Multi-source data thermal power plant load evaluation method

Abstract

The invention provides a multi-source data thermal power plant load evaluation method, which relates to the technical field of power load evaluation, and comprises the following steps: step 1, obtaining historical load data and real-time load data of each unit of a power plant, and outputting historical-real-time load data; step 2, on the basis of the historical-real-time load data, combining with the obtained other-source data output to obtain multi-source data; 3, preprocessing the multi-source data by using a selected preset preprocessing method to obtain to-be-analyzed data; and 4, performing real-time evaluation and prediction analysis on the to-be-analyzed data in combination with a preset load evaluation method to obtain load evaluation-prediction data, and adjusting and optimizing an operation strategy of each unit based on the load evaluation-prediction data. The load data of the power plant can be evaluated and predicted in combination with diversified multi-source data, the accuracy of load evaluation and prediction results is improved, and the operation stability of a unit is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric load assessment, and in particular to a method for assessing load of a thermal power plant using multi-source data. Background Art

[0002] Load assessment of thermal power plants refers to the evaluation and prediction of the load data and load status of the power plant in order to effectively adjust the operation of the generating units and ensure the stable operation of the power grid.

[0003] At present, the load assessment of power plants mainly relies on the historical load information of each unit in the power plant to evaluate and predict the load data in the current or future period. It lacks consideration of the impact of external factors such as meteorological data, environmental data, and fuel market data on the load data of power plants, resulting in poor accuracy of its evaluation and prediction results, which makes it difficult to meet the increasing demand for evaluation and prediction accuracy.

[0004] Therefore, the present invention provides a method for evaluating load of a thermal power plant based on multi-source data. Summary of the invention

[0005] The present invention provides a multi-source data load assessment method for a thermal power plant, which combines diverse multi-source data to assess and predict the load data of a power plant, thereby improving the accuracy of load assessment and prediction results.

[0006] The present invention provides a method for evaluating load of a thermal power plant using multi-source data, comprising:

[0007] Step 1: Obtain historical load data and real-time load data of each unit in the power plant, and output historical-real-time load data;

[0008] Step 2: Based on the historical-real-time load data, multi-source data is obtained in combination with the acquired other source data output;

[0009] Step 3: preprocessing the multi-source data using the selected preset preprocessing method to obtain data to be analyzed;

[0010] Step 4: Perform real-time evaluation and predictive analysis on the data to be analyzed in combination with a preset load evaluation method to obtain load evaluation-prediction data, and adjust and optimize the operation strategy of each unit based on the load evaluation-prediction data.

[0011] Preferably, step 1 includes:

[0012] Obtain the load data of each unit in real time through the preset monitoring system and output the real-time load data;

[0013] Extracting features from the real-time load data, and selecting historical data with a matching degree greater than a first preset degree from a historical database based on the extracted features, and outputting the historical load data;

[0014] Based on the real-time load data and the historical load data, historical-real-time load data is output.

[0015] Preferably, based on the real-time load data and the historical load data, outputting the historical-real-time load data includes:

[0016] Time-aligning the real-time load data and the historical load data within a preset time period to obtain first data;

[0017] Data verification is performed on the first data at each moment in the preset time period, and the first data that passes the verification is output as historical-real-time load data.

[0018] Preferably, step 2 includes:

[0019] Based on the acquired load assessment requirement information, determine the type and accuracy requirements of the data source, and generate source data requirement information;

[0020] Based on the source data demand information, other source data in the power plant are acquired using preset monitoring equipment, and time alignment of other source data is performed based on the time axis of the historical-real-time load data to construct multi-source data.

[0021] Preferably, the other source data includes: meteorological data, fuel data, equipment operation data, environmental monitoring data, operation and maintenance data and market data.

[0022] Preferably, step 3 includes:

[0023] Performing content parsing on the multi-source data to obtain a first data packet corresponding to each data source in the multi-source data;

[0024] Converting a first data packet under the multi-source data into a second data packet in a preset format, and performing feature extraction to obtain a first feature set;

[0025] At the same time, feature matching is performed on the first feature set based on a preset abnormal feature database to obtain a first matching result corresponding to the abnormal data feature and a second matching result corresponding to the normal data feature;

[0026] Based on the first matching result and in combination with a preset feature-factor adaptation table, a first matching factor corresponding to the feature in the first matching result is obtained;

[0027] Inputting the first matching factor into a preset preprocessing method database for method matching, extracting preprocessing methods with a matching degree greater than a second preset degree and putting them into a preset data pool to generate a preprocessing method data pool;

[0028] Based on the first matching result and the first matching factor, a preset preprocessing method corresponding to the corresponding abnormal data feature in the first matching result is selected from the preprocessing method data pool, and the corresponding multi-source data is preprocessed to obtain second data;

[0029] Performing feature recognition on the second matching result, and obtaining a second feature set of the multi-source data based on the recognition result;

[0030] At the same time, the first feature set and the second feature set are combined to perform feature marking on the corresponding data in the second data to obtain data to be analyzed.

[0031] Preferably, step 4 includes:

[0032] Performing data division on the data to be analyzed, and constructing load data to be analyzed and other source data to be analyzed;

[0033] Extracting features of the load data to be analyzed, and selecting load assessment indicators from an indicator database to construct a first indicator set;

[0034] Based on the first indicator set, in combination with a preset indicator-method mapping table, a first evaluation method having a matching degree greater than a third preset degree is selected from a method database;

[0035] At the same time, feature extraction is performed on the other source data to be analyzed, and a mapping relationship between the other source data and each load assessment index in the first index set is obtained, and then a second assessment method is selected from the method database;

[0036] Based on the first evaluation method, real-time analysis and predictive analysis are performed on the load data to be analyzed to generate a master data load analysis result;

[0037] At the same time, the source data to be analyzed is analyzed based on the second evaluation method to generate a secondary data load analysis result;

[0038] Use the preset consensus algorithm to reach consensus on the primary data load analysis results and the secondary data load analysis results in the same period, and output the load assessment-forecast data;

[0039] Performing differential analysis on the load assessment-prediction data and the acquired standard control data, and outputting a load comparison result;

[0040] Based on the load comparison result and in combination with the preset result-index comparison table, the index to be optimized is selected in the index database, and an optimized index set is constructed;

[0041] Based on the optimization index set, optimizing and analyzing the primary data load analysis result and the secondary data load analysis result to obtain optimizable items;

[0042] Based on the optimizable items, the corresponding optimization method is selected from the method database, and the operation strategy of each unit is adjusted and optimized, and an optimization log is generated.

[0043] Preferably, based on the optimizable items, a corresponding optimization method is selected from a method database, and the operation strategy of each unit is adjusted and optimized, and an optimization log is generated, including:

[0044] Conduct source tracing analysis on each optimizable item, obtain the data source of each optimizable item and the historical change data of the corresponding optimizable item, and construct the data to be optimized for the corresponding optimizable item;

[0045] Obtaining the correlation between the data to be optimized and the load data, and analyzing the correlation corresponding to each optimizable item through a preset model to obtain a correlation coefficient between the corresponding optimizable item and the load;

[0046] Based on the to-be-optimized data and the correlation coefficient of each optimizable item, a first prediction curve of the corresponding optimizable item in a preset future period is constructed;

[0047] Based on the load data to be analyzed and the main data load analysis result, a first historical curve and a second prediction curve are generated respectively;

[0048] Extracting features from the first historical curve, the first prediction curve, and the second prediction curve respectively to construct a curve feature set, and matching the first optimization method in a preset method database based on the curve feature set;

[0049] In combination with the optimization index set, the first optimization method is used to optimize the data to be optimized corresponding to each optimizable item to obtain an adjustment amount and optimized data;

[0050] Use the preset evaluation model to analyze the data to be optimized, the adjustment amount, and the optimized data under each optimizable item to generate the load optimization coefficient of the power plant;

[0051]

[0052] Among them, μ represents the load optimization coefficient of the power plant; Δαi represents the adjustment amount of the i-th optimizable item; n represents the total number of optimizable items; represents the gain coefficient of the adjustment amount Δαi of the i-th optimizable item on the power plant load; Δβ1i represents the impact value of the i-th optimizable item on the demand side of the power plant; θi represents the impact coefficient of the demand side on the load; Δβ2i represents the impact value of the i-th optimizable item on the supply side of the power plant; δi represents the impact coefficient of the supply side on the load; σ1 represents the adjustment coefficient of the external response; σ2 represents the adjustment coefficient of the internal response of the unit;

[0053] If the load optimization coefficient is within the preset standard threshold range, all optimizable items are regarded as items to be adjusted;

[0054] If the load optimization coefficient is not within the preset standard threshold range, the optimizable item whose adjustment amount is greater than the preset amount is extracted as the item to be adjusted;

[0055] The item to be adjusted and the adjustment amount are matched with a preset item-amount-method table to obtain a corresponding second optimization method, and the obtained current operation strategy is adjusted and optimized based on the second optimization method.

[0056] The present invention provides a multi-source data load assessment method for a thermal power plant. First, historical load data and real-time load data of each unit in the power plant are obtained to generate historical-real-time load data, and then multi-source data containing diversified data related to the power plant load is obtained by combining the obtained other source data output. Subsequently, the multi-source data is preprocessed by a selected preset preprocessing method, thereby improving the data quality of the multi-source data and providing convenience for subsequent data analysis and prediction. Subsequently, load assessment and prediction analysis are performed by using the preset load assessment method, which not only improves the accuracy of the load assessment results, but also can timely discover abnormal situations such as potential overload through the prediction results, and adjusts the power generation plan and operation strategy of each unit in the power plant in combination with the assessment results and the prediction results, which not only reduces the energy consumption level of the power plant, improves the resource utilization rate of the power plant, but also can ensure the smooth operation of the power plant. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0058] Figure 1 It is a flow chart of a method for load assessment of a thermal power plant based on multi-source data provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] Combine the following Figure 1 The present invention describes a method for load assessment of a thermal power plant based on multi-source data.

[0061] like Figure 1 As shown, an embodiment of the present invention provides a method for evaluating load of a thermal power plant using multi-source data, including:

[0062] Step 1: Obtain historical load data and real-time load data of each unit in the power plant, and output historical-real-time load data;

[0063] Step 2: Based on historical and real-time load data, multi-source data is obtained by combining the acquired other source data output;

[0064] Step 3: Use the selected preset preprocessing method to preprocess the multi-source data to obtain the data to be analyzed;

[0065] Step 4: Combine the preset load assessment method to conduct real-time assessment and predictive analysis on the data to be analyzed, obtain load assessment-prediction data, and adjust and optimize the operation strategy of each unit based on the load assessment-prediction data.

[0066] In this embodiment, the historical load data: that is, the load change data of each unit in the electric field within a certain period of time in the past;

[0067] In this embodiment, real-time load data refers to the load data of each unit obtained through real-time collection;

[0068] In this embodiment, historical-real-time load data: data including historical load data and real-time load data of each unit;

[0069] In this embodiment, other source data refers to data composed of multiple data other than load-related data directly obtained in the power plant;

[0070] In this embodiment, multi-source data: summary data consisting of various data related to power plant load;

[0071] In this embodiment, the preset preprocessing method: that is, the method for preprocessing various types of data in the multi-source data is preset, including but not limited to steps such as data cleaning, data conversion, and data integration;

[0072] In this embodiment, the data to be analyzed: that is, the multi-source data obtained after preprocessing;

[0073] In this embodiment, a load assessment method is preset: a method for real-time assessment and predictive analysis of the data to be analyzed, so as to determine the real-time load status and load distribution in the power plant, and predict the future trend of the load data of the power plant, so as to facilitate the subsequent timely adjustment of the power generation plan and operation strategy of the power plant to ensure the smooth operation of the power plant;

[0074] In this embodiment, load assessment-forecast data: data including load assessment results and load forecast results.

[0075] The implementation principle and beneficial effects of the embodiments of the present invention: The present invention first obtains the historical load data and real-time load data of each unit in the power plant to generate historical-real-time load data, and then combines the obtained other source data output to obtain multi-source data containing diversified data related to the power plant load. The multi-source data is then preprocessed by a selected preset preprocessing method, thereby improving the data quality of the multi-source data and facilitating subsequent data analysis and prediction. Subsequently, load evaluation and prediction analysis are performed by a preset load evaluation method, which not only improves the accuracy of the load evaluation results, but also can timely discover abnormal situations such as potential overload through the prediction results, and adjusts the power generation plan and operation strategy of each unit in the power plant in combination with the evaluation results and the prediction results, which not only reduces the energy consumption level of the power plant, improves the resource utilization rate of the power plant, but also can ensure the smooth operation of the power plant.

[0076] An embodiment of the present invention provides a method for evaluating load of a thermal power plant using multi-source data, wherein step 1 includes:

[0077] Obtain the load data of each unit in real time through the preset monitoring system and output the real-time load data;

[0078] Extracting features from the real-time load data, and selecting historical data with a matching degree greater than a first preset degree from a historical database based on the extracted features, and outputting the historical load data;

[0079] Based on the real-time load data and the historical load data, the historical-real-time load data is output.

[0080] In this embodiment, the preset monitoring system: a system for collecting and monitoring the load data of each unit is pre-set, such as a smart meter, an EMS energy management system, etc.;

[0081] In this embodiment, the first preset degree is used to screen out threshold parameters whose matching degree with the current real-time load data is greater than a preset condition. The historical load data whose matching degree is greater than the first preset degree has reference and analysis value, while the historical load data whose matching degree is lower than the first preset degree has a low correlation with the current real-time load data and has no reference value.

[0082] The implementation principle and beneficial effects of the embodiments of the present invention: The present invention can collect and monitor load data such as load type, load status, load distribution, etc. of each unit in a power plant in real time through a preset monitoring system, and output real-time load data, and then filter out corresponding historical load data in the historical database based on the extracted features obtained by feature extraction of the real-time load data. While ensuring the accuracy of data screening, the speed of obtaining historical load data is improved, and data support is provided for subsequent load evaluation and predictive analysis.

[0083] An embodiment of the present invention provides a method for evaluating load of a thermal power plant using multi-source data, which outputs historical-real-time load data based on real-time load data and historical load data, and includes:

[0084] Time-aligning the real-time load data and the historical load data within a preset time period to obtain first data;

[0085] Data verification is performed on the first data at each moment in the preset time period, and the first data that passes the verification is output as historical-real-time load data.

[0086] In this embodiment, the preset time period is a pre-selected time period used to analyze the real-time load data and historical load data within the selected time period;

[0087] In this embodiment, time alignment: that is, combining the timestamps of real-time load data and historical load data to align the load data within a preset period, not only can the continuity of the data be ensured, but also abnormal data such as missing values ​​and duplicate values ​​in the data can be discovered in time;

[0088] In this embodiment, the first data: namely, the real-time load data and the historical load data obtained after time alignment processing;

[0089] In this embodiment, data verification is an operation used to verify the accuracy, completeness and consistency of data corresponding to each time scale within a preset time period.

[0090] The implementation principle and beneficial effects of the embodiments of the present invention: The present invention can timely discover abnormal data in the load data by time-aligning the real-time load data and the historical load data within a preset time period, which facilitates the subsequent selection of the corresponding preprocessing method to process it. At the same time, the accuracy, completeness and consistency of the load data within the preset time period can be checked through data verification, thereby ensuring the data quality of the load data.

[0091] An embodiment of the present invention provides a method for evaluating load of a thermal power plant using multi-source data, wherein step 2 includes:

[0092] Based on the acquired load assessment requirement information, determine the type and accuracy requirements of the data source, and generate source data requirement information;

[0093] Based on the source data demand information, the preset monitoring equipment is used to obtain other source data in the power plant, and the other source data are time-aligned based on the time axis of historical-real-time load data to construct multi-source data.

[0094] In this embodiment, the load assessment demand information is demand information used to assess the load of the power plant, such as prediction performance requirements, accuracy requirements, energy consumption optimization requirements, etc. Generally, the higher the requirements for the load assessment demand information, the more types of data sources are required, and the higher the sampling accuracy requirements of the corresponding data sources are;

[0095] In this embodiment, the category requirement refers to the requirement on the category of data source;

[0096] In this embodiment, the accuracy requirement: that is, the sampling accuracy requirement for the data of the corresponding data source;

[0097] In this embodiment, the preset monitoring device: a device used to sample, collect and monitor data from each data source, is pre-set and usually has a high degree of matching with the data source.

[0098] Implementation principles and beneficial effects of the embodiments of the present invention: The present invention analyzes and processes the acquired load assessment demand information, thereby determining the type and accuracy requirements of the load assessment on the data source based on the analysis results, and generates source data demand information, and then accurately acquires the corresponding other source data in combination with the preset monitoring equipment, which not only improves the diversity of data sources, but also enables the acquired data type and data quality to match the load assessment requirements, thereby ensuring the accuracy of the load assessment results.

[0099] The embodiment of the present invention provides a method for load assessment of a thermal power plant based on multi-source data, wherein the multi-source data includes: meteorological data, fuel data, equipment operation data, environmental monitoring data, operation and maintenance data, and market data.

[0100] In this embodiment, meteorological data includes meteorological information such as temperature, humidity, and wind speed. These data have an important impact on the operation of the cooling system and air preheating system of the thermal power plant. In addition, the demand side of the power grid will also change greatly under different meteorological conditions. For example, the power demand will rise sharply during the high temperature period in summer.

[0101] In this embodiment, fuel data: for example, data such as fuel type and combustion efficiency;

[0102] In this embodiment, the equipment operation data includes the operation data of the equipment such as the generator set, boiler, steam turbine, etc., for example, parameters such as temperature, pressure, and rotation speed;

[0103] In this embodiment, the environmental monitoring data mainly includes the emission monitoring data of various pollutants and environmental impact data such as noise and vibration, for example, sulfur dioxide, nitrogen oxides, decibels, etc.;

[0104] In this embodiment, operation and maintenance data: i.e., data such as equipment inspection and maintenance records;

[0105] In this embodiment, market data refers to market-related data such as electricity market prices and fuel market prices.

[0106] The implementation principles and beneficial effects of the embodiments of the present invention: The present invention can comprehensively acquire meteorological data, fuel data, equipment operation data, environmental monitoring data, operation and maintenance data and market data, and obtain diversified data types and load-related other source data, which not only provides accurate and diverse data support for subsequent load evaluation and forecasting analysis, but also ensures the accuracy of subsequent load evaluation and forecasting analysis results, and reduces the occurrence of low accuracy and large errors in load evaluation results due to single data.

[0107] An embodiment of the present invention provides a method for evaluating load of a thermal power plant using multi-source data, wherein step 3 includes:

[0108] Performing content parsing on the multi-source data to obtain a first data packet corresponding to each data source in the multi-source data;

[0109] Converting a first data packet under the multi-source data into a second data packet in a preset format, and performing feature extraction to obtain a first feature set;

[0110] At the same time, feature matching is performed on the first feature set based on a preset abnormal feature database to obtain a first matching result corresponding to the abnormal data feature and a second matching result corresponding to the normal data feature;

[0111] Based on the first matching result and in combination with a preset feature-factor adaptation table, a first matching factor corresponding to the feature in the first matching result is obtained;

[0112] Inputting the first matching factor into a preset preprocessing method database for method matching, extracting preprocessing methods with a matching degree greater than a second preset degree and putting them into a preset data pool, thereby generating a preprocessing method data pool;

[0113] Based on the first matching result and the first matching factor, a preset preprocessing method corresponding to the corresponding abnormal data feature in the first matching result is selected from the preprocessing method data pool, and the corresponding multi-source data is preprocessed to obtain second data;

[0114] Performing feature recognition on the second matching result, and obtaining a second feature set of the multi-source data based on the recognition result;

[0115] At the same time, the first feature set and the second feature set are combined to perform feature marking on the corresponding data in the second data to obtain the data to be analyzed.

[0116] In this embodiment, the first data packet: that is, the data content parsing packet corresponding to each data source in the multi-source data:

[0117] In this embodiment, the preset format: a pre-set unified data format, which facilitates the subsequent extraction of data features in the data packet;

[0118] In this embodiment, the second data packet: that is, the first data packet converted into a preset format;

[0119] In this embodiment, the first feature set is a set consisting of a plurality of data features obtained by extracting features from the second data packet;

[0120] In this embodiment, the preset abnormal feature database: a database containing various abnormal data features, which is preset and used for feature matching with the abnormal data features in the second data packet;

[0121] In this embodiment, the first matching result: that is, the matching result including abnormal data features;

[0122] In this embodiment, the second matching result: that is, the matching result including normal data features;

[0123] In this embodiment, the preset feature-factor adaptation table: an adaptation table including the correspondence between the abnormal data features and the first matching factors, is preset;

[0124] In this embodiment, the first matching factor is a matching factor obtained according to a preset feature-factor adaptation table and used to select a preprocessing method;

[0125] In this embodiment, the preset preprocessing method database: a database containing a variety of preprocessing methods and steps, including but not limited to methods or steps such as data cleaning, data integration, data conversion, and data smoothing;

[0126] In this embodiment, the second preset degree is used to select a threshold value of a preprocessing method that meets the preset conditions from a preset preprocessing method database;

[0127] In this embodiment, the preprocessing method data pool: a data pool including a plurality of preprocessing methods, used for temporarily storing the preprocessing methods;

[0128] In this embodiment, the second data: that is, the multi-source data obtained after preprocessing;

[0129] In this embodiment, the second feature set is a set of normal data features corresponding to the second matching result.

[0130] The implementation principle and beneficial effects of the embodiments of the present invention are as follows: The present invention performs content analysis on multi-source data and converts it into a first data packet corresponding to each data source, and then converts the first data packet into a second data packet in a preset format, so as to facilitate the subsequent extraction of data features in the data packets corresponding to each data source; then the extracted features are classified through a preset abnormal feature database to obtain a first matching result corresponding to the abnormal data feature and a first matching result corresponding to the normal data feature, respectively, and then a preprocessing method for preprocessing the corresponding abnormal data feature in the first matching result is selected, and the multi-source data is preprocessed, which not only improves the matching degree between the preprocessing method and the abnormal data, but also improves the preprocessing effect of the abnormal data corresponding to the abnormal data feature, thereby improving the data quality of the multi-source data.

[0131] An embodiment of the present invention provides a method for evaluating load of a thermal power plant using multi-source data, wherein step 4 includes:

[0132] Divide the data to be analyzed, and construct the load data to be analyzed and the other source data to be analyzed;

[0133] Extracting features of the load data to be analyzed, and selecting load assessment indicators from an indicator database to construct a first indicator set;

[0134] Based on the first indicator set, in combination with the preset indicator-method mapping table, a first evaluation method having a matching degree greater than a third preset degree is selected from the method database;

[0135] At the same time, feature extraction is performed on the other source data to be analyzed, and a mapping relationship between the load assessment indicators in the first indicator set is obtained, and then a second assessment method is selected in the method database;

[0136] Based on the first evaluation method, real-time analysis and prediction analysis are performed on the load data to be analyzed to generate a master data load analysis result;

[0137] At the same time, the other source data to be analyzed is analyzed based on the second evaluation method to generate secondary data load analysis results;

[0138] Use the preset consensus algorithm to reach consensus on the primary data load analysis results and the secondary data load analysis results in the same period, and output the load assessment-forecast data;

[0139] Perform differential analysis on the load assessment-prediction data and the acquired standard control data, and output the load comparison results;

[0140] Based on the load comparison results and in combination with the preset result-index comparison table, the index to be optimized is selected in the index database, and an optimized index set is constructed;

[0141] Based on the optimization index set, the primary data load analysis results and the secondary data load analysis results are optimized and analyzed to obtain the optimizable items;

[0142] Based on the optimizable items, the corresponding optimization method is selected from the method database, and the operation strategy of each unit is adjusted and optimized, and an optimization log is generated.

[0143] In this embodiment, the load data to be analyzed includes load quantity, load distribution and other data directly related to the load.

[0144] In this embodiment, the other source data to be analyzed includes data to be analyzed from other data sources indirectly related to the load, such as temperature;

[0145] In this embodiment, the first indicator set includes a set of multiple load assessment indicators selected based on the characteristics of the load data to be analyzed, such as load factor, load rate, load peak-to-valley ratio, etc.;

[0146] In this embodiment, the preset indicator-method mapping table: a data table containing the mapping relationship between each load assessment indicator and the corresponding first assessment method, which is preset;

[0147] In this embodiment, the third preset degree is used to filter out a threshold value of the first evaluation method that matches the load evaluation index in the method database;

[0148] In this embodiment, the first evaluation method is a method obtained by screening in a method database according to load evaluation indicators and used for load evaluation of load data to be analyzed;

[0149] In this embodiment, the second evaluation method is a method selected from the method database and used to perform load evaluation on the other source data to be analyzed;

[0150] In this embodiment, the main data load analysis result: that is, the analysis result obtained by performing load evaluation and prediction analysis on the load data to be analyzed by the first evaluation method;

[0151] In this embodiment, the secondary data load analysis result is the analysis result obtained by performing load evaluation and prediction analysis on the source data to be analyzed by the second evaluation method;

[0152] In this embodiment, the preset consensus algorithm: the algorithm used for reaching consensus on the primary data load analysis result and the secondary data load analysis result is preset, for example, AHP hierarchical analysis method, TOPSIS decision algorithm, etc.;

[0153] In this embodiment, the load comparison result is the analysis result obtained after performing a difference analysis on the load assessment-prediction data;

[0154] In this embodiment, the preset result-index comparison table: a data table containing the comparison relationship between the load comparison results and the index to be optimized, which is preset;

[0155] In this embodiment, the index set to be optimized is a set consisting of multiple indexes to be optimized;

[0156] In this embodiment, the optimizable items are items that can be optimized;

[0157] In this embodiment, the optimization log is a log file generated by recording the process data and time characteristics of adjusting and optimizing the operation strategy of each unit according to the optimization method, so as to facilitate subsequent review.

[0158] The implementation principle and beneficial effects of the embodiments of the present invention: The present invention constructs the load data to be analyzed and the other source data to be analyzed through data partitioning, evaluates and predicts the power plant load from two aspects: the load data to be analyzed containing the main load data and the other source data to be analyzed containing the secondary load data, and then uses a preset consensus algorithm to reach a consensus on the load analysis results to obtain load evaluation-prediction data, thereby improving the accuracy of load evaluation; and then obtains the load comparison result through difference analysis, and then optimizes and analyzes the main and auxiliary load analysis results respectively based on the set of indicators to be optimized obtained based on the load comparison result, and adjusts and optimizes the operation strategy of the power plant units based on the optimizable items and the corresponding optimization methods, thereby continuously improving the stability and safety of the power plant.

[0159] The embodiment of the present invention provides a method for evaluating the load of a thermal power plant based on multi-source data. Based on the optimizable items, a corresponding optimization method is selected from a method database, and the operation strategy of each unit is adjusted and optimized, and an optimization log is generated, including:

[0160] Conduct source tracing analysis on each optimizable item, obtain the data source of each optimizable item and the historical change data of the corresponding optimizable item, and construct the data to be optimized for the corresponding optimizable item;

[0161] Obtain the correlation between the data to be optimized and the load data, and analyze the correlation corresponding to each optimizable item through a preset model to obtain the correlation coefficient between the corresponding optimizable item and the load;

[0162] Based on the to-be-optimized data and the correlation coefficient of each optimizable item, a first prediction curve of the corresponding optimizable item in a preset future period is constructed;

[0163] Based on the load data to be analyzed and the main data load analysis result, a first historical curve and a second prediction curve are generated respectively;

[0164] Extracting features from the first historical curve, the first prediction curve, and the second prediction curve respectively, constructing a curve feature set, and matching the first optimization method in a preset method database based on the curve feature set;

[0165] In combination with the optimization index set, the first optimization method is used to optimize the data to be optimized corresponding to each optimizable item to obtain an adjustment amount and optimized data;

[0166] Use the preset evaluation model to analyze the data to be optimized, the adjustment amount, and the optimized data under each optimizable item to generate the load optimization coefficient of the power plant;

[0167]

[0168] Among them, μ represents the load optimization coefficient of the power plant; Δαi represents the adjustment amount of the i-th optimizable item; n represents the total number of optimizable items; represents the gain coefficient of the adjustment amount Δαi of the i-th optimizable item on the power plant load; Δβ1i represents the impact value of the i-th optimizable item on the demand side of the power plant; θi represents the impact coefficient of the demand side on the load; Δβ2i represents the impact value of the i-th optimizable item on the supply side of the power plant; δi represents the impact coefficient of the supply side on the load; σ1 represents the adjustment coefficient of the external response; σ2 represents the adjustment coefficient of the internal response of the unit;

[0169] If the load optimization factor is within the preset standard threshold, all optimizable items are considered to be adjusted;

[0170] If the load optimization coefficient is not within the preset standard threshold range, the optimizable items with an adjustment amount greater than the preset amount are taken as items to be adjusted;

[0171] The item to be adjusted and the adjustment amount are matched with the preset item-amount-method table to obtain the corresponding second optimization method, and the obtained current operation strategy is adjusted and optimized based on the second optimization method.

[0172] In this embodiment, source tracing analysis: a method for analyzing the data source corresponding to each optimizable item and the corresponding historical evolution data;

[0173] In this embodiment, historical change data: that is, historical evolution data corresponding to each optimization item obtained through source tracing analysis;

[0174] In this embodiment, the data to be optimized includes the data source corresponding to the optimization item and the corresponding historical change data, and the data to be optimized;

[0175] In this embodiment, correlation coefficient: a parameter used to characterize the strength of the correlation relationship between each optimizable item and the load;

[0176] In this embodiment, the first prediction curve is a prediction curve constructed by using the data to be optimized and the correlation coefficients of each optimizable item;

[0177] In this embodiment, the first historical curve is a historical data change curve generated according to the load data to be analyzed and the main data load analysis result;

[0178] In this embodiment, the second prediction curve: that is, the prediction curve constructed based on the load data to be analyzed and the main data load analysis result;

[0179] In this embodiment, the curve feature set is a set of extracted features obtained by extracting features from the first historical curve, the first predicted curve, and the second predicted curve, including but not limited to curvature, inflection point, inflection point, slope, etc.;

[0180] In this embodiment, the first optimization method is a method selected from a preset method database based on a curve feature set and used to optimize each optimizable item;

[0181] In this embodiment, the preset standard threshold range is used to determine whether each optimizable item corresponding to the load optimization coefficient can be calibrated as a threshold range of the item to be adjusted;

[0182] In this embodiment, the items to be adjusted are items that meet the preset load optimization conditions among the optimizable items and are subsequently used to screen the second optimization method to adjust and optimize the unit operation strategy;

[0183] In this embodiment, the preset amount: when the load optimization coefficient is not within a preset threshold range, a threshold condition for determining whether the optimizable item can be regarded as an item to be adjusted;

[0184] In this embodiment, the preset item-quantity-method table: a table including the mapping relationship between the item to be adjusted, the adjustment amount and the second optimization method, is preset;

[0185] In this embodiment, the second optimization method is a method for adjusting and optimizing the current operation strategies of each unit.

[0186] The implementation principle and beneficial effects of the embodiments of the present invention: The present invention performs a traceability analysis on each optimizable item to obtain the data to be optimized corresponding to the optimizable item, and then performs feature extraction on the acquired first prediction curve, the first historical curve and the second prediction curve to obtain a curve feature set, and selects a corresponding first optimization method based on the curve feature set to perform optimization analysis on the data to be optimized corresponding to each optimization item, and then uses a preset evaluation model to evaluate the load optimization, and determines the optimizable items that meet the preset threshold conditions as items to be adjusted, and combines the first optimization method to obtain the adjustment amount corresponding to each adjustment item to obtain the second optimization method for adjusting and optimizing the unit operation strategy, thereby realizing continuous optimization of the unit operation strategy and ensuring stable operation of the unit.

[0187] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for evaluating load of a thermal power plant based on multi-source data, characterized in that: include: Step 1: Obtain historical load data and real-time load data of each unit in the power plant, and output historical-real-time load data; Step 2: Based on the historical-real-time load data, multi-source data is obtained in combination with the acquired other source data output; Step 3: preprocessing the multi-source data using the selected preset preprocessing method to obtain data to be analyzed; Step 4: Perform real-time evaluation and predictive analysis on the data to be analyzed in combination with a preset load evaluation method to obtain load evaluation-prediction data, and adjust and optimize the operation strategy of each unit based on the load evaluation-prediction data.

2. The method for load assessment of a thermal power plant based on multi-source data according to claim 1, characterized in that: Step 1 includes: Obtain the load data of each unit in real time through the preset monitoring system and output the real-time load data; Extracting features from the real-time load data, and selecting historical data with a matching degree greater than a first preset degree from a historical database based on the extracted features, and outputting the historical load data; Based on the real-time load data and the historical load data, historical-real-time load data is output.

3. The method for load assessment of a thermal power plant based on multi-source data according to claim 2, characterized in that: Based on the real-time load data and the historical load data, historical-real-time load data is output, including: Time-aligning the real-time load data and the historical load data within a preset time period to obtain first data; Data verification is performed on the first data at each moment in the preset time period, and the first data that passes the verification is output as historical-real-time load data.

4. The method for load assessment of a thermal power plant based on multi-source data according to claim 1, characterized in that: Step 2 includes: Based on the acquired load assessment requirement information, determine the type and accuracy requirements of the data source, and generate source data requirement information; Based on the source data demand information, other source data in the power plant are acquired using preset monitoring equipment, and time alignment of other source data is performed based on the time axis of the historical-real-time load data to construct multi-source data.

5. The method for load assessment of a thermal power plant based on multi-source data according to claim 1, characterized in that: The other source data include: meteorological data, fuel data, equipment operation data, environmental monitoring data, operation and maintenance data and market data.

6. The method for load assessment of a thermal power plant based on multi-source data according to claim 1, characterized in that: Step 3 includes: Performing content parsing on the multi-source data to obtain a first data packet corresponding to each data source in the multi-source data; Converting a first data packet under the multi-source data into a second data packet in a preset format, and performing feature extraction to obtain a first feature set; At the same time, feature matching is performed on the first feature set based on a preset abnormal feature database to obtain a first matching result corresponding to the abnormal data feature and a second matching result corresponding to the normal data feature; Based on the first matching result and in combination with a preset feature-factor adaptation table, a first matching factor corresponding to the feature in the first matching result is obtained; Inputting the first matching factor into a preset preprocessing method database for method matching, extracting preprocessing methods with a matching degree greater than a second preset degree and putting them into a preset data pool to generate a preprocessing method data pool; Based on the first matching result and the first matching factor, a preset preprocessing method corresponding to the corresponding abnormal data feature in the first matching result is selected from the preprocessing method data pool, and the corresponding multi-source data is preprocessed to obtain second data; Performing feature recognition on the second matching result, and obtaining a second feature set of the multi-source data based on the recognition result; At the same time, the first feature set and the second feature set are combined to perform feature marking on the corresponding data in the second data to obtain data to be analyzed.

7. The method for load assessment of a thermal power plant based on multi-source data according to claim 1, characterized in that: Step 4 includes: Performing data division on the data to be analyzed, and constructing load data to be analyzed and other source data to be analyzed; Extracting features of the load data to be analyzed, and selecting load assessment indicators from an indicator database to construct a first indicator set; Based on the first indicator set, in combination with a preset indicator-method mapping table, a first evaluation method having a matching degree greater than a third preset degree is selected from a method database; At the same time, feature extraction is performed on the other source data to be analyzed, and a mapping relationship between the other source data and each load assessment index in the first index set is obtained, and then a second assessment method is selected from the method database; Based on the first evaluation method, real-time analysis and predictive analysis are performed on the load data to be analyzed to generate a master data load analysis result; At the same time, the source data to be analyzed is analyzed based on the second evaluation method to generate a secondary data load analysis result; Use the preset consensus algorithm to reach consensus on the primary data load analysis results and the secondary data load analysis results in the same period, and output the load assessment-forecast data; Performing differential analysis on the load assessment-prediction data and the acquired standard control data, and outputting a load comparison result; Based on the load comparison result and in combination with the preset result-index comparison table, the index to be optimized is selected in the index database, and an optimized index set is constructed; Based on the optimization index set, optimizing and analyzing the primary data load analysis result and the secondary data load analysis result to obtain optimizable items; Based on the optimizable items, the corresponding optimization method is selected from the method database, and the operation strategy of each unit is adjusted and optimized, and an optimization log is generated.

8. The method for load assessment of a thermal power plant based on multi-source data according to claim 7, characterized in that: Based on the optimizable items, the corresponding optimization method is selected from the method database, and the operation strategy of each unit is adjusted and optimized, and an optimization log is generated, including: Conduct source tracing analysis on each optimizable item, obtain the data source of each optimizable item and the historical change data of the corresponding optimizable item, and construct the data to be optimized for the corresponding optimizable item; Obtaining the correlation between the data to be optimized and the load data, and analyzing the correlation corresponding to each optimizable item through a preset model to obtain a correlation coefficient between the corresponding optimizable item and the load; Based on the to-be-optimized data and the correlation coefficient of each optimizable item, a first prediction curve of the corresponding optimizable item in a preset future period is constructed; Based on the load data to be analyzed and the main data load analysis result, a first historical curve and a second prediction curve are generated respectively; Extracting features from the first historical curve, the first prediction curve, and the second prediction curve respectively to construct a curve feature set, and matching the first optimization method in a preset method database based on the curve feature set; In combination with the optimization index set, the first optimization method is used to optimize the data to be optimized corresponding to each optimizable item to obtain an adjustment amount and optimized data; Use the preset evaluation model to analyze the data to be optimized, the adjustment amount, and the optimized data under each optimizable item to generate the load optimization coefficient of the power plant; Among them, μ represents the load optimization coefficient of the power plant; Δαi represents the adjustment amount of the i-th optimizable item; n represents the total number of optimizable items; represents the gain coefficient of the adjustment amount Δαi of the i-th optimizable item on the power plant load; Δβ1i represents the impact value of the i-th optimizable item on the demand side of the power plant; θi represents the impact coefficient of the demand side on the load; Δβ2i represents the impact value of the i-th optimizable item on the supply side of the power plant; δi represents the impact coefficient of the supply side on the load; σ1 represents the adjustment coefficient of the external response; σ2 represents the adjustment coefficient of the internal response of the unit; If the load optimization coefficient is within the preset standard threshold range, all optimizable items are regarded as items to be adjusted; If the load optimization coefficient is not within the preset standard threshold range, the optimizable item whose adjustment amount is greater than the preset amount is extracted as the item to be adjusted; The item to be adjusted and the adjustment amount are matched with a preset item-amount-method table to obtain a corresponding second optimization method, and the obtained current operation strategy is adjusted and optimized based on the second optimization method.

Images