Online intelligent performance evaluation method for large-scale fan group

Abstract

The invention belongs to the technical field of performance analysis of large-scale wind generating set groups, and discloses an online intelligent performance evaluation method for a large-scale fan group, which comprises the following steps of: accessing operation data, fault records and operation reports of fans; cleaning and preprocessing the data; performing index calculation and statistical analysis; constructing a model algorithm according to the low-efficiency unit of the power curve; and carrying out data visualization. Normal operation of the fan can be ensured, the fan performance test can detect and analyze various performance indexes of the fan, the fault problem during operation of the fan can be found and solved in time, normal operation of the fan is ensured, and production stagnation and economic loss caused by the fault are avoided to the maximum extent.

Description

Technical Field

[0001] This application belongs to the technical field of performance analysis of large-scale wind turbine groups, and particularly relates to an online intelligent performance evaluation method for large-scale wind turbine groups. Background Art

[0002] Wind turbine performance testing is a complete process of evaluating the performance of wind turbines and the operating status of the system by testing, monitoring, and analyzing various performance indicators during the operation of wind turbines. It mainly includes detecting and analyzing relevant indicators such as wind speed, power, outdoor temperature, and fault information of wind turbines. Wind turbine performance testing plays an important role in evaluating the operating efficiency of wind turbines and the system performance, optimizing and improving the wind turbine system, and is a necessary means to improve the use efficiency of wind turbines and reduce operating costs.

[0003] The wind turbine performance evaluation based on big data mainly includes the following aspects:

[0004] (1) Data collection and preprocessing

[0005] To conduct wind turbine performance evaluation, it is first necessary to collect wind turbine operation data. These data can be obtained through means such as sensors and monitoring devices. Then, the collected data needs to be preprocessed, including noise filtering, data cleaning, and abnormal data detection, etc. Only the preprocessed data can more accurately reflect the operating status of wind turbines.

[0006] (2) Feature extraction and selection

[0007] After obtaining the preprocessed data, effective features need to be extracted from it. Feature extraction can be carried out through methods such as statistical analysis, index calculation, normalization, and standardization. The extracted features can be used to describe the status and characteristics of wind turbines under the designed standard working conditions.

[0008] (3) Establishment of wind turbine performance evaluation model

[0009] Based on the extracted and selected features, a wind turbine performance evaluation model can be established. These evaluation models include methods such as wind power curve fitting, clustering analysis, and principal component analysis. After calibrating and classifying the data under different working conditions in the statistical cycle dataset, visualization technology is used to display the evaluation results of wind turbines. The evaluation results can help operation and maintenance personnel judge the operating status of wind turbines, promptly discover faults, and take corresponding measures. Summary of the Invention

[0010] The purpose of this application is to provide an online intelligent performance evaluation method for a large-scale wind turbine group, which solves the problems that the performance evaluation of wind turbines relies on manual elimination of abnormal data, with a large workload and low efficiency, and it is impossible to achieve online evaluation of a large-scale wind turbine group. Based on the historical operation data of the wind turbine SCADA system, a complete set of data preprocessing methods are proposed for abnormal state data such as shutdown, invalidity, over-generation, under-generation, and power curtailment to mark and eliminate the original data, thereby generating the operating power curve of the wind turbine.

[0011] To achieve the above object, this application provides the following technical solutions:

[0012] An online intelligent performance evaluation method for a large-scale wind turbine group, including:

[0013] Step 1: Access the operation data, fault records, and operation reports of the wind turbines;

[0014] Step 2: Clean and preprocess the data;

[0015] Step 3: Perform index calculation and statistical analysis;

[0016] Step 4: Construct a model algorithm based on the low-efficiency units of the power curve;

[0017] Step 5: Data visualization.

[0018] In some embodiments, the operation data includes wind speed, power, and rotational speed, the fault records include fault type, time, and cause, and the operation reports include daily reports and monthly reports.

[0019] In some embodiments, the statistical analysis includes time availability analysis, lost power analysis, shutdown duration analysis, fault duration statistical analysis, and energy availability analysis.

[0020] In some embodiments, the historical power generation data and shutdown records of the wind farm are used to calculate the lost power of each type respectively.

[0021] In some embodiments, Step 4 includes:

[0022] Step 4.1: Identify the units with high lost power;

[0023] Step 4.2: Identify the units deviating from the design curve;

[0024] Step 4.3: Identify the units with deteriorated performance;

[0025] Step 4.4: For the processed low-efficiency units, evaluate the improvement of power generation using the operating power curves before and after the processing time point.

[0026] In some embodiments, based on the operating power curve, a mining index based on the difference rate of wind turbine lost power is established.

[0027] In some embodiments, the method for excavating high-loss power units includes:

[0028] Excavate high-loss power units once a week; through regular excavation, continuously track and optimize the performance of the units;

[0029] Mark the wind turbines with the daily average loss hours in the top 20% and a difference rate greater than 5% as high-loss units;

[0030] Mark the wind turbines with the daily average loss hours in the top 40% and a difference rate greater than 5% as medium-loss units.

[0031] In some embodiments, the consistency coefficient f is used to evaluate whether the operating power curve of a wind turbine deviates from the designed power curve. The consistency coefficient f of each wind turbine is defined as follows. Take the 10-minute data of the wind turbine within the statistical time period, and calculate f for the data marked as the normal state according to the following formula:

[0032]

[0033] In the formula, i is the sampling point for ten minutes; n is the number of valid data points within the time period; P d (v i ) is the actual wind speed v at the i-th sampling point i corresponding to the power of the designed power curve; p i is the actual power at the i-th sampling point.

[0034] In some embodiments, the consistency coefficient f is used to evaluate the wind turbines with deteriorated performance, and the excavation strategy is set as follows:

[0035] Calculate the consistency coefficient f of each wind turbine for 12 consecutive weeks, denoted as {F1,..., F 12}, perform linear fitting on {F1,..., F 12} using the least squares method, and use the slope of the fitting as the performance deterioration coefficient a;

[0036] Take the performance deterioration coefficients a of each wind turbine within the same model as a set, calculate the set limits [A l , A u according to the quartile method, and mark the wind turbines with a consistency coefficient less than A l as performance deteriorated units.

[0037] In some embodiments, step 4.4 includes:

[0038] Calculate the operating power curve P before processing a , and the operating power curve P after processing b ;

[0039] Take the actual wind speed and actual power experienced by the wind turbine during the evaluation period, denoted as {v i , p i};

[0040] Calculate the increase value of power generation. The power generation increase is calculated by the following formula:

[0041]

[0042] In the formula, t i refers to the duration of the i-th sampling point;

[0043] Calculate the power generation increase ratio. The power generation increase ratio is calculated by the following formula:

[0044]

[0045] In the formula, P a (v i ) is the power corresponding to the pre-treatment operation power curve at wind speed v i .

[0046] Compared with the prior art, the large-scale wind turbine group online intelligent performance evaluation method provided by the present application has the following beneficial effects:

[0047] The present application is applied to systematically analyze the power generation efficiency of wind turbines, explore the reasons for the low power generation of the units, formulate targeted treatment measures and technical transformation plans, effectively improve the power generation performance of the units, achieve power increase and efficiency improvement, and improve the operating income of power generation enterprises.

[0048] Based on the operation power curve of wind turbines, considering the randomness, complexity and normal distribution of wind turbine operation data, the present application proposes corresponding mechanism models, algorithms and evaluation methods to evaluate the performance of wind turbines and identify and analyze low-efficiency wind turbines. This method can intelligently locate and screen out low-efficiency wind turbines through the longitudinal comparison of the performance of a single wind turbine itself and the horizontal comparison of the performance of multiple wind turbines in the region. By timely finding low-efficiency wind turbines, the efficiency of troubleshooting / hidden danger elimination and inspection management is improved, and the power generation is promoted.

[0049] The present application can ensure the normal operation of wind turbines. The wind turbine performance test can detect and analyze various performance indicators of wind turbines, timely discover and solve the fault problems during the operation of wind turbines, ensure the normal operation of wind turbines, and maximize the avoidance of production stagnation and economic losses caused by faults.

[0050] The present application improves the efficiency of the wind turbine system. The wind turbine performance test can measure the operation efficiency of the wind turbine system. By optimizing and improving the relevant parameters of the wind turbine system, the maximum efficiency of the wind turbine system is achieved, and the efficiency and performance of the wind turbine are improved. Description of the Drawings

[0051] In order to more clearly illustrate the technical solution of the present application, the following is a brief introduction to the drawings required for the technical description.

[0052] Figure 1 A flow chart of the online intelligent performance evaluation method for a large-scale wind turbine group provided for this application;

[0053] Figure 2 A schematic diagram of the implementation flow of the large-scale wind turbine group online intelligent performance evaluation method provided in this application. DETAILED DESCRIPTION

[0054] The following is further explained in detail through specific implementation methods.

[0055] like Figure 1 and Figure 2 As shown, the present application provides a large-scale wind turbine group online intelligent performance evaluation method, which discovers inefficient units and inefficiency reasons based on indicator calculation and statistical analysis and the design and establishment of an inefficient unit mining model based on a power curve. The method includes:

[0056] Step 1: Data access, the large-scale wind turbine group online intelligent performance evaluation data analysis system accesses the wind turbine's operating data, fault records, operating reports, etc. The operating data includes wind speed, power, speed, etc., the fault records include fault type, time, cause, etc., and the operating reports include daily reports, monthly reports, etc.

[0057] Step 2: Data cleaning and preprocessing, abnormal data identification uses physical rule identification, cluster analysis, quartile method, and data statistical analysis. Data interpolation uses methods such as "multi-point cubic spline interpolation method", "operating wind power curve", "sample wind turbine", and "average power generation of the entire field".

[0058] Physical rule recognition determines whether the data is reasonable based on the laws of physics; cluster analysis divides the data into multiple groups and identifies abnormal points that do not match the majority of the data; the quartile method calculates the quartiles of the data and identifies abnormal values that exceed the interquartile range; data statistical analysis, such as mean, variance and other statistical quantities, determines whether the data deviates from the normal range.

[0059] The multi-point cubic spline interpolation method uses known data points to construct a cubic polynomial for interpolation; the operating wind power curve is interpolated according to the relationship between wind speed and power; the model wind turbine selects a wind turbine with good performance as a model and interpolates according to the data of the model wind turbine; the average power generation of the entire wind farm is interpolated according to the average power generation of the entire wind farm.

[0060] Data standardization is to convert the measured wind speed into the wind speed under the standard air density in combination with the actual local air density. The data status includes: shutdown data, under-generation data, over-generation data, power curtailment data, and normal data. Methods such as the quartile method, quadratic clustering, and physical rules can be used for identification, classification, and annotation.

[0061] Step 3: Index calculation and statistical analysis. Statistical analysis includes time availability analysis, lost electricity analysis, shutdown duration analysis, fault duration statistical analysis, energy availability analysis, and other index analyses.

[0062] (1) Time availability analysis

[0063] The availability of a wind turbine is an index that describes the proportion of the time the unit is in an available state during the statistical period. The available hours are the sum of the available hours.

[0064] Among them, the time availability = (the time the unit is in an available state during the statistical period / the total time of the statistical period) × 100%.

[0065]

[0066]

[0067] The scope of fault shutdown includes:

[0068] 1) Unscheduled shutdowns caused by faults or alarms;

[0069] 2) Unscheduled shutdowns caused by major component failures or replacements;

[0070] The scope of scheduled shutdowns includes:

[0071] 1) Scheduled shutdowns that are not regular inspections;

[0072] 2) Scheduled shutdowns for regular inspections;

[0073] 3) Shutdowns for technical transformation;

[0074] The shutdown time caused by the following situations should not be included in the fault shutdown:

[0075] 1) Shutdowns caused by force majeure, such as when meteorological conditions (including ambient temperature, freezing, typhoon, etc.) exceed the designed operating conditions of the unit, and the time when the equipment enters protective shutdown;

[0076] 2) Shutdowns caused by off-site influence, such as grid faults / fluctuations (grid parameters are outside the technical specifications of the wind turbine, such as grid voltage and frequency fluctuations), grid power curtailment shutdowns, maintenance of substations on the opposite side (not the on-site booster station), maintenance of outgoing lines, and off-site influence shutdowns for temporary dispatching coordination with grid tests and defect elimination;

[0077] 3) Note: Some of the outgoing lines of the wind farm are self-owned assets. The maintenance of the outgoing lines is affected within the farm, and it should be treated differently according to the actual situation;

[0078] 4) The shutdown caused by being affected within the farm, such as the shutdown caused by the failure or regular inspection of on-site power transmission and distribution equipment such as box transformers, collector lines, and booster stations.

[0079] (2) Analysis of lost electricity

[0080] The lost electricity counted by this method includes the following types: lost electricity due to fault shutdown, lost electricity due to planned shutdown, lost electricity due to affected shutdown, and lost electricity due to dispatching power rationing.

[0081] In one embodiment, using the historical power generation data and shutdown records of the wind farm, calculate the lost electricity of each type respectively, and analyze its proportion in the total lost electricity.

[0082] (3) Analysis of shutdown duration

[0083] The shutdown types include fault shutdown, planned shutdown, and affected shutdown. Count the duration of each type of shutdown, and analyze its trend and periodicity.

[0084] (4) Statistical analysis of fault duration

[0085] Count the changes in the shutdown duration of each cycle type and the lost power generation caused by fault shutdown. Use the fault record data to analyze the frequency, duration, and losses caused by faults.

[0086] (5) Analysis of energy availability

[0087] The energy availability is equal to the ratio of the actual power generation to the theoretical power generation, which is an important parameter for the power generation performance of the wind turbine. During the operation of the wind turbine, wind turbine faults, wind turbine maintenance, substation faults and tests, grid power rationing, power curves, etc. will all cause the difference between the theoretical power generation and the actual power generation. Based on the large-scale wind turbine group online intelligent performance evaluation system, accurately count and analyze the power generation losses of the wind turbines caused by the above problems, and provide a basis for improving the power generation efficiency of the wind turbine units and the profits of the wind farm.

[0088] Among them, energy availability = (actual power generation / theoretical power generation) × 100%.

[0089] In one embodiment, according to the wind speed data and the unit performance parameters, calculate the theoretical power generation, compare it with the actual power generation, and analyze the difference in energy availability.

[0090] (6) Analysis of other indicators

[0091] Other indicators include power generation and equipment availability. Through horizontal and vertical benchmarking of production operation indicators, they are applied at different management levels of power generation enterprises. On the one hand, problems can be discovered through indicator statistics, and on the other hand, benchmarking assessment of production operation indicators among different levels can be carried out.

[0092] In one embodiment, through horizontal (compared with other wind farms in the same industry) and vertical (compared with its own historical data) comparison of production operation indicators, problems are discovered and benchmarking assessment is carried out.

[0093] Step 4: Construct a model algorithm based on low-efficiency units in the power curve, specifically including:

[0094] Step 4.1: Identify units with high power loss.

[0095] Step 4.2: Identify units deviating from the design curve.

[0096] Step 4.3: Identify units with deteriorating performance.

[0097] Step 4.4: Evaluation method for power generation improvement. For the processed low-efficiency units, the power generation improvement is evaluated by using the operation power curves before and after the processing time point.

[0098] Low-efficiency units include three forms: 1. Units with high power loss; 2. Units deviating from the design curve; 3. Units with deteriorating performance. To verify the reliability of the algorithm, after these three anomalies are discovered, the site is notified to require on-site rectification. After the rectification is completed, the accuracy of the algorithm and the effectiveness of on-site rectification are verified through the power generation improvement evaluation method.

[0099] In Step 4.1, based on the operation power curve, a mining index based on the difference rate of power loss of the fan is established. After 0:00 on Monday every week, the large-scale online intelligent performance evaluation data analysis system of the fan group automatically mines the high-loss units of the whole company in the last week according to the latest mining conditions. The definition of the difference rate of power loss is as follows:

[0100]

[0101] In the formula:

[0102] D: Difference rate of power loss;

[0103] Y t : Daily average theoretical power generation hours;

[0104] Y loss : Daily average power loss hours;

[0105] E t : Total theoretical power generation;

[0106] E loss:Total lost power;

[0107] P rated : Rated power;

[0108] Days: Number of days within the statistical period.

[0109] The theoretical power generation and lost power are calculated based on the operating power curve of the wind turbine, using the wind speed measured by the anemometer of the wind turbine. When the wind turbine has no communication or fails to stop with no wind speed signal, the average wind speed of the whole field is used for calculation.

[0110] The process of excavating high lost-power units is as follows. High lost-power units are a form of inefficient units. The purpose is to discover inefficient units, as follows:

[0111] 1) Excavate high lost-power units once a week; through regular excavation, continuously track and optimize the performance of the units;

[0112] 2) Mark the wind turbines with the daily average lost hours in the top 20% and a difference rate greater than 5% as high lost-power units; the daily average lost hours in the top 20% indicates that the power loss of these units is relatively large, and a difference rate greater than 5% confirms the significant difference between these units and the average level. Mark the wind turbines that meet the screening criteria as high lost-power units for subsequent key analysis and processing;

[0113] 3) Mark the wind turbines with the daily average lost hours in the top 40% and a difference rate greater than 5% as intermediate lost-power units, but do not process the wind turbines that have been marked as high lost-power units.

[0114] During the excavation process, it is necessary to exclude the wind turbines that have been marked as long-term shutdown units. Due to long-term shutdown, the power loss situation of long-term shutdown units is no longer the focus of current attention. These units need to be excluded during the excavation process to avoid interference with the results.

[0115] Through the excavation of high lost-power units, identify the units with large power loss and low operating efficiency, and conduct in-depth analysis and optimization processing on them to improve the operation efficiency and economic benefits of the entire wind farm.

[0116] In step 4.2, the consistency coefficient f is used to evaluate whether the operating power curve of the wind turbine deviates from the design power curve. The consistency coefficient f of each wind turbine is defined as follows. Take the 10-minute data of the wind turbine within the statistical period, and calculate f for the data marked as normal state according to the following formula:

[0117]

[0118] In the formula:

[0119] i is the sampling point every ten minutes;

[0120] n is the number of valid data points within a time period;

[0121] P d (v i ) is the actual wind speed v at the i-th sampling point i corresponding to the power of the designed power curve;

[0122] p i is the actual power at the i-th sampling point.

[0123] The designed excavation strategy is as follows: Calculate the consistency coefficient f of all wind turbines. Taking the consistency coefficient f of each wind turbine within the same model as a set, calculate the set limits [F l , F u according to the quartile method. Mark the wind turbines with a consistency coefficient less than F l as the units deviating from the design curve. Similarly, during the excavation process, the wind turbines that have been marked as long-term shutdown units need to be excluded.

[0124] In step 4.3, the consistency coefficient f is used to evaluate the wind turbine units with performance degradation. Set the excavation strategy as follows: Calculate the consistency coefficient f of each wind turbine for 12 consecutive weeks, denoted as {F1,..., F 12}}, perform linear fitting on {F1,..., F 12} using the least squares method, and use the slope of the fitting as the performance degradation coefficient α. Taking the performance degradation coefficient α of each wind turbine within the same model as a set, calculate the set limits [A l , A u according to the quartile method. Mark the wind turbines with a consistency coefficient less than A l as the performance degradation units. Similarly, during the excavation process, the wind turbines that have been marked as long-term shutdown units need to be excluded.

[0125] In step 4.4, the evaluation methods include the following:

[0126] (1) Calculate the pre-treatment operating power curve P a , and the post-treatment operating power curve P b ;

[0127] (2) Take the actual wind speed and actual power experienced by the wind turbine during the evaluation period, denoted as {v i , p i};

[0128] (3) Calculate the power generation improvement value. The power generation improvement is calculated using the following formula:

[0129]

[0130] where: t i refers to the duration of the i-th sampling point;

[0131] (4) Calculate the power generation increase ratio, which is calculated using the following formula:

[0132]

[0133] In the formula: P b (v i ) is the power corresponding to the pre - processing operating power curve at wind speed v. i

[0134] (5) Exclude the time periods of fan failures and fan shutdowns during the evaluation period.

[0135] Step 5: Data visualization. Display the index calculation results and the mining model recognition results on a web page. This includes chart display, interactive interface, report generation, etc.

[0136] 1. Graphical visualization: scatter plot, line chart, histogram, bar chart, meter chart, time - series chart, etc.

[0137] 2. Interactive interface: data query, data download.

[0138] Through intuitive charts, the distribution, trend, and correlation of data can be quickly understood, thus enabling accurate performance evaluation. Display the data after cleaning and pre - processing, the calculated performance indicators, the results of data mining, etc. in the form of charts, such as line charts, bar charts, scatter plots, box plots, etc. For example, a box plot shows the distribution characteristics of data, including the median, quartiles, and outliers.

[0139] Ensure the usability of the system through the interactive interface, enabling users to conveniently view and analyze data for performance evaluation.

[0140] Automatically generate a performance evaluation report according to the user's needs, including the results of data cleaning and pre - processing, the calculation results of performance indicators, the results of data mining, etc.

[0141] The large - scale fan group online intelligent performance analysis system accesses data such as the fan operation data, fan alarm data, and wind farm reported power generation in the centralized control center, cleans and filters the data, calculates all the indicators related to the power generation, loss power, and performance of the wind farm and the units, and helps the operation and analysis personnel to more comprehensively, deeply, and professionally master the operation efficiency of the wind farm units through the data analysis method of the inefficient unit mining model.

[0142] ​This application uses methods such as cluster analysis, association analysis, and power curve fitting, which helps to better understand the internal laws and characteristics of data, and reveals the temporal, random, multivariate, and correlation characteristics of basic data such as wind speed, wind direction, rotational speed, and power. Through data monitoring and analysis technologies, the operation of the wind turbine can be analyzed and evaluated, and corresponding control measures can be taken to ensure the stable operation and power generation efficiency of the wind turbine.

[0143] The above are only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present application should be covered by the protection scope of the present application.

Claims

1. An online intelligent performance evaluation method for a large-scale wind turbine group, characterized in that, Including: Step 1: Access the operation data, fault records, and operation reports of the wind turbines; Step 2: Clean and preprocess the data; Step 3: Conduct index calculation and statistical analysis; Step 4: Construct a model algorithm for inefficient units based on the power curve; Step 5: Data visualization.

2. The online intelligent performance evaluation method for large-scale wind turbine groups according to claim 1, wherein The operation data includes wind speed, power, and rotation speed. The fault records include fault type, time, and cause. The operation reports include daily reports and monthly reports.

3. The online intelligent performance evaluation method for large-scale wind turbine groups according to claim 1, characterized in that The statistical analysis includes time availability analysis, lost electricity analysis, downtime analysis, fault duration statistical analysis, and energy availability analysis.

4. The online intelligent performance evaluation method for large-scale wind turbine groups according to claim 3, characterized in that, Use the historical power generation data and downtime records of the wind farm to calculate the lost electricity of each type respectively.

5. The online intelligent performance evaluation method for large-scale wind turbine groups according to claim 1, characterized in that, Step 4 includes: Step 4.1: Identify the units with high lost electricity; Step 4.2: Identify the units that deviate from the design curve; Step 4.3: Identify the units with deteriorated performance; Step 4.4: For the processed inefficient units, evaluate the improvement of power generation using the operation power curves before and after the processing time point.

6. The online intelligent performance evaluation method for large-scale wind turbine groups according to claim 5, characterized in that In Step 4.1, based on the operation power curve, establish a mining index based on the difference rate of lost electricity of the wind turbines.

7. The online intelligent performance evaluation method for large-scale wind turbine groups according to claim 5, wherein The process of mining the units with high lost electricity includes: Conduct mining of the units with high lost electricity once a week; through regular mining, continuously track and optimize the performance of the units; Mark the wind turbines with the daily average lost hours in the top 20% and a difference rate greater than 5% as high-loss units; Mark the wind turbines with the daily average lost hours in the top 40% and a difference rate greater than 5% as medium-loss units.

8. The online intelligent performance evaluation method for large-scale wind turbine groups according to claim 5, characterized in that In Step 4.2, use the consistency coefficient f to evaluate whether the operation power curve of the wind turbine deviates from the design power curve. The consistency coefficient f of each wind turbine is defined as follows. Take the 10-minute data of the wind turbine within the statistical time period and calculate f for the data marked as normal state according to the following formula: Where, i is the sampling point for ten minutes; n is the number of valid data points within the time period; P d (v i ) is the actual wind speed v i corresponding to the power of the designed power curve; p i is the actual power at the i-th sampling point.

9. The online intelligent performance evaluation method for large-scale wind turbine groups according to claim 5, characterized in that In Step 4.3, use the consistency coefficient f to evaluate the wind turbines with deteriorated performance and set the following mining strategy: Calculate the consistency coefficient f for each fan for 12 consecutive weeks, denoted as {F1, …, F 12}, and perform linear fitting on {F1, …, F 12} using the least squares method. The slope of the fitting is used as the performance degradation coefficient a; Taking the performance degradation coefficient a of each fan within the same model as a set, calculate the inner limit [A of the set according to the quartile method l , A u , and mark the fans with a consistency coefficient less than A l as performance degradation units.

10. The online intelligent performance evaluation method for large-scale wind turbine groups according to claim 5, characterized in that, Step 4.4 includes: The operating power curve P before calculation and processing a , and the operating power curve P after processing b ; Take the actual wind speed and actual power experienced by the wind turbine during the evaluation time period, denoted as {v i , p i}; Calculate the power generation improvement value. The power generation improvement is calculated using the following formula: where t i denotes the duration of the i-th sampling point; Calculate the power generation improvement ratio. The power generation improvement ratio is calculated using the following formula: Where P a (v i ) is the wind speed v i corresponding to the power of the operating power curve before treatment.

Images