverse the original data sequence in the order of sampling time, and the number of consecutive occurrences of the current value is the characterization value of the current continuous segment; traverse all subsequences to calculate the empirical cumulative distribution function of the continuous segment in the process data of the target monitoring parameter; in, is the empirical cumulative distribution function; I(·) is the indicator function; n is the number of samples; x represents any possible value, an integer greater than 0; a threshold α is set for the empirical cumulative distribution function to obtain a threshold L corresponding to the length of continuous constant values; when the representation value of the continuous segment exceeds the threshold L, it is considered that a stuck value has occurred.

2. The coal-fired power unit process data cleaning method according to claim 1, characterized in that: The coal-fired power unit process data cleaning method comprises the following steps: Step 1: Use the SIS historical database interface to obtain historical data; Step 2: singular value processing based on segmented box plot; Step 3: Perform hysteresis value processing based on the empirical cumulative distribution function; Step 4: Process missing values based on spline interpolation.

3. The coal-fired power unit process data cleaning method according to claim 2, characterized in that: The historical data acquisition in step 1 includes: The SIS historical database interface is used to obtain the original data of the target monitoring parameter during its historical operation. Based on the operation records, invalid data from equipment not in operation is eliminated to obtain the operation process data reflecting the health status of the equipment. This data is expressed as several continuous time series and recorded as "operation data 1". The equipment operation record is obtained based on the operation switch value, where 0 indicates shutdown and 1 indicates operation. For equipment without an operation switch value, the operation record is obtained when the monitored parameter reaches a certain value. When the motor current is greater than 20A, the equipment is running.

4. The coal-fired power unit process data cleaning method according to claim 2, characterized in that: The missing value processing based on spline interpolation in step 4 includes: For single missing values, the mean of the previous and next values is used to fill in the missing values; for continuous missing values, if the number of continuous missing values is less than the threshold N, cubic spline interpolation is used, otherwise it is directly deleted; The missing value processing based on spline interpolation specifically includes: When acquiring data, missing values are represented as NaN values. The NaN values at a single moment are filled with the mean of the previous and next moments. Continuous NaN values within the maximum allowable range are filled with spline interpolation, otherwise they are directly deleted. The effective process data of the target parameters are obtained, which are represented as several continuous time series.

5. A coal-fired power unit process data cleaning system using the coal-fired power unit process data cleaning method according to any one of claims 1 to 4, characterized in that: The coal-fired power unit process data cleaning system includes: A historical data acquisition module is used to acquire historical data using the SIS historical database interface; Singular value processing module, used for singular value processing based on segmented box plot; A hysteresis value processing module is used to perform hysteresis value processing based on the empirical cumulative distribution function; Missing value processing module, used to process missing values based on spline interpolation.

6. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the coal-fired power unit process data cleaning method according to any one of claims 1 to 4.

7. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the method for cleaning process data of a coal-fired power plant according to any one of claims 1 to 4.

8. An information data processing terminal, characterized in that: The information data processing terminal is used to implement the coal-fired power unit process data cleaning system as described in claim 5.

Images