Comprehensive evaluation system for multi-dimensional operation indexes of wind power plant

Abstract

The invention relates to the technical field of data analysis, in particular to a wind power plant multi-dimensional operation index comprehensive evaluation system which comprises a deviation benchmark construction module, a gradient environment mapping module, a torque ripple calculation module, a dynamic interval construction module and an index generation module. In the method, a dynamic power confidence interval changing along with the environment in real time is generated by constructing a gradient vector containing a vertical height layer air density difference, finely simulating a torque pulsation waveform generated by environmental heterogeneity in a blade rotation period, quantifying a theoretical power fluctuation amplitude caused by the torque pulsation waveform, and superposing the theoretical power fluctuation amplitude to a reference error band; according to the method, a running state judgment boundary with more physical interpretation is established, meanwhile, a sensitivity weight is introduced by combining statistical morphological characteristics of wind energy flow distribution, a physical yaw angle is corrected and weighted, and multi-dimensional coupling evaluation of wind energy capturing efficiency and control precision of a unit on the basis of eliminating reasonable environmental disturbance is achieved.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to a comprehensive evaluation system for multi-dimensional operation indicators of wind farms. Background Technology

[0002] The field of data analytics involves processing massive and diverse information assets. Through data collection, storage management, and analysis and mining, valuable patterns are extracted from messy data to assist in decision-making and business optimization. This field covers the entire process from the underlying database construction to the upper-level visualization display.

[0003] Among them, the wind farm multi-dimensional operation index comprehensive evaluation system mainly relies on viewing the real-time power, wind speed and fault logs of wind turbines recorded by the data acquisition and monitoring control system to calculate the total power generation, average wind speed and equipment downtime of the wind farm, and directly compare the statistical values ​​with the standard values ​​in the wind turbine design specifications or historical records of previous years, and judge whether each indicator is qualified based on the set fixed value range.

[0004] Existing technologies rely solely on a single static design standard or historical averages for comparison, neglecting the non-uniform distribution of meteorological parameters in the vertical space of the actual wind farm operating environment. In particular, they fail to consider the real-time pulsating impact of air density changes with altitude gradients on the rotational forces of long blades. This coarse judgment method based on a fixed numerical range cannot accurately isolate reasonable power fluctuations caused by environmental factors, and it is difficult to effectively distinguish the essential differences between the unit's own performance degradation and the limitations of external environmental conditions. As a result, the assessment of yaw control deviation lacks physical mechanism support, which can easily lead to misjudgments in assessment conclusions and reduce the pertinence of operation and maintenance strategy formulation. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a comprehensive evaluation system for multi-dimensional operation indicators of wind farms.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a comprehensive evaluation system for multi-dimensional operational indicators of wind farms includes: The deviation from the baseline construction module calculates the wind energy flow distribution histogram based on the current wind speed and direction, determines the optimal wind angle of the wind energy flow distribution histogram, and calculates the deviation sensitivity weight. The gradient environment mapping module constructs an air density gradient vector based on the current air pressure and temperature, and simultaneously determines the blade's spatial coordinates under the azimuth step size. The torque pulsation calculation module extracts the local air density from the air density gradient vector based on the blade spatial coordinates, calculates the theoretical torque waveform by combining the currently collected inflow wind speed and aerodynamic parameters, and calculates the density-induced torque fluctuation amplitude based on the extreme value of the theoretical torque waveform. The dynamic interval construction module combines the density-induced torque fluctuation amplitude with the collected wind turbine reference error band width to construct a dynamic power confidence interval, compares it with the current wind turbine output power data and speed data, and generates a power status label. The index generation module calculates a weighted yaw deviation value based on the difference between the cabin orientation angle and the optimal windward angle and the deviation sensitivity weight, and matches the weighted yaw deviation value with the power status label to generate a multi-dimensional operation evaluation index.

[0007] As a further aspect of the present invention, the deviation sensitivity weight is specifically a quantized coefficient determined by retrieving a preset coefficient weight mapping table based on the kurtosis coefficient of the wind energy flow distribution histogram; the air density gradient vector is specifically the air density value arranged in vertical height order and corresponding to each height layer within the swept surface of the wind turbine rotor; the blade spatial coordinates include the discrete azimuth angle position of the wind turbine blade on the rotation circumference and the corresponding height layer index in the vertical direction; the density-induced torque fluctuation amplitude is specifically the difference calculated based on the peak and valley data of the theoretical torque waveform within a single rotation cycle; the power status label is specifically a logical status identifier generated based on the real-time operating point of the wind turbine and the dynamic power confidence interval; and the multi-dimensional operation evaluation index includes a weighted yaw deviation value combined with the degree of wind direction dispersion and a power status judgment result characterizing the conformity of the power curve.

[0008] As a further aspect of the present invention, the deviation from the benchmark construction module includes: The histogram construction submodule collects wind speed and wind direction data within the target sector of the wind farm, divides the wind speed and wind direction data into statistical intervals to obtain wind direction frequency, maps the wind speed magnitude to each wind direction frequency interval for weighted accumulation, reflects the distribution pattern of wind energy density in multiple directions, and obtains a wind energy flow distribution histogram. The optimal parameter extraction submodule locates the density peak position in the wind energy flow distribution histogram and determines the optimal wind angle through the mean drift algorithm. At the same time, it obtains the distribution pattern value of the wind energy flow distribution histogram and calculates the fourth standard moment of the distribution pattern value as the wind energy flow distribution kurtosis coefficient. The weight coefficient quantization submodule presets multiple numerical segment intervals of the wind energy flow distribution kurtosis coefficient, constructs a coefficient weight mapping table by matching the corresponding evaluation weight for each numerical segment interval, retrieves the current wind energy flow distribution kurtosis coefficient in the coefficient weight mapping table, matches the weight value corresponding to the numerical segment interval, and outputs the deviation sensitivity weight.

[0009] As a further aspect of the present invention, the gradient environment mapping module includes: The gradient vector construction submodule collects air pressure and temperature data from the lidar at multiple height levels within the swept surface of the wind turbine rotor, calculates the air density corresponding to each height level, arranges the air density in order of vertical height and records the height level index, and obtains the air density gradient vector. The circumference step size discrete submodule obtains the geometric parameters of the wind turbine rotor rotation circumference, divides the wind turbine rotor rotation circumference into multiple discrete azimuth step sizes, defines the angle position corresponding to each azimuth step size, and generates the corresponding azimuth step size sequence. The spatial coordinate positioning submodule, in conjunction with the azimuth step sequence and the obtained blade length and hub height parameters of the wind turbine, calculates the three-dimensional spatial position of the blade tip at each azimuth step and determines the blade spatial coordinates of the wind turbine blade relative to the ground within the rotation circumference.

[0010] As a further aspect of the present invention, the torque ripple calculation module includes: The density parameter extraction submodule locates the corresponding height layer index in the air density gradient vector based on the vertical component of the spatial coordinates of the wind turbine blades, extracts the local air density at each coordinate point according to the height layer index, and arranges the local air density in the order of azimuth step size to generate a local air density sequence. The torque waveform calculation submodule obtains the aerodynamic parameters of the wind turbine blades and the real-time collected inflow wind speed. It combines the local air density sequence to simulate the force fluctuations caused by changes in air density during the rotation of the blades and calculates the torque waveform at each azimuth step. The fluctuation amplitude calculation submodule extracts the peak and valley data of the torque waveform within a single rotation cycle, performs difference calculation on the peak and valley data, quantifies the torque change range, and obtains the density-induced torque fluctuation amplitude.

[0011] As a further aspect of the present invention, the dynamic interval construction module includes: The dynamic boundary setting submodule obtains the reference error band width of the standard power curve of the wind turbine, converts the density-induced torque fluctuation amplitude into the corresponding power fluctuation amplitude, and superimposes it onto the reference error band width. It then calculates the upper and lower limits of the fluctuation of the standard power curve of the wind turbine under the current environmental density gradient and determines the dynamic boundary values. The confidence interval generation submodule performs bidirectional numerical bias calculations on the standard power curve of the wind turbine based on the dynamic boundary values, constructs dynamic power envelopes on both sides of the standard power curve of the wind turbine, defines the distribution range of the wind turbine power data at multiple speeds based on the dynamic power envelopes, and generates dynamic power confidence intervals. The status label determination submodule collects real-time output power and speed data of the wind turbine, constructs the current operating point, and determines that the wind turbine is in a normal wind energy capture state if the current operating point is within the dynamic power confidence interval, and generates a power status label.

[0012] As a further aspect of the present invention, the indicator generation module includes: The angle deviation calculation submodule collects the current orientation angle of the wind turbine nacelle, calculates the numerical difference between the current orientation angle of the wind turbine nacelle and the optimal wind angle, quantifies the degree of deviation between the current orientation of the nacelle and the maximum wind energy capture direction of the optimal wind angle, and obtains the angle deviation value. The weighted deviation quantization submodule performs a weighted operation on the deviation sensitivity weight and the angle deviation value to obtain the weighted yaw deviation value. The evaluation index synthesis submodule associates and matches the weighted yaw deviation value with the power status label, and rates the yaw deviation in conjunction with the power output status to reflect the unit's yaw control performance and obtain multi-dimensional operation evaluation indexes.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by constructing a gradient vector that includes the differences in air density at vertical height levels, the torque pulsation waveform caused by environmental non-uniformity during the blade rotation cycle is simulated in a refined manner. The theoretical power fluctuation amplitude caused by this is quantified and superimposed onto the benchmark error band, generating a dynamic power confidence interval that changes with the environment in real time. This establishes a more physically interpretable boundary for determining the operating state. At the same time, by introducing sensitivity weights in combination with the statistical morphological characteristics of wind energy flow distribution, the physical yaw angle is corrected and weighted. This achieves a multi-dimensional coupled evaluation of the unit's wind energy capture efficiency and control accuracy based on the elimination of reasonable environmental disturbances, significantly improving the robustness of unit performance diagnosis and the scientific nature of operation and maintenance decisions. Attached Figure Description

[0014] Figure 1 This is a system flowchart of the present invention; Figure 2 This is a flowchart of the deviation from the baseline construction module of the present invention; Figure 3 This is a flowchart of the gradient environment mapping module of the present invention; Figure 4 This is a flowchart of the torque ripple calculation module of the present invention; Figure 5 This is a flowchart of the dynamic range construction module of the present invention; Figure 6 This is a flowchart of the indicator generation module of the present invention. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0016] Please see Figure 1 The comprehensive evaluation system for multi-dimensional operation indicators of wind farms includes: The deviation from the baseline construction module calculates the wind energy flow distribution histogram based on the current wind speed and direction, determines the optimal wind angle of the wind energy flow distribution histogram, and calculates the deviation sensitivity weight. The gradient environment mapping module constructs an air density gradient vector based on the current air pressure and temperature, and simultaneously determines the blade's spatial coordinates under the azimuth step size. The torque pulsation calculation module extracts the local air density from the air density gradient vector based on the blade spatial coordinates, calculates the theoretical torque waveform by combining the currently collected inflow wind speed and aerodynamic parameters, and calculates the density-induced torque fluctuation amplitude based on the extreme value of the theoretical torque waveform. The dynamic range construction module combines the density-induced torque fluctuation amplitude with the collected wind turbine reference error band width to construct a dynamic power confidence range, which is then compared with the current wind turbine output power data and speed data to generate a power status label. The indicator generation module calculates the weighted yaw deviation value based on the difference between the cabin orientation angle and the optimal wind angle and the deviation sensitivity weight, and matches the weighted yaw deviation value with the power status label to generate multi-dimensional operation evaluation indicators.

[0017] The deviation sensitivity weight is specifically a quantified coefficient determined by retrieving the kurtosis coefficient of the wind energy flow distribution histogram from a preset coefficient weight mapping table. The air density gradient vector is specifically the air density value arranged in vertical order and corresponding to each height layer within the swept surface of the wind turbine rotor. The blade spatial coordinates include the discrete azimuth angle position of the wind turbine blade on the rotation circumference and the corresponding height layer index in the vertical direction. The density-induced torque fluctuation amplitude is specifically the difference calculated based on the peak and valley data of the theoretical torque waveform in a single rotation cycle. The power status label is specifically a logical status identifier generated based on the real-time operating point of the wind turbine and the dynamic power confidence interval. The multi-dimensional operation evaluation index includes a weighted yaw deviation value combined with the degree of wind direction dispersion and a power status judgment result characterizing the conformity of the power curve.

[0018] Please see Figure 2 The deviation from the baseline building block includes: The histogram construction submodule collects wind speed and wind direction data within the target sector of the wind farm, divides the wind speed and wind direction data into statistical intervals to obtain wind direction frequency, maps the wind speed magnitude to each wind direction frequency interval for weighted accumulation, reflects the distribution pattern of wind energy density in multiple directions, and obtains a wind energy flow distribution histogram. Ultrasonic sensors deployed on wind farm anemometer towers collect wind speed and direction data within the target sector at a preset high-frequency sampling rate. Wind speed data originates from the sensors' real-time capture of airflow velocity, while wind direction data comes from the sensors' sensing of the incoming airflow direction. The acquired raw data sequence undergoes a cleaning process to remove invalid data generated by sensor malfunctions. Subsequently, a statistical interval division is performed, dividing the 360-degree azimuth angle on the horizontal plane into 72 continuous and non-overlapping sectors, each sector corresponding to an angle span of 5 degrees. For each divided wind direction sector, all wind speed sample points falling within the angular range of that sector are selected. To accurately reflect wind energy density, the wind speed data undergoes a cubic operation, and the cube values ​​of all wind speed samples within the sector are summed to obtain the cumulative wind energy for that azimuth. This cumulative amount is then divided by the total number of samples within the sector to obtain the weighted wind energy density index for that wind direction sector. For example, within a sector ranging from 0 degrees due north to 5 degrees north of east, three sets of wind speed data were collected: 5 m / s, 8 m / s, and 10 m / s. Calculate the cube value for each of these three sets of data: , , Then, the cube values ​​are accumulated: Finally, this accumulated value is mapped to the interval, forming the bar height of that angle interval in the histogram.

[0019] The optimal parameter extraction submodule locates the density peak position in the wind energy flow distribution histogram and determines the optimal wind angle through the mean drift algorithm. At the same time, it obtains the distribution pattern values ​​of the wind energy flow distribution histogram and calculates the fourth standard moment of the distribution pattern values ​​as the wind energy flow distribution kurtosis coefficient. The system receives a dataset of wind energy flow distribution histograms and uses a mean-shift algorithm to iteratively optimize the density distribution within the histogram. This process initializes a kernel window with a preset bandwidth, randomly centers the window at a position on the histogram, calculates the probability density-weighted mean of the data points within the window, and moves the window center towards this mean position. This process is repeated until the window center converges to the position of maximum density; the angle corresponding to this convergence position is determined as the peak position of wind energy flow density. The angle value corresponding to this peak position is then locked as the optimal windward angle. After determining the peak position, the overall distribution pattern values ​​of the histogram are extracted, including the frequency and amplitude information of each sector, and the fourth standard moment of the distribution pattern values ​​is calculated based on these values. The calculation formula is as follows: In the formula, The fourth standard moment representing the numerical distribution pattern is used as the kurtosis coefficient of wind energy flow distribution. The total number of sectors represents the total number of sectors, which is determined by the settings when the histogram is constructed. Index markers indicating sectors; Indicates the first The wind energy density values ​​for each sector are directly extracted from histogram data. This represents the average wind energy density across all sectors. This represents the standard deviation of wind energy density distribution; This represents the data quality weighting coefficient, which is set based on the proportion of valid data in the real-time status log of the data acquisition system. The higher the proportion of valid data, the closer the weight is to 1. It is used to correct distribution deviations caused by sensor aging or missing data. This represents the baseline value for the kurtosis of the normal distribution. It is based on the fourth central moment value of the standard normal distribution in statistics and is usually constant at 3. This parameter is introduced to calculate the excess kurtosis, so that the kurtosis coefficient of the standard normal distribution is zero, thereby more intuitively judging whether the wind energy flow distribution is steeper or flatter than the normal distribution.

[0020] The parameters are set as follows: Total number of sectors Set to 72; Data quality weighting coefficient Based on the log setting of 1; normal distribution kurtosis baseline value Set to 3; assume the calculated average value The standard deviation is 500. The value is 100. This is to calculate the summation term. This requires traversing all 72 sectors. Three typical sectors are selected to illustrate the calculation process: 1. Peak sector: wind energy density value. The value is 800. The standardized and fourth-power calculation is as follows: 2. Mean sector: wind energy density value The value is 500. The standardized and fourth-power calculation is as follows: 3. Edge sector: wind energy density value The value is 200. The standardized and fourth-power calculation is as follows: The values ​​for the remaining 69 sectors are calculated using the same logic, assuming the sum of all other terms is 162. The overall calculation process for the summation term is as follows: Finally, substitute all parameters into the main formula to calculate: The calculated result 1.5 is the fourth standard moment of the distribution pattern, which is used as the kurtosis coefficient of the wind energy flow distribution.

[0021] The weight coefficient quantification submodule presets multiple numerical segment intervals of the wind energy flow distribution kurtosis coefficient, matches the corresponding evaluation weight for each numerical segment interval to construct a coefficient weight mapping table, retrieves the current wind energy flow distribution kurtosis coefficient in the coefficient weight mapping table, matches the weight value corresponding to the numerical segment interval, and outputs the deviation sensitivity weight. Based on the analysis of historical long-term wind data, multiple numerical intervals for the kurtosis coefficient of wind energy flow distribution are predefined. The boundary values ​​for these intervals are set according to the probability distribution of the kurtosis coefficient in the past three years of operational data, selecting the 25th, 50th, and 75th percentiles as segment thresholds. The interval determination for this coefficient includes three possibilities: the first possibility is a kurtosis coefficient less than 0.5, indicating a relatively dispersed wind direction distribution; the second possibility is a kurtosis coefficient between 0.5 and 1.5, indicating a moderate concentration of wind direction distribution; and the third possibility is a kurtosis coefficient greater than 1.5, indicating a highly concentrated wind direction distribution. For each numerical interval, a corresponding evaluation weight is assigned. The weight setting principle follows that the larger the kurtosis coefficient, the more obvious the dominant wind direction, and the more significant the impact of yaw deviation on power generation, thus assigning a higher weight value. For example, the first interval corresponds to a weight of 0.8, the second interval to a weight of 1.0, and the third interval to a weight of 1.2.

[0022] Please see Figure 3 The gradient environment mapping module includes: The gradient vector construction submodule collects air pressure and temperature data from the lidar at multiple height levels within the swept surface of the wind turbine rotor, calculates the air density corresponding to each height level, arranges the air density in order of vertical height and records the height level index, and obtains the air density gradient vector. A wind-measuring lidar installed on top of the wind turbine nacelle emits a laser beam towards the swept surface of the rotor. Based on the Doppler effect, the echo signal is analyzed to collect meteorological data at different heights within the swept surface. The collected heights cover the vertical range from the lowest to the highest point of the rotor, defined at five characteristic heights: 30 meters, 60 meters, 90 meters, 120 meters, and 150 meters above the ground. At each height, the lidar, in conjunction with temperature and pressure sensors, collects real-time pressure and temperature data. The pressure data is obtained from direct measurements by the barometer, and the temperature data from the temperature sensor readings. The data for each height is processed according to the ideal gas law. The air pressure at the current height is divided by the product of the gas constant and the absolute temperature to calculate the air density at that height. For example, at a height of 150 meters, the collected pressure is 98,000 Pascals, the temperature is 15 degrees Celsius (i.e., 288.15 Kelvin), and the gas constant is taken as 287 joules per kilogram Kelvin. The calculation process is as follows: , The same calculation process is performed on all preset height layers, and the calculated air density values ​​are arranged strictly in ascending order of vertical height. At the same time, the height layer index corresponding to each value is recorded to form an ordered air density gradient vector.

[0023] The circumference step size discrete submodule obtains the geometric parameters of the wind turbine rotor rotation circumference, divides the wind turbine rotor rotation circumference into multiple discrete azimuth step sizes, defines the angle position corresponding to each azimuth step size, and generates the corresponding azimuth step size sequence. The geometric parameters of the wind turbine rotor's rotation circumference are obtained, primarily including the rotor diameter and swept area. These parameters are derived from the turbine's design drawings or technical specification database. To refine the simulation of the blade's positional changes during rotation, the 360-degree circumference of the wind turbine rotor is divided into multiple discrete azimuth angle steps. The step size depends on the accuracy requirements of the simulation calculation and the response frequency of the control system. For example, dividing the circumference into 72 steps means that each step corresponds to an angle increment of 5 degrees. The angular position corresponding to each azimuth angle step is defined, starting from 0 degrees and increasing by 5 degrees sequentially until 355 degrees. This generates an azimuth angle step sequence containing 72 angular values. The elements in the sequence are 0 degrees, 5 degrees, 10 degrees, and up to 355 degrees. This sequence provides a discretized temporal and spatial reference for subsequently determining the instantaneous spatial position of the blade during rotation.

[0024] The spatial coordinate positioning submodule, combined with the azimuth step sequence and the obtained blade length and hub height parameters of the wind turbine, calculates the three-dimensional spatial position of the blade tip at each azimuth step and determines the blade spatial coordinates of the wind turbine blade relative to the ground in the rotation circumference. By combining the azimuth step sequence with the blade length and hub height parameters of the wind turbine obtained from the turbine configuration database, the trajectory of the blade tip in space is located. The calculation formula is as follows: In the formula, Indicates the first The vertical height component of the blade tip at each azimuth step, in meters; This indicates the vertical height of the hub center from the ground, obtained from the unit design parameter table, in meters; This indicates the blade length, obtained from the unit design parameter table, in meters. The blade pre-cone angle is set based on the design value of the installation angle between the blade root and the hub connecting flange, and the unit is degrees. This parameter is used to calculate the effective projected length of the blade length on the plane of rotation. The presence of the pre-cone angle causes the blade tip to tilt slightly towards the leeward side to increase the tower clearance. Indicates the first The angle values ​​for each azimuth step are derived from the azimuth step sequence and are in degrees. This is a cosine trigonometric function used to convert angular relationships into length proportionality coefficients; this term is dimensionless. For each angle value in the azimuth step sequence, this geometric relationship is used to calculate the three-dimensional spatial position of the blade tip. Parameter settings are as follows: hub height. Set to 90 meters; blade length Set to 60 meters; current azimuth step size Select 0 degrees (i.e., the blade is vertically upward); blade pre-cone angle Set the angle to 0 degrees (this is a simplified numerical demonstration; in actual working conditions, it is usually 2 to 4 degrees). Substitute the above parameters into the formula to calculate, first calculating the cosine of the pre-cone angle. Then calculate the cosine value of the azimuth angle. The final numerical calculation process is as follows: The calculated result of 150 meters represents the vertical height of the blade tip at an azimuth angle of 0 degrees. By traversing the entire azimuth step sequence, the vertical height corresponding to each step is calculated one by one, and combined with the horizontal displacement component, the set of spatial coordinates of the wind turbine blade relative to the ground within the rotation circumference is determined.

[0025] Please see Figure 4 The torque ripple calculation module includes: The density parameter extraction submodule locates the corresponding height layer index in the air density gradient vector based on the vertical component of the spatial coordinates of the wind turbine blades, extracts the local air density at each coordinate point according to the height layer index, and arranges the local air density in the order of azimuth step size to generate a local air density sequence. Based on the vertical height component at each azimuth step output by the wind turbine blade spatial coordinate positioning submodule, data matching is performed in the air density gradient vector. Since the gradient vector is constructed based on discrete height layers, and the calculated blade tip height may be a continuously changing value, a linear interpolation algorithm is used to locate the corresponding height layer index or calculate the interpolated density in the air density gradient vector. For example, when the blade tip is at a height of 150 meters, the air density value of 1.185 kg / m³ corresponding to the 150-meter height layer is directly extracted from the gradient vector; when the blade is at a hub height of 90 meters, the air density value of 1.200 kg / m³ at 90 meters is extracted; and when the blade is at a height of 30 meters, the corresponding value of 1.225 kg / m³ is extracted. Following the order of the azimuth step, the local air density at each coordinate point is extracted or calculated sequentially, and these values ​​are arranged in order to generate a local air density sequence.

[0026] The torque waveform calculation submodule obtains the aerodynamic parameters of the wind turbine blades and the real-time collected inflow wind speed. It combines the local air density sequence to simulate the force fluctuations caused by changes in air density during the rotation of the blades and calculates the torque waveform at each azimuth step. The aerodynamic parameters of the wind turbine blades are obtained, mainly including the wind energy utilization coefficient curve, the blade swept area, and the inflow wind speed data collected in real time at the hub height. The inflow wind speed data comes from real-time measurements by the nacelle anemometer. Combined with the generated local air density sequence, the aim is to simulate the force fluctuations caused by changes in air density during blade rotation. For each azimuth step in the sequence, the aerodynamic torque calculation logic is invoked. This logic calculates the instantaneous aerodynamic power by multiplying a constant of 0.5, the current local air density value, the blade swept area, the cube of the inflow wind speed, and the wind energy utilization coefficient corresponding to the current tip speed ratio. This power is then divided by the rotor speed to obtain the aerodynamic torque value at that azimuth step. For example, assuming a swept area of ​​10,000 square meters, an inflow wind speed of 10 meters per second, a wind energy utilization coefficient of 0.45, and a rotor rotational angular velocity of 1 radian per second, at a 0-degree azimuth angle, the air density is 1.185. The calculation process is as follows: (Unit: Watts; numerical value: equivalent to torque in Newton-meters). At a 180-degree azimuth angle, the air density is 1.225. The calculation process is as follows: Perform the above calculations for all azimuth step sizes, arrange the calculation results in angular order, and form a torque waveform that varies with the rotation angle.

[0027] The fluctuation amplitude calculation submodule extracts the peak and valley data of the torque waveform within a single rotation cycle, performs difference calculation on the peak and valley data, quantifies the torque change range, and obtains the density-induced torque fluctuation amplitude. Extreme value analysis is performed on the generated torque waveform. All data points within a single rotation cycle of the torque waveform are traversed, and the maximum value is identified and extracted as the peak value, while the minimum value is extracted as the valley value. In the previous example, the peak value corresponds to 2,756,250 Nm when the blade passes its lowest point, and the valley value corresponds to 2,666,250 Nm when the blade passes its highest point. Subsequently, a difference calculation is performed on the peak and valley values. The calculation process is as follows: This 90,000 Nm difference quantifies the range of torque variation caused solely by the ambient density gradient. Finally, this difference is defined as the density-induced torque ripple amplitude. This amplitude is used to subsequently determine the critical compensation amount for the dynamic power boundary.

[0028] Please see Figure 5 The dynamic range construction module includes: The dynamic boundary setting submodule obtains the reference error band width of the standard power curve of the wind turbine, converts the density-induced torque fluctuation amplitude into the corresponding power fluctuation amplitude, and superimposes it onto the reference error band width. It then calculates the upper and lower limits of the fluctuation of the standard power curve of the wind turbine under the current environmental density gradient and determines the dynamic boundary values. The density-induced torque fluctuation amplitude is converted into the corresponding power fluctuation amplitude and superimposed onto the reference error band width. The upper and lower limits of the standard power curve of the wind turbine under the current environmental density gradient are calculated, and the dynamic boundary values ​​are determined, including: For each discrete speed point in the standard power curve of the wind turbine, extract the corresponding standard power value; The density-induced torque fluctuation amplitude is converted into the corresponding power fluctuation amplitude and superimposed on the reference error band width to obtain the total deviation under the current environment; Add the standard power value to the total deviation to obtain the upper limit boundary value of the power at the corresponding discrete speed point; Subtract the standard power value from the total deviation to obtain the lower limit boundary value of the power at the corresponding discrete speed point; The set of power upper limit boundary values ​​and power lower limit boundary values ​​corresponding to all discrete speed points is defined as dynamic boundary values; The baseline error band width of the standard power curve is obtained from the wind turbine's control system. This baseline error band width is set based on the power curve assessment tolerance specified by the International Electrotechnical Commission (IEC) standard, for example, 5% of the standard power. The calculated density-induced torque fluctuation amplitude is converted into the corresponding power fluctuation amplitude and superimposed on this baseline error band width. Assuming the current speed is 1 radian per second, a torque fluctuation of 90,000 Nm corresponds to a power fluctuation of 90,000 watts, or 90 kilowatts. Under the standard power condition of 2,000 kilowatts, the baseline error band width corresponds to 100 kilowatts (i.e., 2,000 multiplied by 0.05). The calculation process is as follows: Next, the standard power value of 2000 kW is added to the total deviation of 190 kW. The calculation process is as follows: The upper limit boundary value of the power is obtained; the standard power value is subtracted from the total deviation. The calculation process is as follows: This yields the lower limit boundary value of the power. This calculation is repeated for all speed points covered by the power curve, and the set of all calculated upper and lower limit values ​​is defined as the dynamic boundary values.

[0029] The confidence interval generation submodule performs bidirectional numerical bias calculations on the standard power curve of the wind turbine based on dynamic boundary values, constructs dynamic power envelopes on both sides of the standard power curve of the wind turbine, defines the distribution range of wind turbine power data at multiple speeds based on the dynamic power envelopes, and generates dynamic power confidence intervals. Based on the dynamic boundary values ​​output by the dynamic boundary setting submodule, a bidirectional numerical bias calculation is performed on the standard power curve of the wind turbine in a two-dimensional coordinate system. Specifically, the upper power boundary values ​​corresponding to all discrete speed points are connected and smoothed using spline interpolation or linear fitting to construct an upper envelope above the standard power curve; similarly, the lower power boundary values ​​are connected and smoothed to construct a lower envelope below the standard power curve. The closed region enclosed by these two dynamic power envelopes is the dynamic power confidence interval. This interval defines the reasonable range within which the power data of the wind turbine should be distributed at various speeds under the current specific air density gradient environment.

[0030] The status label determination submodule collects real-time output power and speed data of the wind turbine, constructs the current operating point, and determines that the wind turbine is in normal wind energy capture state if the current operating point is within the dynamic power confidence interval, and generates a power status label. The system collects real-time output power and speed data of the wind turbines, maintaining a sampling frequency consistent with the central monitoring system. Using the collected real-time speed as an index, it searches for the corresponding upper and lower power limits within a pre-generated dynamic power confidence interval. Then, it constructs the current operating point and determines whether the power value at this point falls between the found upper and lower limits. This determination includes three possibilities: the first possibility is that the real-time power is between the upper and lower limits; the second possibility is that the real-time power is higher than the upper limit; and the third possibility is that the real-time power is lower than the lower limit. For example, if the currently collected speed is 1 radian per second, the real-time output power is 2100 kW. The confidence interval for this speed is 1810 kW to 2190 kW. Numerical comparisons are then performed. If the result falls under the first possibility, the wind turbine is confirmed to be in normal wind energy capture mode, and a power status label of "Normal" is generated. If the real-time power is 1700 kilowatts, it falls under the third possibility, and an "Abnormal" label is generated.

[0031] Please see Figure 6 The indicator generation module includes: The angle deviation calculation submodule collects the current orientation angle of the wind turbine nacelle, calculates the numerical difference between the current orientation angle of the wind turbine nacelle and the optimal wind angle, quantifies the degree of deviation between the current orientation of the nacelle and the maximum wind energy capture direction of the optimal wind angle, and obtains the angle deviation value. The calculation of the numerical difference between the current orientation angle of the wind turbine nacelle and the optimal windward angle includes: Obtain the algebraic difference between the current orientation angle of the wind turbine nacelle and the optimal windward angle, and determine whether the absolute value of the algebraic difference is greater than 180 degrees. If the absolute value of the algebraic difference is greater than 180 degrees, then the circumferential complementary angle is obtained by subtracting the absolute value of the algebraic difference from 360 degrees, and the circumferential complementary angle is determined as the angle deviation value. If the absolute value of the algebraic difference is less than or equal to 180 degrees, then the absolute value of the algebraic difference is determined as the angular deviation value. The nacelle position sensor collects the current orientation angle of the wind turbine nacelle relative to true north, for example, an angle of 350 degrees. This angle data comes from the reading of the nacelle yaw encoder. Simultaneously, the optimal windward angle output from the optimal parameter extraction submodule is called, for example, 10 degrees. The algebraic difference between the two is calculated, the calculation process is as follows: Next, determine if the absolute value of the algebraic difference is greater than 180 degrees. This determination has two possibilities: the first possibility is that the absolute value is greater than 180 degrees, and the second possibility is that the absolute value is less than or equal to 180 degrees. In this example, 340 is greater than 180, falling under the first possibility. This indicates that the two cross the 0-degree mark in the circumferential direction. Therefore, perform the calculation of the supplementary angle in the circle, subtracting the absolute value of the algebraic difference from 360 degrees. The calculation process is as follows: This 20-degree deviation is the actual physical angle deviation value.

[0032] The weighted deviation quantization submodule performs a weighted operation on the deviation sensitivity weight and the angle deviation value to obtain the weighted yaw deviation value. The system receives the angle deviation value output by the angle deviation calculation submodule and the deviation sensitivity weight output by the weighting coefficient quantization submodule. A weighted multiplication operation is performed, multiplying the angle deviation value by the deviation sensitivity weight. For example, if the angle deviation value is 20 degrees as calculated above, and the deviation sensitivity weight is 1.2 as matched above, the specific calculation process is as follows: The result is defined as a weighted yaw deviation value. By introducing weighting coefficients, it not only considers the magnitude of the physical angle error but also incorporates the morphological characteristics of the current wind energy flow distribution.

[0033] The evaluation index synthesis submodule associates and matches the weighted yaw deviation value with the power status label, and rates the yaw deviation in combination with the power output status to reflect the unit's yaw control performance and obtain multi-dimensional operation evaluation indexes. The calculated weighted yaw deviation value is matched with the power status label generated by the status label determination submodule. First, the power status label is checked. If the label shows "abnormal" or "underpowered," the multi-dimensional operation evaluation index is directly judged as "needs inspection" or "performance degradation." If the power status label shows "normal," the unit's yaw control performance is further rated based on the magnitude of the weighted yaw deviation value. Preset grading thresholds are used, determined based on the statistical quantiles of a large amount of historical weighted deviation data. For example, the 10th and 20th percentiles are set as dividing lines, corresponding to values ​​of 10 and 20, respectively. The rating includes three possibilities: the first possibility is a weighted deviation value less than 10, rated as "excellent"; the second possibility is a weighted deviation value between 10 and 20, rated as "good"; and the third possibility is a weighted deviation value greater than 20, rated as "needs optimization." In the aforementioned example, the weighted yaw deviation value is 24, and the comparison is as follows: The assessment result falls under the third possibility. Combining the "normal" power status with the final multi-dimensional operational evaluation index, the output is "Power is normal, but yaw needs optimization." This result indicates that although the unit's energy conversion efficiency is within the normal range, there is significant room for optimization in yaw control, and it is recommended that maintenance personnel make targeted adjustments.

[0034] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A comprehensive evaluation system for multi-dimensional operational indicators of wind farms, characterized in that, The system includes: The deviation from the baseline construction module calculates the wind energy flow distribution histogram based on the current wind speed and direction, determines the optimal wind angle of the wind energy flow distribution histogram, and calculates the deviation sensitivity weight. The gradient environment mapping module constructs an air density gradient vector based on the current air pressure and temperature, and simultaneously determines the blade's spatial coordinates under the azimuth step size. The torque pulsation calculation module extracts the local air density from the air density gradient vector based on the blade spatial coordinates, calculates the theoretical torque waveform by combining the currently collected inflow wind speed and aerodynamic parameters, and calculates the density-induced torque fluctuation amplitude based on the extreme value of the theoretical torque waveform. The dynamic interval construction module combines the density-induced torque fluctuation amplitude with the collected wind turbine reference error band width to construct a dynamic power confidence interval, compares it with the current wind turbine output power data and speed data, and generates a power status label. The index generation module calculates a weighted yaw deviation value based on the difference between the cabin orientation angle and the optimal windward angle and the deviation sensitivity weight, and matches the weighted yaw deviation value with the power status label to generate a multi-dimensional operation evaluation index.

2. The comprehensive evaluation system for multi-dimensional operation indicators of wind farms according to claim 1, characterized in that, The deviation sensitivity weight is specifically a quantified coefficient determined by retrieving the kurtosis coefficient of the wind energy flow distribution histogram from a preset coefficient weight mapping table. The air density gradient vector is specifically the air density value arranged in vertical order and corresponding to each height layer within the swept surface of the wind turbine rotor. The blade spatial coordinates include the discrete azimuth angle position of the wind turbine blade on the rotation circumference and the corresponding height layer index in the vertical direction. The density-induced torque fluctuation amplitude is specifically the difference calculated based on the peak and valley data of the theoretical torque waveform in a single rotation cycle. The power status label is specifically a logical status identifier generated based on the real-time operating point of the wind turbine and the dynamic power confidence interval. The multi-dimensional operation evaluation index includes a weighted yaw deviation value combined with the degree of wind direction dispersion and a power status judgment result characterizing the conformity of the power curve.

3. The comprehensive evaluation system for multi-dimensional operation indicators of wind farms according to claim 1, characterized in that, The deviation from the baseline construction module includes: The histogram construction submodule collects wind speed and wind direction data within the target sector of the wind farm, divides the wind speed and wind direction data into statistical intervals to obtain wind direction frequency, maps the wind speed magnitude to each wind direction frequency interval for weighted accumulation, reflects the distribution pattern of wind energy density in multiple directions, and obtains a wind energy flow distribution histogram. The optimal parameter extraction submodule locates the density peak position in the wind energy flow distribution histogram and determines the optimal wind angle through the mean drift algorithm. At the same time, it obtains the distribution pattern value of the wind energy flow distribution histogram and calculates the fourth standard moment of the distribution pattern value as the wind energy flow distribution kurtosis coefficient. The weight coefficient quantization submodule presets multiple numerical segment intervals of the wind energy flow distribution kurtosis coefficient, constructs a coefficient weight mapping table by matching the corresponding evaluation weight for each numerical segment interval, retrieves the current wind energy flow distribution kurtosis coefficient in the coefficient weight mapping table, matches the weight value corresponding to the numerical segment interval, and outputs the deviation sensitivity weight.

4. The comprehensive evaluation system for multi-dimensional operation indicators of wind farms according to claim 3, characterized in that, For calculating the fourth standard moment of the distribution shape The formula used is: ； in, This represents the total number of sectors divided in the wind energy flow distribution histogram. Indicates the index marker of the sector, Indicates the first Wind energy density values ​​for each sector This represents the average wind energy density across all sectors. The standard deviation of wind energy density distribution. This represents the data quality weighting coefficient. This represents the baseline value for kurtosis of a normal distribution.

5. The comprehensive evaluation system for multi-dimensional operation indicators of wind farms according to claim 3, characterized in that, The gradient environment mapping module includes: The gradient vector construction submodule collects air pressure and temperature data from the lidar at multiple height levels within the swept surface of the wind turbine rotor, calculates the air density corresponding to each height level, arranges the air density in order of vertical height and records the height level index, and obtains the air density gradient vector. The circumference step size discrete submodule obtains the geometric parameters of the wind turbine rotor rotation circumference, divides the wind turbine rotor rotation circumference into multiple discrete azimuth step sizes, defines the angle position corresponding to each azimuth step size, and generates the corresponding azimuth step size sequence. The spatial coordinate positioning submodule, in conjunction with the azimuth step sequence and the obtained blade length and hub height parameters of the wind turbine, calculates the three-dimensional spatial position of the blade tip at each azimuth step and determines the blade spatial coordinates of the wind turbine blade relative to the ground within the rotation circumference.

6. The comprehensive evaluation system for multi-dimensional operation indicators of wind farms according to claim 5, characterized in that, The torque ripple calculation module includes: The density parameter extraction submodule locates the corresponding height layer index in the air density gradient vector based on the vertical component of the spatial coordinates of the wind turbine blades, extracts the local air density at each coordinate point according to the height layer index, and arranges the local air density in the order of azimuth step size to generate a local air density sequence. The torque waveform calculation submodule obtains the aerodynamic parameters of the wind turbine blades and the real-time collected inflow wind speed. It combines the local air density sequence to simulate the force fluctuations caused by changes in air density during the rotation of the blades and calculates the torque waveform at each azimuth step. The fluctuation amplitude calculation submodule extracts the peak and valley data of the torque waveform within a single rotation cycle, performs difference calculation on the peak and valley data, quantifies the torque change range, and obtains the density-induced torque fluctuation amplitude.

7. The wind farm multi-dimensional operation index comprehensive evaluation system according to claim 6, characterized in that, The dynamic interval construction module includes: The dynamic boundary setting submodule obtains the reference error band width of the standard power curve of the wind turbine, converts the density-induced torque fluctuation amplitude into the corresponding power fluctuation amplitude, and superimposes it onto the reference error band width. It then calculates the upper and lower limits of the fluctuation of the standard power curve of the wind turbine under the current environmental density gradient and determines the dynamic boundary values. The confidence interval generation submodule performs bidirectional numerical bias calculations on the standard power curve of the wind turbine based on the dynamic boundary values, constructs dynamic power envelopes on both sides of the standard power curve of the wind turbine, defines the distribution range of the wind turbine power data at multiple speeds based on the dynamic power envelopes, and generates dynamic power confidence intervals. The status label determination submodule collects real-time output power and speed data of the wind turbine, constructs the current operating point, and determines that the wind turbine is in a normal wind energy capture state if the current operating point is within the dynamic power confidence interval, and generates a power status label.

8. The comprehensive evaluation system for multi-dimensional operation indicators of wind farms according to claim 7, characterized in that, The density-induced torque fluctuation amplitude is converted into the corresponding power fluctuation amplitude and superimposed on the reference error band width. The upper and lower limits of the standard power curve of the wind turbine under the current environmental density gradient are calculated, and the dynamic boundary values ​​are determined, including: For each discrete speed point in the standard power curve of the wind turbine, extract the corresponding standard power value; The density-induced torque fluctuation amplitude is converted into the corresponding power fluctuation amplitude and superimposed on the reference error band width to obtain the total deviation under the current environment; Add the standard power value to the total deviation to obtain the upper limit boundary value of the power at the corresponding discrete speed point; Subtract the standard power value from the total deviation to obtain the lower limit boundary value of the power at the corresponding discrete speed point; The set of upper and lower power boundary values ​​corresponding to all discrete speed points is defined as the dynamic boundary value.

9. The comprehensive evaluation system for multi-dimensional operation indicators of wind farms according to claim 7, characterized in that, The indicator generation module includes: The angle deviation calculation submodule collects the current orientation angle of the wind turbine nacelle, calculates the numerical difference between the current orientation angle of the wind turbine nacelle and the optimal wind angle, quantifies the degree of deviation between the current orientation of the nacelle and the maximum wind energy capture direction of the optimal wind angle, and obtains the angle deviation value. The weighted deviation quantization submodule performs a weighted operation on the deviation sensitivity weight and the angle deviation value to obtain the weighted yaw deviation value. The evaluation index synthesis submodule associates and matches the weighted yaw deviation value with the power status label, and rates the yaw deviation in conjunction with the power output status to reflect the unit's yaw control performance and obtain multi-dimensional operation evaluation indexes.

10. The comprehensive evaluation system for multi-dimensional operation indicators of wind farms according to claim 9, characterized in that, The calculation of the numerical difference between the current orientation angle of the wind turbine nacelle and the optimal windward angle includes: Obtain the algebraic difference between the current orientation angle of the wind turbine nacelle and the optimal windward angle, and determine whether the absolute value of the algebraic difference is greater than 180 degrees. If the absolute value of the algebraic difference is greater than 180 degrees, then the circumferential complementary angle is obtained by subtracting the absolute value of the algebraic difference from 360 degrees, and the circumferential complementary angle is determined as the angle deviation value. If the absolute value of the algebraic difference is less than or equal to 180 degrees, then the absolute value of the algebraic difference is determined as the angular deviation value.

Images