Data processing device and method for power plant

Abstract

The present disclosure provides a data processing device and method for a power plant, wherein the data processing device comprises: a data evaluation module configured to analyze power plant data to generate data quality evaluation information of the power plant data; and a data management module configured to manage the power plant data in response to the data quality evaluation information being generated. The data processing device and method are conducive to the evaluation and management of power plant data.

Description

Technical Field

[0001] The present disclosure relates to the field of power generation, and in particular to a data processing device and method for a power plant. Background Art

[0002] In the field of power generation (e.g., wind power generation and photovoltaic power generation), the data used is particularly important for the control of power generation equipment, power forecasting, etc. Therefore, there is an urgent need to improve the processing of data in the power generation field and overcome the problems of data processing in the power generation field, so as to improve the experience of data users in the power generation field and improve the control of power plants and many other aspects. Summary of the invention

[0003] The purpose of the embodiments of the present disclosure is to provide a data processing device and method for a power plant to overcome the deficiencies in the prior art.

[0004] According to an embodiment of the present disclosure, a data processing device for a power plant is provided, the data processing device comprising: a data evaluation module configured to analyze power plant data to generate data quality evaluation information of the power plant data; and a data governance module configured to govern the power plant data in response to generating the data quality evaluation information.

[0005] Optionally, the data evaluation module is configured to analyze the power plant data according to evaluation indicators, wherein the evaluation indicators include at least one of an integrity indicator, an accuracy indicator, a relevance indicator and a consistency indicator.

[0006] Optionally, the data quality assessment information includes at least one of integrity assessment information, accuracy assessment information, relevance assessment information and consistency assessment information, and the data assessment module is configured to perform at least one of the following operations: analyzing the missingness of the power plant data to generate integrity assessment information; analyzing at least one of the over-threshold, dead number, repetition and fluctuation of the power plant data to generate accuracy assessment information; analyzing the correlation between all fields in the power plant data to generate relevance assessment information; analyzing the internal consistency and / or temporal consistency of the power plant data to generate consistency assessment information.

[0007] Optionally, the data evaluation module is configured to score the quality of the power plant data according to an evaluation index, and the data quality evaluation information includes a data quality score corresponding to the evaluation index.

[0008] Optionally, the data evaluation module is configured to score the power plant data according to the accuracy index and / or the completeness index in the evaluation index, wherein the accuracy index includes at least one of an over-threshold index, a death count index, a repetition index and a fluctuation index, and the completeness index includes a missing index.

[0009] Optionally, the data evaluation module is configured to: count at least one of the over-threshold index, dead number index, duplication index, fluctuation index and missing index of each field in the power plant data to obtain the count value of the corresponding indicator of each field; generate a data quality score corresponding to the accuracy index and / or completeness index based on the total number of samples and the count value of the power plant data.

[0010] Optionally, the data evaluation module is configured to: apply weights to the count values ​​of the corresponding indicators of each field; and generate a data quality score corresponding to the accuracy indicator and / or the completeness indicator based on the total number of samples of the power plant data and the weighted count values.

[0011] Optionally, the data governance module is configured to: in response to generating data quality assessment information, determine the missing type and / or field type corresponding to the missing data in the power plant data; and fill the missing data in the power plant data according to the missing type and / or the field type.

[0012] Optionally, the data missing type includes a continuous missing type and a non-continuous missing type, and the data governance module is configured to use a linear interpolation method to fill the missing data of the non-continuous missing type, and / or the data governance module is configured to fill the missing data of the continuous missing type by at least one of the following operations: selecting non-missing data whose geographical location is within a predetermined range of the geographical location of the missing data from the power plant data to fill the missing data; selecting non-missing data corresponding to the missing data from another power plant data of the power plant to fill the missing data, wherein the data granularity of the other power plant data is smaller than the data granularity of the power plant data; selecting similar day data with the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and using the non-missing data at the time corresponding to the time of the missing data in the similar day data to fill the missing data; using the non-missing data at the time corresponding to the time of the missing data on the day before the missing data to fill the missing data; selecting non-missing data at the same time as the missing data from the power plant data of another power plant within a predetermined distance of the power plant to fill the missing data; using a machine learning algorithm to predict data for filling the missing data, and using the predicted data to fill the missing data.

[0013] Optionally, the field type includes a power type and a non-power type, and the data governance module is configured to fill in the missing data of the power type by at least one of the following operations: selecting non-missing data corresponding to the missing data from another power plant data of the power plant to fill in the missing data, wherein the data granularity of the other power plant data is smaller than the data granularity of the power plant data; selecting similar day data having the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and filling in the missing data with the non-missing data at a time corresponding to the time of the missing data in the similar day data; filling in the missing data with the non-missing data at a time corresponding to the time of the missing data on the day before the missing data; using a machine learning algorithm to predict data for filling in the missing data, and filling in the missing data with the predicted data, and / or, the data governance module is configured to fill in the missing data of the non-power type by at least one of the following operations: selecting non-missing data corresponding to the missing data from the other power plant data of the power plant to fill in the missing data, wherein the data granularity of the other power plant data is smaller than the data granularity of the power plant data; selecting similar day data having the highest similarity to the data of the day corresponding to the time of the missing data from the non-missing historical power plant data of the power plant, and filling in the missing data with the non-missing data at a time corresponding to the time of the missing data in the similar day data; filling in the missing data with the non-missing data at a time corresponding to the time of the missing data on the day before the missing data; using a machine learning algorithm to predict data for filling in the missing data, and filling in the missing data with the predicted data, and / or, the data governance module is configured to fill in the missing data of the non-power type by at least one of the following operations: selecting non-missing data corresponding to the missing data from the other power plant data of the power plant, and filling in the missing data with the predicted data. Select non-missing data whose geographical location is within a predetermined range of the geographical location of the missing data from the power plant data to fill the missing data; select non-missing data corresponding to the missing data from another power plant data of the power plant to fill the missing data, wherein the data granularity of the another power plant data is smaller than the data granularity of the power plant data; select similar day data with the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and use the non-missing data at the time corresponding to the time of the missing data in the similar day data to fill the missing data; use the non-missing data at the time corresponding to the time of the missing data on the day before the missing data to fill the missing data; select non-missing data at the same time as the missing data from the power plant data of another power plant within a predetermined distance from the power plant to fill the missing data; use a machine learning algorithm to predict data for filling the missing data, and use the predicted data to fill the missing data.

[0014] Optionally, the data evaluation module is configured to: analyze the treated power plant data to generate data quality evaluation information of the treated power plant data to compare with the data quality evaluation information of the power plant data before treatment.

[0015] Optionally, the data processing device also includes: a data testing module, configured to apply the power plant data before treatment and the power plant data after treatment to perform output power prediction, so as to generate predicted power before treatment and predicted power after treatment; and compare the predicted power before treatment and the predicted power after treatment with the actual power, so as to determine the prediction accuracy before treatment and the prediction accuracy after treatment.

[0016] Optionally, the data governance module is further configured to: determine whether the data quality assessment information meets a predetermined standard; if the data quality assessment information does not meet the predetermined standard, govern the power plant data.

[0017] Optionally, the power plant data includes: wind power generation data and / or photovoltaic power generation data. The fields of the wind power generation data include: at least one of the active power of wind power generation, the available power of wind power generation, the temperature of the wind measurement location, the wind speed of the wind measurement location, the wind direction of the wind measurement location, the humidity of the wind measurement location, and the air pressure of the wind measurement location. The fields of the photovoltaic power generation data include: at least one of the active power of photovoltaic power generation, the available power of photovoltaic power generation, the temperature of the radiometry measurement location, the direct radiometry of the radiometry measurement location, the total radiometry of the radiometry measurement location, the humidity of the radiometry measurement location, and the air pressure of the radiometry measurement location.

[0018] According to an embodiment of the present disclosure, a data processing method for a power plant is provided, the data processing method comprising: analyzing power plant data to generate data quality assessment information of the power plant data; and managing the power plant data in response to generating the data quality assessment information.

[0019] Optionally, the step of analyzing the power plant data includes: analyzing the power plant data according to an evaluation index, wherein the evaluation index includes at least one of an integrity index, an accuracy index, a relevance index and a consistency index.

[0020] Optionally, the data quality assessment information includes at least one of integrity assessment information, accuracy assessment information, correlation assessment information and consistency assessment information, and the step of analyzing the power plant data to generate data quality assessment information of the power plant data includes at least one of the following operations: analyzing the missing conditions of the power plant data to generate integrity assessment information; analyzing at least one of the over-threshold, dead number, repetition and fluctuation of the power plant data to generate accuracy assessment information; analyzing the correlation between all fields in the power plant data to generate correlation assessment information; analyzing the internal consistency and / or time consistency of the power plant data to generate consistency assessment information.

[0021] Optionally, the step of analyzing the power plant data to generate data quality assessment information of the power plant data includes: scoring the quality of the power plant data according to an assessment indicator, and the data quality assessment information includes a data quality score corresponding to the assessment indicator.

[0022] Optionally, the step of analyzing the power plant data to generate data quality assessment information of the power plant data includes: scoring the power plant data according to the accuracy index and / or the completeness index in the assessment index, wherein the accuracy index includes at least one of an over-threshold index, a dead number index, a repetition index and a fluctuation index, and the completeness index includes a missing index.

[0023] Optionally, the step of analyzing the power plant data to generate data quality assessment information of the power plant data includes: counting at least one of the over-threshold index, dead number index, duplication index, fluctuation index and missing index of each field in the power plant data to obtain the count value of the corresponding indicator of each field; generating a data quality score corresponding to the accuracy index and / or completeness index based on the total number of samples and the count value of the power plant data.

[0024] Optionally, the step of analyzing the power plant data to generate data quality assessment information of the power plant data includes: applying weights to the count values ​​of the corresponding indicators of each field; and generating a data quality score corresponding to the accuracy indicator and / or the completeness indicator based on the total number of samples of the power plant data and the weighted count values.

[0025] Optionally, the step of managing power plant data includes: in response to generating data quality assessment information, determining the missing type and / or field type corresponding to the missing data in the power plant data; and filling the missing data in the power plant data according to the missing type and / or the field type.

[0026] Optionally, the data missing type includes a continuous missing type and a non-continuous missing type, and the step of managing the power plant data includes using a linear interpolation method to fill the missing data of the non-continuous missing type, and / or the step of managing the power plant data includes filling the missing data of the continuous missing type by at least one of the following operations: selecting non-missing data whose geographical location is within a predetermined range of the geographical location of the missing data from the power plant data to fill the missing data; selecting non-missing data corresponding to the missing data from another power plant data of the power plant to fill the missing data, wherein the data granularity of the other power plant data is smaller than the data granularity of the power plant data; selecting similar day data with the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and filling the missing data with the non-missing data at the time corresponding to the time of the missing data in the similar day data; filling the missing data with the non-missing data at the time corresponding to the time of the missing data on the day before the missing data; selecting non-missing data at the same time as the missing data from the power plant data of another power plant within a predetermined distance of the power plant to fill the missing data; using a machine learning algorithm to predict data for filling the missing data, and filling the missing data with the predicted data.

[0027] Optionally, the field types include power types and non-power types, and the step of managing power plant data includes filling missing data of the power type through at least one of the following operations: selecting non-missing data corresponding to the missing data from another power plant data of the power plant to fill the missing data, wherein the data granularity of the other power plant data is smaller than the data granularity of the power plant data; selecting similar day data with the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and filling the missing data with the non-missing data at the moment corresponding to the moment of the missing data in the similar day data; filling the missing data with the non-missing data at the moment corresponding to the moment of the missing data on the day before the missing data; using a machine learning algorithm to predict data for filling the missing data, and filling the missing data with the predicted data. And / or, the step of managing power plant data includes filling missing data of non-power type by at least one of the following operations: selecting non-missing data whose geographical location is within a predetermined range of the geographical location of the missing data from the power plant data to fill the missing data; selecting non-missing data corresponding to the missing data from another power plant data of the power plant to fill the missing data, wherein the data granularity of the other power plant data is smaller than the data granularity of the power plant data; selecting similar day data with the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and filling the missing data with the non-missing data at the time corresponding to the time of the missing data in the similar day data; filling the missing data with the non-missing data at the time corresponding to the time of the missing data on the day before the missing data; selecting non-missing data at the same time as the missing data from the power plant data of another power plant within a predetermined distance of the power plant to fill the missing data; using a machine learning algorithm to predict data for filling the missing data, and filling the missing data with the predicted data.

[0028] Optionally, the data processing method further comprises: analyzing the treated power plant data to generate data quality assessment information of the treated power plant data to compare with the data quality assessment information of the power plant data before the treatment.

[0029] Optionally, the data processing method also includes: applying the power plant data before treatment and the power plant data after treatment to perform output power prediction, so as to generate the predicted power before treatment and the predicted power after treatment; comparing the predicted power before treatment and the predicted power after treatment with the actual power, respectively, to determine the prediction accuracy before treatment and the prediction accuracy after treatment.

[0030] Optionally, the step of managing the power plant data includes: determining whether the data quality assessment information meets a predetermined standard; and managing the power plant data if the data quality assessment information does not meet the predetermined standard.

[0031] Optionally, the power plant data includes: wind power generation data and / or photovoltaic power generation data, and the fields of the wind power generation data include: at least one of the active power of wind power generation, the available power of wind power generation, the temperature at the wind measuring location, the wind speed at the wind measuring location, the wind direction at the wind measuring location, the humidity at the wind measuring location, and the air pressure at the wind measuring location; the fields of the photovoltaic power generation data include: at least one of the active power of photovoltaic power generation, the available power of photovoltaic power generation, the temperature at the radiometry measurement location, the direct radiometry at the radiometry measurement location, the total radiometry at the radiometry measurement location, the humidity at the radiometry measurement location, and the air pressure at the radiometry measurement location.

[0032] According to an embodiment of the present disclosure, a computer-readable storage medium storing a computer program is provided. When the computer program is executed by a processor, the data processing method described above is implemented.

[0033] According to an embodiment of the present disclosure, a computing device is provided, the computing device comprising: a processor; a memory storing a computer program, and when the computer program is executed by the processor, the data processing method as described above is implemented.

[0034] By adopting the data processing equipment and method according to the embodiments of the present disclosure, at least the following technical effects can be achieved: through data quality evaluation and data screening, the power plant data becomes very friendly to subsequent data users, and data users no longer need to pay too much attention to the data itself, but only need to focus on the optimization of research and development. For example, the prediction model requires a large amount of historical data, and algorithm engineers no longer need to devote a large part of their energy to processing data; the evaluation standards and systems for data quality are improved, and at the same time a basis is provided for subsequent data governance or cleaning; an efficient and accurate data governance method is provided for the field of new energy power generation, which can standardize the application of data. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The above and other objects and features of the present disclosure will become more apparent from the following description in conjunction with the accompanying drawings.

[0036] Figure 1 is a schematic diagram of a data processing device of a power plant according to an embodiment of the present disclosure.

[0037] Figure 2 is a schematic diagram of a data processing method for a power plant according to an embodiment of the present disclosure.

[0038] Figure 3 is a schematic diagram of a data processing method for a power plant according to another embodiment of the present disclosure.

[0039] Figure 4is a comparative distribution diagram of integrity assessment information of power plant data of multiple power plants according to an embodiment of the present disclosure.

[0040] Figure 5 is a cumulative distribution graph of integrity assessment information of power plant data of a plurality of power plants according to an embodiment of the present disclosure.

[0041] Figure 6 Detailed description is provided for each distribution diagram of integrity assessment information of power plant data of a plurality of power plants according to an embodiment of the present disclosure.

[0042] Figure 7 and Figure 8 They are statistical distribution diagrams of the number of data points exceeding an upper threshold and a lower threshold in the power plant data of multiple power plants according to embodiments of the present disclosure.

[0043] Figures 9 to 11 They are statistical distribution diagrams of the number of dead points in power plant data of multiple power plants according to embodiments of the present disclosure.

[0044] Fig.12 is a statistical distribution diagram of the number of repeated data points in the power plant data of multiple power plants according to an embodiment of the present disclosure.

[0045] Figures 13 to 15 Schematic diagrams of correlation evaluation information between active power and total radiation in power plant data of multiple power plants according to embodiments of the present disclosure.

[0046] Fig.16 It is a comparative distribution diagram of integrity assessment information of power plant data after treatment of multiple power plants according to an embodiment of the present disclosure.

[0047] Figures 17 to 19 They are statistical distribution diagrams of the number of dead points in power plant data after treatment of multiple power plants according to embodiments of the present disclosure.

[0048] Fig. 20 It is a statistical distribution diagram of the number of repeated data points in the power plant data after treatment of multiple power plants according to an embodiment of the present disclosure.

[0049] Fig.21 It is a comparative statistical chart of data of a specific field in the power plant data before and after treatment according to an embodiment of the present disclosure.

[0050] Fig. 22 It is a comparative statistical chart of the prediction accuracy before and after the treatment according to the embodiment of the present disclosure.

[0051] Fig.23 is a schematic diagram of a computing device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0052] In the field of renewable energy power generation (e.g., wind power generation and photovoltaic power generation), the existing data quality evaluation process is usually simple and rough, and often not detailed enough. The existing technology also lacks a more complete data quality evaluation system. In particular, the data quality of data that will eventually be used for output power prediction is of great importance. Therefore, good data processing technology is needed to conduct a more comprehensive evaluation of data quality.

[0053] On the other hand, in order to make the data meet the basic standards for data application, data governance is needed so that the governed data can meet the basic standards before actual application, providing a better experience for data users (for example, power plant staff, developers of power prediction models).

[0054] The present disclosure proposes a data processing device and method for power plants, which are used to evaluate the quality of power plant data and manage power plant data. Standard data quality assessment is performed on time series data of the power generation industry (for example, the wind power generation industry and the photovoltaic power generation industry), and efforts are made to build a relatively complete data quality evaluation system. At the same time, for unqualified data, data management is performed using a method unique to the industry, so as to provide subsequent data users with higher quality, more effective and more reliable data.

[0055] Embodiments of the present disclosure will now be described more fully with reference to the accompanying drawings. Examples of the embodiments are shown in the accompanying drawings, wherein like reference numerals represent like parts throughout. Although some embodiments are shown in the accompanying drawings, the present disclosure is not limited thereto.

[0056] Figure 1 is a schematic diagram of a data processing device of a power plant according to an embodiment of the present disclosure.

[0057] like Figure 1 As shown, the data processing device 1 may include a data evaluation module 11 and a data management module 12. The data evaluation module 11 may be configured to analyze the power plant data to generate data quality evaluation information of the power plant data. The data management module 12 may be configured to manage the power plant data in response to generating the data quality evaluation information.

[0058] According to an embodiment of the present disclosure, the power plant may include a wind power plant and / or a photovoltaic power plant or similar new energy power plant. Accordingly, the power plant data may include: wind power generation data and / or photovoltaic power generation data or similar new energy power plant data.

[0059] Specifically, the power plant data may be composed of fields. For example, the fields of wind power generation data may include: at least one of the active power of wind power generation, the available power of wind power generation, the temperature of the wind measurement location, the wind speed of the wind measurement location, the wind direction of the wind measurement location, the humidity of the wind measurement location, and the air pressure of the wind measurement location. The fields of photovoltaic power generation data may include: at least one of the active power of photovoltaic power generation, the available power of photovoltaic power generation, the temperature of the radiometry measurement location, the direct radiometry of the radiometry measurement location, the total radiometry of the radiometry measurement location, the humidity of the radiometry measurement location, and the air pressure of the radiometry measurement location. The data processing device and method proposed in the present disclosure can analyze different fields of power plant data from multiple dimensions, and perform quality assessment on power plant data from multiple dimensions, thereby generating quality assessment information of power plant data, which is convenient for objectively and comprehensively assessing the quality of power plant data.

[0060] Optionally, the data processing device 1 may further include a data testing module 13. The data testing module 13 may be configured to respectively apply the power plant data before treatment and the power plant data after treatment to perform output power prediction to generate the predicted power before treatment and the predicted power after treatment; and compare the predicted power before treatment and the predicted power after treatment with the actual power to determine the prediction accuracy before treatment and the prediction accuracy after treatment. In this way, the degree of improvement of data quality by treating the power plant data can be obtained by comparing the prediction accuracy after treatment with the prediction accuracy before treatment.

[0061] Please refer to the following combination Figures 2 to 22 The specific details of the corresponding processing performed by the data processing device 1 and its various modules can be understood through the description, which will not be repeated here.

[0062] Figure 2 is a schematic diagram of a data processing method for a power plant according to an embodiment of the present disclosure.

[0063] In step S201, the power plant data may be analyzed. After the power plant data is analyzed, data quality assessment information of the power plant data may be generated accordingly (step S202). For example, the data assessment module 11 may analyze the power plant data to generate data quality assessment information.

[0064] According to an embodiment of the present disclosure, the operation of analyzing power plant data may include analyzing the power plant data according to an evaluation index, wherein the evaluation index may include at least one of an integrity index, an accuracy index, a correlation index, and a consistency index. According to an embodiment of the present disclosure, the integrity index may represent an index that is statistically calculated based on the degree of completeness or missingness of the data. The accuracy index may represent an index that is statistically calculated based on the normal and / or abnormal conditions of the data. The correlation index may represent an index that is statistically calculated based on the correlation between data of different fields at the same time. The consistency index may represent an index that is statistically calculated based on the consistency between data of different fields at the same time (i.e., internal consistency) and / or the consistency between adjacent moments of data of the same field (i.e., temporal consistency).

[0065] In this way, the power plant data can be analyzed in detail through multiple dimensions corresponding to the evaluation indicators, and the quality of the power plant data can be comprehensively and comprehensively evaluated.

[0066] According to one embodiment of the present disclosure, data quality assessment information may include at least one of integrity assessment information, accuracy assessment information, relevance assessment information, and consistency assessment information. The operation of analyzing power plant data to generate data quality assessment information of power plant data may include at least one of the following operations: analyzing the missing conditions of power plant data to generate integrity assessment information; analyzing at least one of the exceeding threshold, dead number, repetition, and fluctuation of power plant data to generate accuracy assessment information; analyzing the relevance between all fields in the power plant data to generate relevance assessment information; analyzing the internal consistency and / or time consistency of the power plant data to generate consistency assessment information.

[0067] The following description takes the data quality assessment information generated by analyzing the power plant data of multiple power plants as an example, but the present disclosure is not limited to the data quality assessment information shown in the example, and may also include other types or forms of data quality assessment information.

[0068] According to an embodiment of the present disclosure, the missing state of power plant data may be analyzed based on integrity indicators to generate integrity assessment information. Figures 4 to 6 Integrity assessment information of farm data of farms A, B, and C (ie, stations A, B, and C) is shown.

[0069] Figure 4 The power plant data may include power plant data of one or more power plants. Figure 4 The following is a statistical chart showing the percentage of days of existence of power plant data for power plants A, B and C (i.e., stations A, B and C) within 365 days of a certain year, where the horizontal axis represents the cumulative number of existence days and the vertical axis represents the power plant. Figure 4In the example shown, the data granularity of the power plant data corresponds to sampling the data once every 15 minutes, and the number of power plant data points in one day is 96. In the embodiment of the present disclosure, the number of data points may also be referred to as the number of data items.

[0070] like Figure 4 As shown, colors of different depths represent the number of points or point ranges of power plant data that exist in a day. If at least one of all fields of a power plant data exists and its value is not null (empty), the number of points of this power plant data is recorded as 1, indicating that there is a power plant data. If all fields of a power plant data do not exist or their values ​​are null, the number of points of this power plant data is recorded as 0, indicating that there is no power plant data. "96 points" means that there are 96 power plant data in the corresponding day, that is, the power plant data of this day is complete, and there is no missing or lost data. "80-85 points" means that there are 80 to 85 power plant data in the corresponding day, "55-80 points" means that there are 55 to 80 power plant data in the corresponding day, "1-55 points" means that there are 1 to 55 power plant data in the corresponding day, and "0 points" means that there is no power plant data in the corresponding day. Among the power field data of power plant A (i.e., station A) for one year, 325 days of power field data are complete, among the power field data of power plant B (i.e., station B) for one year, 320 days of power field data are complete, and among the power field data of power plant C (i.e., station C) for one year, 339 days of power field data are complete.

[0071] By comparison Figure 4 From the integrity assessment information of the power plant data of the three power plants, it can be concluded that power plant C has the best overall situation, with the most days with complete power plant data. Even if there are days with incomplete power plant data, they are very few; compared with power plant C, power plants A and B have fewer days with complete power plant data, and among the days with incomplete power plant data, the number of days with "80-95 points" accounts for a large proportion, and there are also days with "0 points" (that is, there is no power plant data at all).

[0072] According to the embodiments of the present disclosure, the power plant data of multiple power plants may also be analyzed on a daily basis to more deeply evaluate the integrity of the power plant data.

[0073] Figure 5 is a cumulative distribution diagram of integrity assessment information of power plant data of multiple power plants according to an embodiment of the present disclosure. Figure 5 As shown, the horizontal axis represents the date, and the vertical axis represents the number of stations with complete power plant data (i.e., the number of power plants). Here, the integrity assessment information of power plant data from July 1, 2018 to June 30, 2019 is taken as an example, but the present disclosure is not limited to this.

[0074] exist Figure 5 In the figure, the light-colored part represents the projection of the line connecting the number of stations with no missing data (i.e., 96 power plant data) on the horizontal axis, and the dark-colored part represents the projection of the line connecting the number of stations with no missing data and all fields of the existing power plant data on the horizontal axis (i.e., all fields of the 96 power plant data are non-null). The integrity assessment information represented by the dark-colored part is more stringent than the integrity assessment information represented by the light-colored part.

[0075] pass Figure 5 The integrity assessment information shown shows that the power plant data from the end of November 2018 to the end of February 2019 is relatively complete compared to other periods, with very little data missing. Starting from March 2019, the data missing is more serious overall, with only one power plant having relatively complete data. In June 2019, there were even three power plants with strict missing data (the value of at least one field in each of the 96 power plant data was null).

[0076] According to the embodiments of the present disclosure, the power plant data of each power plant may also be analyzed one by one to more deeply evaluate the integrity of the power plant data.

[0077] Figure 6 It is a distribution diagram of integrity assessment information of power plant data of multiple power plants according to an embodiment of the present disclosure. Wherein, the horizontal axis represents time (for example, the time point on the horizontal axis is set according to the data granularity of the power plant data (for example, corresponding to sampling once every 15 minutes) (for example, a time point is set every 15 minutes)), the value 1 on the vertical axis indicates that the power plant data exists (for example, all fields in all fields of the power plant data exist and their values ​​are not null), and the value 0 on the vertical axis indicates that the power plant data does not exist. Figure 6 The non-white area in represents the projection area on the abscissa of the value representing the existence of the power plant data.

[0078] Figure 6 The integrity assessment information indicating the existence of the power plant data for each power plant is shown, and the Figure 5 , it can be found that the power plant data missing during March 2019 are power plants B and C.

[0079] According to an embodiment of the present disclosure, at least one of the over-threshold, dead number, repetition and fluctuation of the power plant data can be analyzed according to the accuracy index to generate accuracy evaluation information. The over-threshold, dead number, repetition, fluctuation and other conditions in the power plant data can be used to determine whether the power plant data is abnormal, which helps to measure the accuracy of the power plant data. Therefore, the accuracy of the power plant data can be evaluated from different dimensions such as over-threshold, dead number, repetition, fluctuation and so on.

[0080] The over-threshold situation may indicate that there is over-threshold data in the power plant data that exceeds the normal threshold range. For example, if the active power in a power plant data (indicating the actual power generation power of the wind turbines or photovoltaic generators in the entire power plant) exceeds the normal threshold range (for example, the active power is higher than the preset upper limit threshold of the active power or lower than the preset lower limit threshold of the active power), then this power plant data is identified as over-threshold data.

[0081] The dead number situation may indicate that there are dead number data in the power plant data where the data of the same field at different times are repeated continuously. For example, if the active power of multiple consecutive (for example, 4 or other numbers of) power plant data is repeated, the power plant data with these consecutively repeated active powers are identified as dead number data. For the dead number situation in the photovoltaic power plant data, since the radiation before sunrise and after sunset is 0, the photovoltaic power plant data before sunrise and after sunset can be excluded when identifying the dead number data in the photovoltaic power plant data.

[0082] The repetition situation may indicate that there is a complete repetition of the entire data in the power plant data (i.e., the data of all fields at different times are repeated (for example, it may include continuous repetition or discontinuous repetition)). For example, if two or more power plant data are repeated, these repeated power plant data are identified as repeated data.

[0083] The fluctuation situation may indicate that there is fluctuation data with abnormal fluctuation (e.g., abnormal standard deviation) in the power plant data. For example, the standard deviation of the active power in the power plant data may be used to identify the active power with abnormal fluctuation. Specifically, for example, if the standard deviation between the active power in the power plant data at a certain moment and the active power in the power plant data at a previous moment is greater than a preset standard deviation threshold, the power plant data at the certain moment is identified as fluctuation data. Among them, the standard deviation threshold may be set according to the overall change of the active power in the power plant data within a specific time period (e.g., within a corresponding day). For example, the standard deviation threshold is three times or other multiples of the standard deviation of the active power within the corresponding day.

[0084] Figures 7 to 12 Accuracy assessment information of power plant data of power plants A, B, and C (ie, stations A, B, and C) is shown.

[0085] Figure 7 and Figure 8 1 and 2 are statistical distribution diagrams of the number of data points exceeding the upper threshold and the number of data points exceeding the lower threshold in the power plant data of multiple power plants according to an embodiment of the present disclosure. The accuracy evaluation information of the power plant data of power plants A, B and C (i.e., stations A, B and C) from July 2018 to June 2019 is used as an example for illustration, but the present disclosure is not limited thereto. Figure 7 and Figure 8 The horizontal axis is the date. Figure 7 The vertical axis in is the number of data points that exceed the upper threshold (that is, higher than the upper threshold). Figure 8 The vertical axis in is the number of data points that exceed the lower threshold (that is, lower than the upper threshold). Figure 7 and Figure 8 In the illustrated embodiment, the upper threshold may include an upper threshold set separately for all fields, and the lower threshold may include a lower threshold set separately for all fields. Accordingly, data exceeding the upper threshold indicates that the data exceeds the upper threshold of the corresponding field, and data exceeding the lower threshold indicates that the data exceeds the lower threshold of the corresponding field.

[0086] from Figure 7 and Figure 8 It can be seen that: the data exceeding the upper threshold in the power field data of station C were relatively large before October 2018, and almost none after October 2018; the data exceeding the upper threshold in the power field data of stations A and B basically did not exceed the upper threshold; the data below the lower threshold in the power field data of station B was relatively large, followed by the power field data of station C, and finally the power field data of station A; On the whole, station B had the more serious errors such as exceeding the threshold, followed by station C, and finally station A.

[0087] On the other hand, there will be death counts in the power plant data of almost every power plant. For the convenience of observation, the death counts of each power plant can be analyzed one by one.

[0088] Figures 9 to 11 1 and 2 are statistical distribution diagrams of the number of dead points in the power plant data of power plants A, B and C (i.e., stations A, B and C) according to embodiments of the present disclosure. Figures 9 to 11 In the figure, the horizontal axis represents the date, and the vertical axis represents the number of death data points (which can be simply referred to as death points).

[0089] exist Figures 9 to 11 In the embodiment shown, power plants A, B and C are all photovoltaic power plants. Figures 9 to 11Only examples of counting dead points for active power, ambient temperature, total radiance (which may be referred to as total radiation) and direct radiance (which may be referred to as direct radiation) are shown, but the present disclosure is not limited to this. Specifically, active power may represent the actual power generation of the entire power plant, ambient temperature may represent the temperature of the radiance measurement position, direct radiance may represent the direct radiance of the radiance measurement position, and total radiance may represent the total radiance of the radiance measurement position. Each power plant includes a radiance measurement position. For example, the ambient temperature, total radiance and direct radiance may be obtained by an environmental detector in the power plant.

[0090] from Figures 9 to 11 The figure shows that the death rate situation of the three stations is generally serious, but the distribution of death rate data in different categories or fields of each station is different. For example, for station A, the death rate situation is more serious in active power; for station B, the death rate situation is more serious in ambient temperature, and the death rate situation of active power on some dates is more serious; for station C, the death rate situation is more serious in active power, direct radiation and total radiation.

[0091] On the other hand, the repetition situation of each power plant can be analyzed one by one. Fig.12 is a statistical distribution diagram of the number of repeated data points in the power plant data of multiple power plants according to an embodiment of the present disclosure. Fig.12 In the figure, the horizontal axis represents the date and the vertical axis represents the number of repeated data points.

[0092] from Fig.12 It can be seen that the duplication situation at station B is the most serious, followed by station C, while there is almost no duplication at station A.

[0093] Reference Figures 7 to 12 , the error conditions (e.g., over-threshold conditions, dead count conditions, and duplication conditions) in the power plant data of multiple power plants can be comprehensively analyzed. According to the accuracy evaluation information of the three stations, it can be found that stations B and C have more errors, and station A has fewer errors. Relatively speaking, the quality of the power plant data of station A is the highest.

[0094] According to an embodiment of the present disclosure, the power plant data can be analyzed according to the correlation index to generate correlation evaluation information. The correlation between the data of different fields at the same time in the power plant data can be analyzed to determine whether the power plant data is logical. For example, the correlation between the wind speed, wind direction, temperature, humidity at the wind measurement location, the average wind speed of a single power generation device, the active power of a single power generation device, the active power or available power of the entire power plant, the total radiance or direct radiance at the radiance measurement location, and other fields can be analyzed.

[0095] Specifically, at least one of the correlations between the data in the following fields can be analyzed: the correlation between the wind speed at the hub height of the wind tower and the average wind speed of a single power generation device (applicable to wind farms); the correlation between the wind speed at the hub height of the wind tower and the active power of the entire field (applicable to wind farms); the correlation between the average wind speed of a single power generation device and the active power of the entire field (applicable to wind farms); the correlation between the total radiation of the environmental detector and the active power of the entire field (applicable to photovoltaic power fields); the correlation between the active power of a single power generation device and the active power of the entire field (applicable to wind farms); the correlation between the active power of a single power generation device and the theoretical power of the entire field (applicable to wind farms); the correlation between the active power of a single power generation device and the available power of the entire field (applicable to wind farms); the correlation between the active power of a reference station and the active power of the entire field (applicable to photovoltaic power fields). Among them, active power can represent actual power generation, available power can represent theoretical power generation minus shutdown capacity, and shutdown capacity represents the power loss caused by shutdown of power generation equipment due to failure or maintenance.

[0096] Figures 13 to 15 Schematic diagrams of correlation evaluation information between active power and total radiation in power plant data of multiple power plants according to embodiments of the present disclosure.

[0097] exist Figures 13 to 15 In the illustrated embodiment, the power generation fields A, B and C are all photovoltaic power generation fields, the abscissa is the total radiation (may be referred to as total radiation for short), and the ordinate is the active power. Figures 13 to 15 The scatter plot of the correlation between the active power and total radiation of each power plant is displayed. It can be seen that the correlation between the total radiation and active power of each station is very strong. Except for the situation of high radiation and low power in individual data points (which may be caused by power rationing and other reasons), the data are basically normal.

[0098] Based on the above-mentioned integrity assessment information, accuracy assessment information and correlation assessment information, the correlation of the power field data of power plants A, B and C is good. The number of existence points of the power field data of power plant C is relatively large, but the accuracy is not high. Although the number of existence points of the power field data of power plants A and B is less than that of power plant C, the accuracy of the power field data of power plants A and B is relatively good, especially the accuracy of the power field data of power plant A is relatively the best.

[0099] According to an embodiment of the present disclosure, power plant data can be analyzed according to consistency indicators to generate consistency evaluation information. The consistency indicators may include internal consistency indicators and / or time consistency indicators. The internal consistency indicators can be used to evaluate the consistency between data of different fields at the same time, and the time consistency indicators can be used to evaluate the consistency between adjacent moments of data of the same field. For example, the wind speed, wind direction, temperature, humidity at the wind measurement location, average wind speed of a single power generation device, active power of a single power generation device, active power or available power of the entire power plant, temperature at the radiometry measurement location, humidity at the radiometry measurement location, total radiometry or direct radiometry at the radiometry measurement location, etc., and / or the time consistency of each field can be analyzed.

[0100] Specifically, at least one of the internal consistencies between the data in the following fields can be analyzed: the consistency between the wind speeds at each floor height of the wind tower (applicable to wind farms); the consistency between the scattered radiation of the environmental monitor and the horizontal total radiation (applicable to photovoltaic power plants); the consistency between the direct radiation of the environmental monitor and the horizontal total radiation (applicable to photovoltaic power plants).

[0101] In addition, at least one of the temporal consistency between the data of the following fields can also be analyzed: wind speed of the wind tower (applicable to wind farms); wind direction of the wind tower (applicable to wind farms); temperature of the wind tower (applicable to wind farms); humidity of the wind tower (applicable to wind farms); air pressure of the wind tower (applicable to wind farms); direct radiation of the environmental detector (applicable to photovoltaic power farms); diffuse radiation of the environmental detector (applicable to photovoltaic power farms); total horizontal radiation of the environmental detector (applicable to photovoltaic power farms); active power of the entire field (applicable to any power farm, such as wind power farms and photovoltaic power farms); available power of the entire field (applicable to any power farm, such as wind power farms and photovoltaic power farms).

[0102] As described above, the power plant data may be analyzed from the consistency dimension, so that the quality of the data of each field in the consistency dimension may be analyzed in detail.

[0103] In addition to the above-mentioned intuitive and visual evaluation of the quality of power plant data in the form of statistical charts, the quality of power plant data can also be evaluated in a quantitative manner.

[0104] According to an embodiment of the present disclosure, the operation of analyzing the power plant data to generate data quality assessment information of the power plant data may include scoring the quality of the power plant data according to the assessment index, and the data quality assessment information includes a data quality score corresponding to the assessment index. In this way, the power plant data can be quantitatively assessed, which is convenient for objectively assessing the data quality and is conducive to completing the entire data quality assessment system.

[0105] According to an embodiment of the present disclosure, the power plant data can be scored according to the accuracy index and / or the completeness index in the evaluation index, wherein the accuracy index may include at least one of the indicators such as the over-threshold index, the dead number index, the repetition index and the fluctuation index, and the completeness index includes the missing index. Therefore, the data quality score can be calculated for each field of the power plant data and / or each indicator used for evaluation, and the data quality can be quantified comprehensively and comprehensively.

[0106] For example, at least one of the indicators such as the over-threshold index, the dead number index, the repetition index, the fluctuation index and the missing index of each field in the power plant data can be counted to obtain the count value of the corresponding indicator of each field; according to the total number of samples and the count value of the power plant data, a data quality score corresponding to the accuracy index and / or the completeness index is generated. Here, the evaluation of the over-threshold index, the dead number index, the repetition index, the fluctuation index and the missing index for each field is taken as an example, but the present disclosure is not limited to this.

[0107] When calculating the data quality score, a score reduction process may be performed for at least one of the over-threshold index, the dead number index, the duplicate index, the fluctuation index, and the missing index, for example, the count value of at least one of the over-threshold index, the dead number index, the duplicate index, the fluctuation index, and the missing index is subtracted. Then, the data quality score may be calculated by dividing the count value by the total number of samples.

[0108] In addition, weights may be applied to the count values ​​of the corresponding indicators of each field; and a data quality score corresponding to the accuracy indicator and / or the completeness indicator may be generated based on the total number of samples of the power plant data and the weighted count values.

[0109] Specifically, the following formula can be set to calculate the data quality score:

[0110]

[0111]

[0112] score=score1×score2 (3)

[0113] Among them, score1 represents the accuracy score, score2 represents the completeness score, and score represents the final data quality score. The subscript i represents the number of the field. For example, the number of active power is 1, the number of temperature is 2, and m represents the maximum value among the numbers of all fields, so that each field in the power plant data can be traversed to calculate the score. The subscript j represents the number of the indicator. For example, the number of the over-threshold indicator is 1, the number of the dead number indicator is 2, and the number of the fluctuation indicator is 3. n represents the maximum value among the numbers of indicators except for repeated indicators and missing indicators. In addition, the subscripts "missing" and "repeated" represent missing indicators and repeated indicators, respectively. In this way, each indicator in the power plant data can be traversed to calculate the score.

[0114] W represents the weight of a field or indicator. For example, Wi is the weight of the i-th field, Wj is the weight of the j-th indicator, and W 重复 is the weight of the repeated indicator.

[0115] C represents the count value of the field under the indicator, for example, C ij represents the count value of the jth index of the i-th field, C i缺失 represents the count value of the missing indicator of the i-th field, C 重复 Indicates the count value of repeated data in the power plant data.

[0116] S i represents the total number of samples in the i-th field, S 总体 Represents the total number of samples of power plant data, the total number of samples of the i-th field S i It can be equal to the total number of samples S of the power plant data 总体 For example, taking the power plant data sampled every 15 minutes, the total number of samples of the power plant data for one day can be 96, so the quality of the power plant data for that day can be analyzed. In addition, the quality of power plant data in other time periods (for example, one week, one month, half a year, or one year, etc.) can also be analyzed to effectively screen data with high quality scores.

[0117] In the embodiments of the present disclosure, the weight of each indicator and / or field can be set according to the degree of influence of each indicator and / or each field on the data quality score, and the weight of each indicator and / or each field can be independent of each other and do not affect each other. If new fields or new indicators are added later, the set weights do not need to be changed, and the weights only need to be assigned according to the importance of the new fields or new indicators.

[0118] According to the above formulas (1) to (3), the data quality scores of the power plant data of one or more power plants can be calculated. According to the data quality scores, at least the integrity and accuracy of the power plant data can be effectively evaluated, so that the quality of the data can be judged according to the quality scores.

[0119] Reference above Figures 4 to 15 The operation of analyzing power plant data and generating data quality assessment information is described. Figure 2 In step S203, in response to generating the data quality assessment information, the power plant data may be governed.

[0120] According to an embodiment of the present disclosure, step S203 may include: in response to generating data quality assessment information, determining the missing type and / or field type corresponding to the missing data in the power plant data; and filling the missing data in the power plant data according to the missing type and / or the field type.

[0121] According to an embodiment of the present disclosure, the data missing type may include a continuous missing type and a non-continuous missing type, and the field type may include a power type and a non-power type.

[0122] Specifically, the power plant data before treatment can be constructed into complete time series data, and the value of the field corresponding to the missing data is set to None (indicating no value). Then, according to the missing type and / or field type corresponding to the missing data, a suitable operation is selected to fill or fill the data. For example, all data with None values ​​can be extracted, and whether the missing data belongs to the continuous missing type or the discontinuous missing type can be determined based on whether the number of missing points of the None value data is continuous.

[0123] The operation of remediating the power plant data may include using a linear interpolation method to fill in the missing data of the non-continuous missing type. For example, a linear interpolation method of a time series may be used to fill in the missing data of the non-continuous missing type.

[0124] Optionally, the operation of managing power plant data may include filling missing data of a continuous missing type by at least one of the following operations: selecting non-missing data whose geographical location is within a predetermined range of the geographical location of the missing data from the power plant data to fill the missing data; selecting non-missing data corresponding to the missing data from another power plant data of the power plant to fill the missing data, wherein the data granularity of the other power plant data is smaller than the data granularity of the power plant data; selecting similar day data with the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and filling the missing data with the non-missing data at the time corresponding to the time of the missing data in the similar day data; filling the missing data with the non-missing data at the time corresponding to the time of the missing data on the day before the missing data; selecting non-missing data at the same time as the missing data from the power plant data of another power plant within a predetermined distance of the power plant to fill the missing data; using a machine learning algorithm to predict data for filling the missing data, and filling the missing data with the predicted data.

[0125] For example, the wind speed data of multiple storeys measured by the wind tower can be used to fill in the missing data, and the geographical location used to fill in the missing data should be within the predetermined range of the geographical location of the missing data. For example, when the wind speed data at a height of 10 meters is missing, the wind speed data at a height of 30 meters or 50 meters within the predetermined range (height difference range of 20 meters) can be selected for filling, and the wind speed data at a height of 30 meters can be preferentially selected, but the wind speed data at a height of 70 meters cannot be used. The above data filling method is particularly suitable for power plant data.

[0126] Optionally, non-missing data corresponding to the missing data can be selected from another power plant data of the power plant to fill the missing data, the data granularity of the other power plant data corresponds to sampling once every first time period (for example, 5 minutes), and the data granularity of the power plant data to be filled corresponds to sampling once every second time period (for example, 15 minutes), wherein the first time period is smaller than the second time period.

[0127] Optionally, non-missing data at the same time as the missing data may be selected from the power plant data of another power plant within a predetermined distance of the power plant to fill the missing data. For example, multiple power plants within a predetermined distance of the power plant may be determined, and the correlation between the data of each field of the power plant data of the multiple power plants and the data of the corresponding field of the power plant data to be filled may be calculated (for example, the correlation is calculated every day, month, or year). When the power plant data of the power plant is missing, the data of the field with the greatest correlation with the field corresponding to the missing data may be selected from the power plant data of the multiple power plants to fill the missing data. In addition, the data granularity of the power plant may be equal to the data granularity of the other power plant. The above data filling method is particularly suitable for power plant data.

[0128] According to an embodiment of the present disclosure, the field type may include a power type and a non-power type. Accordingly, step S203 may include filling in missing data of the power type (e.g., active power, available power, power generation, etc. of the power plant) by at least one of the following operations: selecting non-missing data corresponding to the missing data from another power plant data of the power plant to fill in the missing data, wherein the data granularity of the other power plant data is smaller than the data granularity of the power plant data; selecting similar day data with the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and filling in the missing data using non-missing data at the moment corresponding to the moment of the missing data in the similar day data; filling in the missing data using non-missing data at the moment corresponding to the moment of the missing data on the day before the missing data; using a machine learning algorithm to predict data for filling in the missing data, and filling in the missing data using the predicted data.

[0129] Optionally, step S203 may include filling in missing data of non-power type (for example, various meteorological data such as wind direction, wind speed, temperature, humidity, atmospheric pressure, total radiation, direct radiation, scattered radiation, etc.) by at least one of the following operations: selecting non-missing data whose geographical location is within a predetermined range of the geographical location of the missing data from the power plant data to fill in the missing data; selecting non-missing data corresponding to the missing data from another power plant data of the power plant to fill in the missing data, wherein the data granularity of the other power plant data is smaller than the data granularity of the power plant data; selecting similar day data with the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and using the non-missing data at the time corresponding to the time of the missing data in the similar day data to fill in the missing data; using the non-missing data at the time corresponding to the time of the missing data on the day before the missing data to fill in the missing data; selecting non-missing data at the same time as the missing data from the power plant data of another power plant within a predetermined distance of the power plant to fill in the missing data; using a machine learning algorithm to predict data for filling in the missing data, and using the predicted data to fill in the missing data.

[0130] The specific data filling method has been described above and will not be repeated here. As described above, by managing the power plant data, at least one of the characteristics of the power plant data, such as integrity, accuracy, consistency and relevance, can be effectively improved. In addition, according to the data quality assessment information of the power plant data, the aspects of the power plant data with poor quality can be effectively managed, which is conducive to accurately and efficiently managing the power plant data.

[0131] After completing the governance of the power plant data, the governed power plant data can be analyzed to generate data quality assessment information of the governed power plant data, which can be compared with the data quality assessment information of the power plant data before governance to understand the data changes before and after governance, and determine the degree and specific characteristics of the improvement in power plant data quality through data governance.

[0132] Therefore, the data processing method may also include: applying the power plant data before treatment and the power plant data after treatment to predict the output power, so as to generate the predicted power before treatment and the predicted power after treatment; comparing the predicted power before treatment and the predicted power after treatment with the actual power, so as to determine the prediction accuracy before treatment and the prediction accuracy after treatment.

[0133] Regarding the timing of regulating power plant data, it is also possible to choose to execute only when the data quality assessment information does not meet the predetermined standards. Figure 3is a schematic diagram of a data processing method for a power plant according to another embodiment of the present disclosure.

[0134] like Figure 3 As shown, in step S301, the power plant data may be analyzed. In step S302, data quality assessment information of the power plant data may be generated. The implementation of these two steps has been described above and will not be repeated here.

[0135] In step S303, it may be determined whether the data quality assessment information meets a predetermined standard. If the data quality assessment information does not meet the predetermined standard, the power plant data is managed (step S304). If the data quality assessment information meets the predetermined standard, the power plant data may not be managed (step S305).

[0136] According to an embodiment of the present disclosure, after the governance of the power plant data is completed, the governed power plant data can be analyzed to generate data quality assessment information of the governed power plant data to be compared with the data quality assessment information of the power plant data before the governance. In addition, it can be determined again whether the data quality assessment information of the governed power plant data meets the predetermined standard, that is, steps S303 to S304 or S305 are performed again.

[0137] In an embodiment of the present disclosure, the predetermined standard may be set according to the data usage standard of the power prediction model. Optionally, if the data quality assessment information includes a data quality score, the predetermined standard may include a standard score value. Therefore, when the data quality score is lower than the standard score value, the power plant data may be governed. In this way, it is possible to reasonably select whether the power plant data needs to be governed according to the set predetermined standard, which helps to improve production efficiency.

[0138] After completing the data governance, the governed power plant data can be analyzed again, that is, the above data governance method is repeatedly executed.

[0139] Fig.16 It is a comparative distribution diagram of integrity assessment information of power plant data after treatment of multiple power plants according to an embodiment of the present disclosure. Fig.16 Shows Figure 4 The example shown is a statistical chart of the percentage of days of existence of the power plant data after treatment for power plants A, B and C (i.e., stations A, B and C) within 365 days of a certain year, where the horizontal axis represents the cumulative number of days and the vertical axis represents the power plant. Fig.16 In the use of Figure 4 The plotting method shown is used to present the integrity information of the treated power plant data. Fig.16 and Figure 4 By comparison, it can be found that the power plant data of power plants A, B and C after treatment have been fully supplemented, and the power plant data for each day of the year are complete.

[0140] In addition, accuracy assessment information can be generated by analyzing the remediated power plant data.

[0141] Figures 17 to 19 The following are statistical distribution diagrams of the number of dead points in the power plant data after treatment of power plants A, B and C (i.e., stations A, B and C) according to the embodiments of the present disclosure. Figures 17 to 19 In the figure, the horizontal axis represents the date, and the vertical axis represents the number of death data points (which can be simply referred to as death points). Figures 17 to 19 Only an example of counting the number of dead points for active power, ambient temperature, total radiation (may be simply referred to as total radiation) and direct radiation (may be simply referred to as direct radiation) is shown, but the present disclosure is not limited thereto.

[0142] Fig. 20 The present invention is a statistical distribution diagram of the number of repeated data points in the power plant data after treatment of multiple power plants according to an embodiment of the present invention. By comparing with the power plant data before treatment, it can be found that the error conditions (e.g., over-threshold conditions, dead count conditions, and repeated conditions) of the power plant data after treatment are significantly reduced.

[0143] On the other hand, the changes before and after the treatment of specific fields in the power plant data can also be compared and analyzed. Fig.21 It is a comparative statistical chart of data of a specific field in the power plant data before and after treatment according to an embodiment of the present disclosure. Fig.21 As shown, it can be seen that the power plant data before treatment were missing or straightened abnormally, while the power plant data after treatment became more continuous and closer to the normal data change trend.

[0144] On the other hand, according to the embodiments of the present disclosure, the power plant data before governance and the power plant data after governance can be used to predict the output power respectively to generate the predicted power before governance and the predicted power after governance; the predicted power before governance and the predicted power after governance are compared with the actual power respectively to determine the prediction accuracy before governance and the prediction accuracy after governance.

[0145] Fig. 22 It is a comparative statistical chart of the prediction accuracy before and after the treatment according to the embodiment of the present disclosure. Fig. 22 The horizontal axis represents time, and the vertical axis represents prediction accuracy.

[0146] In the field of renewable energy power generation (e.g., wind power generation, photovoltaic power generation), power prediction is indispensable, and a power prediction model can be established using machine learning algorithms. For example, a power prediction model can be established using an ultra-short-term algorithm. For example, the power plant data before and after treatment from June 2018 to May 2019 can be used as training samples to establish a power prediction model, and then the power plant data before and after treatment in June 2019 can be used to test the power prediction, and then compared with the actual power to obtain Fig. 22 The prediction accuracy comparison chart shown in FIG. 1 is a graph showing the prediction accuracy of the data. In an embodiment, data governance can be referred to as data cleaning. The accuracy before cleaning can be used to represent the prediction accuracy before governance, and the accuracy after cleaning can be used to represent the prediction accuracy after governance. Fig. 22 It can be seen that the accuracy of modeling and prediction using the power plant data after treatment (i.e., the prediction accuracy after treatment) is mostly higher than the accuracy of modeling and prediction using the power plant data before treatment (i.e., the prediction accuracy before treatment), which fully demonstrates that the power plant data after treatment is more conducive to accurate power prediction.

[0147] In addition, other data testing methods applicable to power plant data can be used to test the power plant data before and after treatment to compare and analyze the changes in the power plant data before and after treatment.

[0148] According to an embodiment of the present disclosure, there is further provided a computer-readable storage medium on which a computer program is stored. When the computer program is executed, the data processing method according to the embodiment of the present disclosure is implemented.

[0149] In the embodiment of the present disclosure, the computer-readable storage medium may carry one or more programs, and when the computer program is executed, the reference Figures 2 to 22 The steps described include: analyzing the power plant data to generate data quality assessment information for the power plant data; and remediating the power plant data in response to generating the data quality assessment information.

[0150] The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In an embodiment of the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a computer program that may be used by or in conjunction with an instruction execution system, device or device. The computer program contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination thereof. The computer-readable storage medium may be contained in any device; it may also exist alone without being assembled into the device.

[0151] The above has been combined Figures 2 to 12 The data processing method according to the embodiment of the present disclosure is described. Fig.13 A computing device according to an embodiment of the present disclosure is described.

[0152] Fig.23 is a schematic diagram of a computing device according to an embodiment of the present disclosure.

[0153] Reference Fig.23 According to an embodiment of the present disclosure, the computing device 3 may include a memory 31 and a processor 32. A computer program 33 is stored in the memory 31. When the computer program 33 is executed by the processor 32, the data processing method according to an embodiment of the present disclosure is implemented.

[0154] In the embodiment of the present disclosure, when the computer program 33 is executed by the processor 32, reference can be made to Figures 2 to 22 The described data processing method operates by: analyzing the power plant data to generate data quality assessment information of the power plant data; and in response to generating the data quality assessment information, remediating the power plant data.

[0155] Fig.23 The computing device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0156] The above has been referred to Figures 1 to 23 The data processing device and method and computing apparatus of the power plant according to the embodiments of the present disclosure are described. However, it should be understood that: Figure 1The data processing devices and modules thereof shown in the figure may be respectively configured as software, hardware, firmware or any combination of the above items to perform specific functions. Fig.23 The computing device shown in is not limited to including the components shown above, but some components may be added or deleted as needed, and the above components may also be combined.

[0157] By adopting the data processing equipment and method according to the embodiments of the present disclosure, at least the following technical effects can be achieved: through data quality evaluation and data screening, the power plant data becomes very friendly to subsequent data users, and data users no longer need to pay too much attention to the data itself, but only need to focus on the optimization of research and development. For example, the prediction model requires a large amount of historical data, and algorithm engineers no longer need to devote a large part of their energy to processing data; the evaluation standards and systems for data quality are improved, and at the same time a basis is provided for subsequent data governance or cleaning; an efficient and accurate data governance method is provided for the field of new energy power generation, which can standardize the application of data.

[0158] The control logic or function performed by each component or controller in the control system can be represented by a flow chart or similar diagram in one or more figures. These figures provide representative control strategies and / or logic, which can be implemented using one or more processing strategies (such as event-driven, interrupt-driven, multi-tasking, multi-threading, etc.). Therefore, the various steps or functions shown can be executed in the order shown, executed in parallel, or omitted in some cases. Although not always clearly shown, it will be appreciated by those of ordinary skill in the art that one or more steps or functions shown can be repeatedly executed according to the specific processing strategy used.

[0159] While the present disclosure has been shown and described with reference to preferred embodiments, it will be understood by those skilled in the art that various modifications and variations may be made to these embodiments without departing from the spirit and scope of the disclosure as defined by the appended claims.

Claims

1. A data processing device for a power plant, characterized in that: The data processing device comprises: a data evaluation module configured to analyze the power plant data to generate data quality evaluation information of the power plant data; and a data governance module configured to govern the power plant data in response to generating the data quality assessment information; The data governance module is configured to: In response to generating data quality assessment information, determining missing types and / or field types corresponding to missing data in the power plant data; Filling missing data in the farm data according to the missing type and / or the field type, The missing type includes a continuous missing type and a non-continuous missing type, and the field type includes a power type and a non-power type. The power plant data includes: wind power generation data and / or photovoltaic power generation data, the fields of the wind power generation data include: at least one of the active power of wind power generation, the available power of wind power generation, the temperature of the wind measurement location, the wind speed of the wind measurement location, the wind direction of the wind measurement location, the humidity of the wind measurement location, and the air pressure of the wind measurement location; the fields of the photovoltaic power generation data include: at least one of the active power of photovoltaic power generation, the available power of photovoltaic power generation, the temperature of the radiometry measurement location, the direct radiometry of the radiometry measurement location, the total radiometry of the radiometry measurement location, the humidity of the radiometry measurement location, and the air pressure of the radiometry measurement location; wherein the data evaluation module is further configured to: score the quality of the power plant data according to the evaluation index, the evaluation index includes the integrity index, the accuracy index, and at least one of the correlation index and the consistency index; the data quality evaluation information includes the data quality score corresponding to the evaluation index; the accuracy index includes at least one of the over-threshold index, the dead number index, the repetition index, and the fluctuation index; the integrity index includes the missing index, The calculation formula of the data quality score includes: score=score1×score2, Among them, score1 represents the accuracy score, score2 represents the completeness score, score represents the final data quality score, i represents the field number, m represents the maximum value among all field numbers, j represents the indicator number, n represents the maximum value among the indicator numbers except for duplicate indicators and missing indicators, and W represents the maximum value among all indicator numbers. i represents the weight of the i-th field, W j represents the weight of the jth index, W 重复 represents the weight of repeated indicators, C ij represents the count value of the jth index of the i-th field, C i缺失 represents the count value of the missing indicator of the i-th field, C 重复 Represents the count value of repeated data in the power plant data, S i represents the total number of samples in the i-th field, S 总体 Represents the total number of samples of power plant data.

2. The data processing device according to claim 1, characterized in that The data governance module is configured to use a linear interpolation method to fill in missing data of non-continuous missing types, and / or, The data governance module is configured to fill missing data of the continuous missing type by at least one of the following operations: Selecting non-missing data whose geographical location is within a predetermined range of the geographical location of the missing data from the power plant data to fill the missing data; Selecting non-missing data corresponding to the missing data from another power plant data of the power plant to fill the missing data, wherein the data granularity of the another power plant data is smaller than the data granularity of the power plant data; Select similar day data having the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and use the non-missing data at the time corresponding to the time of the missing data in the similar day data to fill the missing data; Fill missing data with non-missing data at the time corresponding to the time of missing data on the day before the missing data; Selecting non-missing data at the same time as the missing data from the power plant data of another power plant within a predetermined distance of the power plant to fill the missing data; Use machine learning algorithms to predict data to fill in missing data, and use the predicted data to fill in missing data.

3. The data processing device according to claim 1, characterized in that The data quality assessment information includes at least one of completeness assessment information, accuracy assessment information, relevance assessment information and consistency assessment information. The data evaluation module is configured to perform the first two of the following operations and at least one of the last two operations: Analyze missing data at the power plant to generate completeness assessment information; analyzing the power plant data for at least one of over-threshold, dead count, duplication, and fluctuation to generate accuracy assessment information; Analyzing correlations between all fields in the power plant data to generate correlation assessment information; The farm data is analyzed for internal consistency and / or temporal consistency to generate consistency assessment information.

4. The data processing device according to claim 1, characterized in that The data governance module is configured to fill in the missing data of the power type by at least one of the following operations: Selecting non-missing data corresponding to the missing data from another power plant data of the power plant to fill the missing data, wherein the data granularity of the another power plant data is smaller than the data granularity of the power plant data; Select similar day data having the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and use the non-missing data at the time corresponding to the time of the missing data in the similar day data to fill the missing data; Fill missing data with non-missing data at the time corresponding to the time of missing data on the day before the missing data; Use machine learning algorithms to predict data to fill in missing data, use the predicted data to fill in missing data, and / or, The data governance module is configured to fill in the missing data of non-power type by at least one of the following operations: Selecting non-missing data whose geographical location is within a predetermined range of the geographical location of the missing data from the power plant data to fill the missing data; Selecting non-missing data corresponding to the missing data from another power plant data of the power plant to fill the missing data, wherein the data granularity of the another power plant data is smaller than the data granularity of the power plant data; Select similar day data having the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and use the non-missing data at the time corresponding to the time of the missing data in the similar day data to fill the missing data; Fill missing data with non-missing data at the time corresponding to the time of missing data on the day before the missing data; Selecting non-missing data at the same time as the missing data from the power plant data of another power plant within a predetermined distance of the power plant to fill the missing data; Use machine learning algorithms to predict data to fill in missing data, and use the predicted data to fill in missing data.

5. The data processing device according to claim 1, characterized in that The data evaluation module is configured to: analyze the treated power plant data to generate data quality evaluation information of the treated power plant data to compare with the data quality evaluation information of the power plant data before treatment.

6. The data processing device according to claim 1, characterized in that The data processing device further comprises: a data testing module configured to respectively apply the power plant data before treatment and the power plant data after treatment to perform output power prediction to generate the predicted power before treatment and the predicted power after treatment; The predicted power before governance and the predicted power after governance are compared with the actual power respectively to determine the prediction accuracy before governance and the prediction accuracy after governance.

7. The data processing device according to claim 1, characterized in that The data governance module is further configured to: determine whether the data quality assessment information meets a predetermined standard; If the data quality assessment information does not meet the predetermined standards, the power plant data is governed.

8. A data processing method for a power plant, characterized in that: The data processing method comprises: analyzing the farm data to generate data quality assessment information for the farm data; and In response to generating the data quality assessment information, remediating the power plant data; The step of managing power plant data includes: In response to generating data quality assessment information, determining missing types and / or field types corresponding to missing data in the power plant data; Filling missing data in the farm data according to the missing type and / or the field type, The missing type includes a continuous missing type and a non-continuous missing type, and the field type includes a power type and a non-power type. The power plant data includes: wind power generation data and / or photovoltaic power generation data, the fields of the wind power generation data include: at least one of the active power of wind power generation, the available power of wind power generation, the temperature of the wind measurement location, the wind speed of the wind measurement location, the wind direction of the wind measurement location, the humidity of the wind measurement location, and the air pressure of the wind measurement location; the fields of the photovoltaic power generation data include: at least one of the active power of photovoltaic power generation, the available power of photovoltaic power generation, the temperature of the radiometry measurement location, the direct radiometry of the radiometry measurement location, the total radiometry of the radiometry measurement location, the humidity of the radiometry measurement location, and the air pressure of the radiometry measurement location; wherein the data evaluation module is further configured to: score the quality of the power plant data according to the evaluation index, the evaluation index includes the integrity index, the accuracy index, and at least one of the correlation index and the consistency index; the data quality evaluation information includes the data quality score corresponding to the evaluation index; the accuracy index includes at least one of the over-threshold index, the dead number index, the repetition index, and the fluctuation index; the integrity index includes the missing index, The calculation formula of the data quality score includes: score=score1×score2, Among them, score1 represents the accuracy score, score2 represents the completeness score, score represents the final data quality score, i represents the field number, m represents the maximum value among all field numbers, j represents the indicator number, n represents the maximum value among the indicator numbers except for duplicate indicators and missing indicators, and W represents the maximum value among all indicator numbers. i represents the weight of the i-th field, W j represents the weight of the jth index, W 重复 represents the weight of repeated indicators, C ij represents the count value of the jth index of the i-th field, C i缺失 represents the count value of the missing indicator of the i-th field, C 重复 Represents the count value of repeated data in the power plant data, S i represents the total number of samples in the i-th field, S 总体 Represents the total number of samples of power plant data.

9. The data processing method according to claim 8, characterized in that: The step of managing the power plant data includes using a linear interpolation method to fill in the missing data of a non-continuous missing type, and / or, The step of managing the power plant data includes filling missing data of a continuous missing type by at least one of the following operations: Selecting non-missing data whose geographical location is within a predetermined range of the geographical location of the missing data from the power plant data to fill the missing data; Selecting non-missing data corresponding to the missing data from another power plant data of the power plant to fill the missing data, wherein the data granularity of the another power plant data is smaller than the data granularity of the power plant data; Select similar day data having the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and use the non-missing data at the time corresponding to the time of the missing data in the similar day data to fill the missing data; Fill missing data with non-missing data at the time corresponding to the time of missing data on the day before the missing data; Selecting non-missing data at the same time as the missing data from the power plant data of another power plant within a predetermined distance of the power plant to fill the missing data; Use machine learning algorithms to predict data to fill in missing data, and use the predicted data to fill in missing data.

10. The data processing method according to claim 8, characterized in that: The data quality assessment information includes at least one of completeness assessment information, accuracy assessment information, relevance assessment information and consistency assessment information. The step of analyzing the power plant data to generate data quality assessment information of the power plant data includes at least one of the first two operations and the last two operations in the following operations: Analyze missing data at the power plant to generate completeness assessment information; analyzing the power plant data for at least one of over-threshold, dead count, duplication, and fluctuation to generate accuracy assessment information; Analyzing correlations between all fields in the power plant data to generate correlation assessment information; The farm data is analyzed for internal consistency and / or temporal consistency to generate consistency assessment information.

11. The data processing method according to claim 8, characterized in that: The step of managing the power plant data includes filling in missing data of the power type by at least one of the following operations: Selecting non-missing data corresponding to the missing data from another power plant data of the power plant to fill the missing data, wherein the data granularity of the another power plant data is smaller than the data granularity of the power plant data; Select similar day data having the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and use the non-missing data at the time corresponding to the time of the missing data in the similar day data to fill the missing data; Fill missing data with non-missing data at the time corresponding to the time of missing data on the day before the missing data; Use machine learning algorithms to predict data to fill in missing data, use the predicted data to fill in missing data, and / or, The step of managing the power plant data includes filling in missing data of non-power type by at least one of the following operations: Selecting non-missing data whose geographical location is within a predetermined range of the geographical location of the missing data from the power plant data to fill the missing data; Selecting non-missing data corresponding to the missing data from another power plant data of the power plant to fill the missing data, wherein the data granularity of the another power plant data is smaller than the data granularity of the power plant data; Select similar day data having the highest similarity to the data of the day corresponding to the missing data from the non-missing historical power plant data of the power plant, and use the non-missing data at the time corresponding to the time of the missing data in the similar day data to fill the missing data; Fill missing data with non-missing data at the time corresponding to the time of missing data on the day before the missing data; Selecting non-missing data at the same time as the missing data from the power plant data of another power plant within a predetermined distance of the power plant to fill the missing data; Use machine learning algorithms to predict data to fill in missing data, and use the predicted data to fill in missing data.

12. The data processing method according to claim 8, characterized in that: The data processing method further includes: analyzing the treated power plant data to generate data quality assessment information of the treated power plant data to compare with the data quality assessment information of the power plant data before the treatment.

13. The data processing method according to claim 8, characterized in that: The data processing method further comprises: respectively applying the power plant data before treatment and the power plant data after treatment to perform output power prediction to generate the predicted power before treatment and the predicted power after treatment; The predicted power before governance and the predicted power after governance are compared with the actual power respectively to determine the prediction accuracy before governance and the prediction accuracy after governance.

14. The data processing method according to claim 8, characterized in that: The steps of governing power plant data include: determining whether data quality assessment information meets predetermined standards; If the data quality assessment information does not meet the predetermined standards, the power plant data is governed.

15. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the data processing method according to any one of claims 8 to 14 is implemented.

16. A computing device, characterized in that: The computing device comprises: processor; A memory storing a computer program, which, when executed by a processor, implements the data processing method according to any one of claims 8 to 14.

Images