Desert wind field power curve dynamic correction method considering sand and dust abrasion

Abstract

The invention relates to the technical field of curve correction, in particular to a desert wind field power curve dynamic correction method considering sand and dust abrasion, and the method comprises the steps: collecting operation state data and sand and dust concentration data, and carrying out the fusion to generate a sand and dust abrasion real-time evaluation index; based on a deviation value between the index and a standard power value, calculating an attenuation compensation amount and carrying out real-time compensation on the initial output power to generate a corrected output power; meanwhile, wind speed response characteristic parameters of the initial power curve are corrected in a segmented mode according to the indexes, and a primary corrected desert wind field power curve is generated; and finally, combining the corrected output power and the current wind speed into a verification data point, and generating a secondary corrected desert wind field power curve capable of continuously tracking performance degradation by fitting and optimizing the primary corrected curve, thereby realizing the self-adaption and dynamic calibration of the power curve in an unstable wear environment. And the problem of dynamic mismatch between a theoretical model and physical properties is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of curve correction technology, specifically to a method for dynamic correction of desert wind field power curves that takes into account sand and dust abrasion. Background Technology

[0002] In wind farm operations in desert regions, wind turbines are constantly exposed to high concentrations of dust, leading to continuous wear on blades and critical components. This wear alters the aerodynamic characteristics of the blades, such as increasing surface roughness and drag, thereby reducing the turbine's conversion efficiency and causing the actual output power to deviate from design specifications. The power curve, as the core model of wind turbine performance, defines the mapping relationship between wind speed and output power, and is crucial for wind farm power prediction, load control, and economic benefit assessment. However, under the influence of dust wear, the power curve drifts, especially in desert environments where dust concentration changes dynamically, and the wear effect exhibits cumulative and nonlinear characteristics, posing challenges to traditional power curve correction methods. Existing technologies mostly employ offline correction or periodic maintenance strategies based on historical data, failing to capture the instantaneous changes in dust wear in real time, resulting in correction lag and insufficient accuracy. Some improved methods attempt to integrate environmental sensor data, but are often limited to single parameter adjustments, such as directly scaling power output based solely on dust concentration, ignoring the differentiated impact of wear on different wind speed ranges of the power curve (e.g., low-wind-speed start-up zone, medium-wind-speed operating zone, and rated wind speed zone). For example, dust abrasion can significantly affect the blade starting performance in low wind speed areas, while mainly increasing mechanical losses in high wind speed areas. However, traditional methods lack detailed corrections for wind speed response characteristics.

[0003] Furthermore, existing methods often rely on static compensation models, which are ill-suited to the frequent fluctuations in dust concentration and the gradual effects of wear, leading to accumulated power prediction errors and impacting the overall power generation efficiency and reliability of wind farms. Therefore, there is an urgent need for a comprehensive method capable of real-time assessment of dust wear and dynamic adjustment of power curves to improve the operational adaptability and accuracy of desert wind farms. Summary of the Invention

[0004] The purpose of this invention is to provide a dynamic correction method for desert wind field power curves that takes into account sand and dust abrasion, in order to solve the problems mentioned in the background art. Specific technical problems include how to achieve autonomous adaptation and dynamic calibration of desert wind field power curves under unsteady abrasion environments, in order to solve the common industry problem of dynamic mismatch between physical performance degradation and theoretical models.

[0005] To achieve the above objectives, the present invention aims to provide a method for dynamically correcting desert wind field power curves that takes into account sand and dust abrasion, specifically including the following method steps: S1. Real-time data collection of operating status, including output power, speed, blade pitch angle, gearbox temperature, vibration amplitude, and generator efficiency parameters, is achieved through a sensor system integrated on the wind turbine. Simultaneously, dust concentration monitoring equipment deployed in the wind farm environment is used to continuously measure the dust concentration index of the air around the turbine, which serves as dust concentration data.

[0006] Step S1 provides real-time, synchronous raw data input for the entire dynamic correction system, solving the problem of data completeness necessary for autonomous adaptation. Specifically, it not only collects core parameters such as output power and speed that directly reflect the unit's performance, but also captures deep state information such as vibration and temperature that are closely related to wear, and combines external dust concentration as a key environmental stress indicator.

[0007] S2. Based on operational status data and dust concentration data, a real-time assessment index for dust wear is generated using a fusion processing method; the fusion processing method specifically includes: The collected operational status data and dust concentration data are preprocessed, with mean filtering used to smooth data fluctuations. Key feature parameters are extracted from the preprocessed operational status data and dust concentration data. A feature weighting allocation model based on entropy weighting is used to assign weight coefficients to each key feature parameter. The weighted feature parameters are then linearly combined to calculate real-time dust wear assessment indicators. The key feature parameters include power output fluctuation variance, high-frequency energy value of vibration signal, and moving average value of dust concentration. The specific operation process of the feature weighting allocation model based on entropy weighting includes: Collect historical sample data of all key feature parameters within a fixed time period to form the original data matrix; The data in each column of the original data matrix is ​​standardized to convert it into a dimensionless value between zero and one. Calculate the proportion of each sample value in the total sample values ​​of each key feature parameter; Based on the definition of information entropy, the information entropy of each key feature parameter is calculated using the logarithm of the weight. Calculate the difference coefficient for each key feature parameter. The difference coefficient is equal to one minus the information entropy of that key feature parameter. Divide the difference coefficient of each key feature parameter by the sum of the difference coefficients of all key feature parameters to obtain the weight coefficient of that key feature parameter.

[0008] The core contribution of step S2 lies in transforming the multidimensional raw data collected in S1 into a single real-time assessment index that comprehensively and objectively quantifies the current degree of sand and dust wear through fusion processing based on the entropy weight method. It solves the technical problem of how to condense the complex, multidimensional physical state degradation into a key signal that can drive model correction. By extracting features strongly correlated with wear, such as power fluctuations and high-frequency vibration energy, and objectively assigning weights based on the data's inherent volatility (information content) using the entropy weight method, this index avoids subjective assumptions and can more sensitively capture early signs and dynamic changes in wear. This index, as the perception center of the entire system, provides a unified and reliable quantitative basis for subsequent model correction and is the cornerstone of dynamic calibration decisions.

[0009] S3. By analyzing the deviation between the real-time assessment index of sand and dust wear and the standard power value, the attenuation compensation amount of the initial desert wind field power curve is calculated, and the attenuation compensation amount is applied to the output power value obtained from the initial desert wind field power curve to generate the corrected output power; the specific process of generating the corrected output power includes: The standard power value corresponding to the current wind speed is retrieved from the wind turbine control system. The standard power value is pre-stored in a wind speed and standard power lookup table constructed from the initial desert wind field power curve of the unit. Calculate the deviation between the real-time assessment index of sand and dust wear and the preset standard wear index threshold; Based on the specific value of the deviation, consult the preset deviation and compensation comparison table to obtain the attenuation compensation amount; The attenuation compensation is algebraically added to the initial output power value obtained from the initial desert wind field power curve based on the current wind speed to generate the corrected output power.

[0010] Step S3 utilizes the real-time dust wear assessment index generated in S2 to rapidly compensate and correct the theoretical output power at the current wind speed, generating a corrected output power that is closer to the true value. Its technical advantage lies in achieving instantaneous response and feedforward compensation to power curve deviations, solving the challenge of real-time correction in dynamic mismatch problems. It does not alter the power curve model itself, but instead quickly calculates a compensation amount based on real-time wear deviations using a lookup table method, directly adding it to the output of the initial model. This method is computationally efficient and responds rapidly, immediately compensating for instantaneous power drops caused by sudden increases or accumulations of dust, ensuring the accuracy of short-term power prediction and control of the wind field under harsh conditions, and providing time for more complex model parameter corrections and a verification benchmark.

[0011] S4. By analyzing the influence of real-time sand and dust abrasion assessment indicators on the changing trend of the initial desert wind field power curve in different wind speed ranges, the wind speed response characteristic parameters of the initial desert wind field power curve are corrected to generate a first-corrected desert wind field power curve; the specific process of generating the first-corrected desert wind field power curve includes: The entire operating wind speed range of the wind turbine is divided into three continuous intervals: low wind speed zone, medium wind speed zone, and rated wind speed zone. For each wind speed range, based on the current value of the real-time dust wear assessment index, the preset wear index and parameter correction table is consulted to obtain the power curve slope correction and power curve intercept correction. Add the queried slope correction amount to the original slope parameter of the initial desert wind field power curve in the current wind speed range to obtain the corrected slope; add the queried intercept correction amount to the original intercept parameter to obtain the corrected intercept. Using the corrected slope and intercept parameters for each wind speed range, a piecewise linear power curve is reconstructed as the first-corrected desert wind field power curve.

[0012] The corrected output power is combined with the current wind speed to form validation data points. Based on these validation data points, the primary corrected desert wind field power curve is fitted and optimized to generate the secondary corrected desert wind field power curve. The specific process of generating the secondary corrected desert wind field power curve includes: The real-time generated corrected output power is combined with the synchronously measured current wind speed to form verification data points, which are then stored in a verification dataset with a fixed capacity and following a first-in-first-out (FIFO) principle. The FIFO principle execution process specifically includes: To verify the dataset, a fixed-length sequential storage queue is maintained, which is arranged in the order of the timestamps generated by the data points; When a new verification data point is generated and needs to be stored, first check if the current queue is full; if the queue is not full, insert the new data point directly into the tail of the queue; if the queue is full, first remove the oldest data point stored at the head of the queue, and then insert the new verification data point into the tail of the queue. The weighted least squares fitting algorithm is invoked, and weights are assigned to each data point in the validation dataset based on an exponentially decaying weight model. The execution process of the exponentially decaying weight model specifically includes: Based on the current system time, the weight of each data point is uniquely determined by the time difference between its timestamp and the current time. The weight of a data point is equal to the power of the decay base, where the decay base is a fixed constant less than one and greater than zero. Using a modified desert wind field power curve as the initial model and a weighted validation dataset as the fitting basis, recursive calculations are performed. The optimization objective is to minimize the sum of squared weighted power residuals for all data points, and output a set of optimized power curve parameters. The optimized power curve parameters are used to replace the corresponding parameters in the primary correction curve to form the secondary correction desert wind field power curve.

[0013] Step S4 systematically addresses the "continuous self-adaptation" problem of the power curve model through two stages: primary correction and secondary correction. Primary correction focuses on structural adjustments to the model, adjusting the slope and intercept parameters of the power curve differently for different wind speed ranges (e.g., low, medium, and high) based on wear assessment indicators. This solves the nonlinear problem of wear's impact on aerodynamic characteristics varying with operating conditions, enabling the model to finely adapt to overall performance drift caused by wear. Secondary correction further introduces a data-driven optimization mechanism. It uses the corrected output power generated in S3 as verification data to approximate the true value, recursively fitting the curve after primary correction using weighted least squares with exponential decay weights. This not only continuously optimizes model parameters using the latest operating data but also ensures, through a first-in-first-out queue and exponential decay weights, that the model can both track the latest performance changes of the equipment and mitigate the influence of outdated historical data. This achieves autonomous evolution and dynamic calibration of the power curve under unsteady wear conditions, thus solving the long-standing mismatch between theoretical models and physical performance degradation.

[0014] Compared with the prior art, the beneficial effects of the present invention are: This invention integrates multi-source data into a real-time assessment index for sand and dust wear. This index not only enables rapid real-time compensation of output power to cope with instantaneous fluctuations, but also allows for refined segmented parameter correction of the power curve in different wind speed ranges. Finally, by introducing a weighted least squares fitting algorithm based on an exponential decay weight model, and using real-time verification data to recursively optimize the primary correction curve, this invention ensures that the power curve model can continuously track the gradual changes in equipment performance. This solves the problem of theoretical model mismatch caused by physical performance degradation, and significantly improves wind farm power generation efficiency, power prediction accuracy, and operational reliability. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the overall method steps of the present invention; Figure 2 This is the core flowchart of step S4 of the present invention. Detailed Implementation

[0016] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Next, please refer to Figure 1 The purpose of this embodiment is to provide a dynamic correction method for desert wind field power curves considering sand and dust abrasion, which includes the following steps: S1. Real-time operational status data is collected through a sensor system integrated on the wind turbine (such as power sensors, tachometers, temperature sensors, vibration accelerometers, etc.), including but not limited to output power, rotational speed, blade pitch angle, gearbox temperature, vibration amplitude, and generator efficiency parameters. Simultaneously, dust concentration monitoring equipment deployed in the wind farm environment (such as laser scattering particulate matter sensors) is used to continuously measure the dust concentration index of the air around the turbine, which serves as dust concentration data. Operational status data and dust concentration data are collected synchronously at a fixed sampling frequency (such as once per second), and are preprocessed, timestamped, and stored by the data acquisition unit to ensure data integrity and real-time performance, providing reliable input for subsequent fusion processing.

[0018] S2. Based on operational status data and dust concentration data, a fusion processing method is used to generate real-time assessment indicators for dust wear, specifically including: First, the collected operational status data and dust concentration data are preprocessed. Mean filtering is used to smooth data fluctuations, and data standardization is employed to eliminate the influence of dimensions. Mean filtering maintains a fixed-length data buffer for each data sequence (e.g., power, vibration), forming a filtering window. When a new data point arrives, it is added to the end of the buffer, while the oldest data point is removed. Then, the arithmetic mean of all data points in the current buffer is immediately calculated. This calculated mean is the smoothed output value at that moment. This process is repeated cyclically with each new data point, thus achieving point-by-point smoothing filtering of the entire data sequence.

[0019] Next, key feature parameters are extracted from the preprocessed operating status data and dust concentration data, including but not limited to power output fluctuation variance, high-frequency energy value of vibration signal, and moving average value of dust concentration.

[0020] Then, a feature weight allocation model based on entropy weighting is adopted to assign a fixed weight coefficient to each key feature parameter.

[0021] Finally, the weighted feature parameters are linearly combined to calculate a comprehensive, dimensionless real-time assessment index for sand and dust wear. This real-time assessment index for sand and dust wear is a continuous real value, and its increase directly indicates that the wear of the wind turbine is aggravated by sand and dust.

[0022] The specific operation process of the feature weight allocation model based on the entropy weight method is as follows: Historical sample data of all key characteristic parameters (power output fluctuation variance, high-frequency energy value of vibration signal, and moving average value of dust concentration) within a fixed time period are collected to form an original data matrix; then, data standardization processing is performed on each column of data in the original data matrix, that is, all sample values ​​of a single characteristic parameter, to convert them all into dimensionless values ​​between zero and one. Calculate the proportion of each sample value in the total sample values ​​of each key feature parameter; then, according to the definition of information entropy, use the logarithm of the proportion to calculate the information entropy of each key feature parameter; the greater the data dispersion of a key feature parameter, the smaller its information entropy. Calculate the difference coefficient for each key feature parameter. The difference coefficient is equal to one minus the information entropy of that feature parameter. The larger the difference coefficient, the greater the amount of information that the parameter provides in the evaluation, and the higher its importance. Divide the difference coefficient of each difference coefficient feature parameter by the sum of the difference coefficients of all feature parameters, and the quotient is the fixed weight coefficient of that difference coefficient feature parameter; the sum of the weight coefficients of all key feature parameters is always equal to one.

[0023] S3. By analyzing the deviation between the real-time assessment index of sand and dust wear and the standard power value, the attenuation compensation amount of the initial desert wind field power curve is calculated, and the attenuation compensation amount is applied to the output power value obtained from the initial desert wind field power curve to generate the corrected output power, specifically including: First, the standard power value corresponding to the current wind speed is retrieved from the wind turbine control system. This value is pre-stored in a wind speed and standard power lookup table constructed from the initial desert wind field power curve of the unit. Next, the deviation between the real-time dust wear assessment index generated in step S2 and the preset standard wear index threshold is calculated. Then, based on the specific value of the deviation, look up the preset deviation and compensation table. This table clearly specifies the unique and definite attenuation compensation for different deviation values. Finally, the attenuation compensation obtained from the query is algebraically added to the initial output power value obtained from the initial desert wind field power curve based on the current wind speed. The result is the corrected output power.

[0024] S4, please refer to Figure 2By analyzing the influence of real-time dust abrasion assessment indicators on the variation trend of the initial desert wind field power curve in different wind speed ranges, the wind speed response characteristic parameters of the initial desert wind field power curve are corrected, and a first-corrected desert wind field power curve is generated, specifically including: First, the entire operating wind speed range of the wind turbine is clearly divided into three continuous intervals: low wind speed zone, medium wind speed zone, and rated wind speed zone. For each wind speed interval, based on the current value of the real-time dust wear assessment index generated in step S2, a preset wear index and parameter correction table is directly queried. This table presets a unique power curve slope correction amount and power curve intercept correction amount for each wind speed interval and each wear index level. Then, the original slope parameter of the initial desert wind field power curve in the current wind speed range is added to the slope correction amount obtained by query to obtain the corrected slope; the original intercept parameter is added to the intercept correction amount obtained by query to obtain the corrected intercept. Finally, using the corrected slope and intercept parameters for each wind speed range, a piecewise linear power curve is reconstructed, which is the first-corrected desert wind field power curve.

[0025] After generating the primary corrected desert wind field power curve, a fitting optimization process is performed to generate the secondary corrected desert wind field power curve. The corrected output power is combined with the current wind speed as validation data points, and the primary corrected desert wind field power curve is fitted and optimized based on the validation data points to generate the secondary corrected desert wind field power curve. Specifically, this includes: First, the corrected output power generated in real time during step S3 is combined with the synchronously measured current wind speed to form a verification data point, which is then stored in a verification dataset with a fixed capacity and following a first-in, first-out (FIFO) principle. The FIFO principle is implemented through the following defined queue data structure and management logic, specifically including: To validate the dataset, a fixed-length sequential storage queue is maintained, strictly arranged according to the timestamp order of the data points. When a new validation data point is generated and needs to be stored, the system first checks whether the current queue is full (i.e., the number of data points has reached the preset capacity limit). If the queue is not full, the new data point is directly inserted into the tail of the queue. If the queue is full, the system performs a dequeue operation, permanently removing the oldest data point stored at the head of the queue, and then performs an enqueue operation, inserting the new validation data point into the tail of the queue. This mechanism ensures that the dataset always contains the latest specific number of data points, and that the storage order of its internal data points is strictly consistent with the time order.

[0026] Subsequently, the weighted least squares fitting algorithm is invoked, and weights are assigned to each data point in the validation dataset based on the exponentially decaying weight model. The specific execution process of the exponentially decaying weight model is as follows: To verify the dataset, a fixed time window length is preset, and a queue of data points strictly sorted by timestamp is continuously maintained. Weight allocation is based on a defined exponential decay weight model. This model stipulates that, with the current system time as the benchmark, the weight value of each data point is uniquely determined by the time difference between its timestamp and the current time. The specific calculation formula is as follows: The weight of a data point is equal to the decay base raised to the power of (time difference divided by the preset decay time constant); where the decay base is a fixed constant less than 1 and greater than zero, and the preset decay time constant is a fixed unit of time. According to this exponential decay weight model, the more recent the timestamp of a data point (i.e., the smaller the time difference with the current time), the larger the calculated weight value; the larger the time difference, the weight value decreases exponentially. Each data point is calculated with a specific weight value according to this rule, which is used for subsequent weighted least squares fitting calculations.

[0027] Next, the weighted least squares fitting algorithm uses the first-corrected desert wind field power curve as the initial model and the weighted validation dataset as the fitting basis to perform recursive calculations. The optimization objective is to minimize the sum of squared weighted power residuals of all data points. After the algorithm is executed, it outputs a set of optimized power curve parameters.

[0028] Finally, the optimized power curve parameters are used to directly replace the corresponding parameters in the primary correction curve, thus forming the secondary correction desert wind farm power curve for actual power control of the wind farm.

[0029] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for dynamic correction of desert wind field power curves considering sand and dust abrasion, characterized in that, The methods and steps include the following: S1. Collect operating status data of the wind turbine and dust concentration data of the surrounding environment; S2. Based on the operating status data and the dust concentration data, a real-time dust wear assessment index is generated using a fusion processing method; S3. By analyzing the deviation between the real-time evaluation index of sand and dust wear and the standard power value, calculate the attenuation compensation amount of the initial desert wind field power curve, and apply the attenuation compensation amount to the output power value obtained from the initial desert wind field power curve to generate the corrected output power. S4. By analyzing the influence of the real-time sand and dust wear assessment index on the changing trend of the initial desert wind field power curve in different wind speed ranges, the wind speed response characteristic parameters of the initial desert wind field power curve are corrected to generate a first-corrected desert wind field power curve. The corrected output power is combined with the current wind speed to form a verification data point, and the primary corrected desert wind field power curve is fitted and optimized based on the verification data point to generate a secondary corrected desert wind field power curve.

2. The method for dynamic correction of desert wind field power curves considering sand and dust abrasion according to claim 1, characterized in that, The fusion processing method specifically includes the following steps: The collected operational status data and dust concentration data were preprocessed; Key feature parameters were extracted from the preprocessed operational status data and dust concentration data; A feature weight allocation model based on entropy weighting is used to assign weight coefficients to each key feature parameter. By linearly combining the weighted characteristic parameters, a real-time assessment index for sand and dust wear is calculated.

3. The method for dynamic correction of desert wind field power curve considering sand and dust abrasion according to claim 2, characterized in that, The key characteristic parameters include power output fluctuation variance, high-frequency energy value of vibration signal, and moving average value of dust concentration.

4. The method for dynamic correction of desert wind field power curve considering sand and dust abrasion according to claim 2, characterized in that, The operation process of the feature weight allocation model based on the entropy weight method specifically includes: Collect historical sample data of all key feature parameters within a fixed time period to form the original data matrix; The data in each column of the original data matrix is ​​standardized to convert it into a dimensionless value between zero and one. Calculate the proportion of each sample value in the total sample values ​​of each key feature parameter; Based on the definition of information entropy, the information entropy of each key feature parameter is calculated using the logarithm of the weight. Calculate the difference coefficient for each key feature parameter, where the difference coefficient is equal to one minus the information entropy of that key feature parameter; Divide the difference coefficient of each key feature parameter by the sum of the difference coefficients of all key feature parameters to obtain the weight coefficient of that key feature parameter.

5. The method for dynamic correction of desert wind field power curve considering sand and dust abrasion according to claim 1, characterized in that, The process of generating the corrected output power specifically includes: The standard power value corresponding to the current wind speed is retrieved from the wind turbine control system. The standard power value is pre-stored in a wind speed and standard power lookup table constructed from the initial desert wind field power curve of the unit. Calculate the deviation between the real-time sand and dust wear assessment index and the preset standard wear index threshold; Based on the specific value of the deviation, the attenuation compensation amount is obtained by consulting the preset deviation and compensation amount comparison table. The attenuation compensation amount is algebraically added to the initial output power value obtained from the initial desert wind field power curve based on the current wind speed to generate the corrected output power.

6. The method for dynamic correction of desert wind field power curves considering sand and dust abrasion according to claim 1, characterized in that, The process of generating the modified desert wind field power curve specifically includes: The entire operating wind speed range of the wind turbine is divided into three continuous intervals: low wind speed zone, medium wind speed zone, and rated wind speed zone. For each wind speed range, based on the current value of the real-time dust wear assessment index, the preset wear index and parameter correction table is queried to obtain the power curve slope correction and power curve intercept correction. Add the queried slope correction amount to the original slope parameter of the initial desert wind field power curve in the current wind speed range to obtain the corrected slope; add the queried intercept correction amount to the original intercept parameter to obtain the corrected intercept. Using the corrected slope and intercept parameters for each wind speed range, a piecewise linear power curve is reconstructed as the first-corrected desert wind field power curve.

7. The method for dynamic correction of desert wind field power curve considering sand and dust abrasion according to claim 1, characterized in that, The process of generating the secondary corrected desert wind field power curve specifically includes: The real-time generated corrected output power is combined with the synchronously measured current wind speed to form a verification data point, and the verification data point is stored in a verification dataset with a fixed capacity and following the first-in-first-out principle. The weighted least squares fitting algorithm is invoked, and weights are assigned to each data point in the validation dataset based on the exponentially decaying weight model. Using a modified desert wind field power curve as the initial model and a weighted validation dataset as the fitting basis, recursive calculations are performed. The optimization objective is to minimize the sum of squared weighted power residuals for all data points, and output a set of optimized power curve parameters. The optimized power curve parameters are used to replace the corresponding parameters in the primary correction curve to form the secondary correction desert wind field power curve.

8. The method for dynamic correction of desert wind field power curves considering sand and dust abrasion according to claim 7, characterized in that, The execution process of the exponential decay weight model specifically includes: Based on the current system time, the weight value of each data point is uniquely determined by the time difference between its timestamp and the current time. The weight value of the data point is equal to the power of the decay base, where the decay base is a fixed constant less than one and greater than zero.

9. The method for dynamic correction of desert wind field power curve considering sand and dust abrasion according to claim 7, characterized in that, The execution process of the first-in, first-out (FIFO) principle specifically includes: To verify the dataset, a fixed-length sequential storage queue is maintained, which is arranged in the order of the timestamps generated by the data points; When a new verification data point is generated and needs to be stored, first check if the current queue is full. If the queue is not full, insert the new data point directly into the tail of the queue. If the queue is full, remove the oldest data point stored at the head of the queue and then insert the new verification data point into the tail of the queue.

10. The method for dynamic correction of desert wind field power curve considering sand and dust abrasion according to claim 2, characterized in that, The preprocessing process uses mean filtering to smooth data fluctuations.

Images