Wind turbine generator power generation efficiency optimization method and system based on wind direction and wind speed monitoring

Abstract

The invention relates to the field of efficiency optimization, and discloses a wind turbine generator power generation efficiency optimization method and system based on wind direction and wind speed monitoring, which are used for realizing long-term stable and efficient operation of a wind turbine generator. Comprising the following steps: acquiring an original wind speed and wind direction sequence by using a sensor, generating an enhanced standardized wind speed sequence containing turbulence intensity characteristics and a standardized wind direction sequence containing variable quantity characteristics through calculation, outputting a standardized wind speed and wind direction prediction sequence, and combining turbulence intensity correction to improve prediction precision and practicability. A continuous differentiable energy capture efficiency curved surface is constructed, a yaw angle instruction sequence and a pitch angle instruction sequence are obtained through analysis in the gradient direction of the curved surface, global optimization control is achieved, and instructions can be dynamically adjusted to adapt to wind regime changes. According to the method, the fan wind power efficiency evaluation report can be generated, and the power generation efficiency is optimized according to the fan wind power efficiency evaluation report.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of efficiency optimization, and particularly relates to a wind turbine generator efficiency optimization method and system based on wind direction and speed monitoring. BACKGROUND

[0002] With the continuous growth of global energy demand and the increasing depletion of traditional fossil energy, wind energy as a clean and renewable energy has attracted widespread attention. Wind turbines, as the key equipment for converting wind energy into electrical energy, their power generation efficiency directly affects the economic and environmental benefits of wind energy utilization. Therefore, how to improve the power generation efficiency of wind turbines has become an important topic in the field of wind energy research.

[0003] Wind energy capture efficiency is significantly affected by dynamic wind speed and direction. Traditional strategies are difficult to accurately predict changes in wind conditions, resulting in a significant gap between actual power output and theoretical maximum of the unit. Especially in sites with high turbulence intensity and frequent changes in wind direction, how to obtain a wind power efficiency evaluation report to assist in optimizing power generation efficiency is an effective way to improve the power generation efficiency of wind turbines.

[0004] Although the existing technology has made certain progress in the optimization of wind turbine power generation efficiency, there are still the following shortcomings: The existing technology often processes the original wind direction and speed data simply, usually only basic normalization is performed, while ignoring the turbulence characteristics of wind speed and the continuity change characteristics of wind direction; In terms of wind speed and direction prediction, the existing technology mostly uses traditional statistical methods or simple machine learning models. These methods have low prediction accuracy when dealing with nonlinear and non-stationary wind speed and direction data, and are difficult to meet the actual engineering needs.

[0005] Therefore, we propose a wind turbine generator efficiency optimization method and system based on wind direction and speed monitoring to solve the above problems. SUMMARY

[0006] The present application provides a wind turbine generator efficiency optimization method and system based on wind direction and speed monitoring, which is used to realize long-term stable and efficient operation of wind turbines.

[0007] The first aspect of the present application provides a wind turbine power generation efficiency optimization method based on wind direction and wind speed monitoring, which comprises: obtaining an original wind speed sequence and an original wind direction sequence, calling a rated wind speed parameter to process the original wind speed sequence to generate a standardized wind speed sequence, and calling a circular angle constant to process the original wind direction sequence to generate a standardized wind direction sequence; taking the standardized wind speed sequence and the standardized wind direction sequence as input sources, constructing a three-dimensional phase space vector, generating a dynamic triangular mesh topology based on real-time updated historical data flow; locating the current phase space vector in the dynamic triangular mesh topology to extract the historical standardized wind speed and wind direction data of all vertices of the simplex, performing closed calculation on the vertex data, and outputting a standardized wind speed prediction sequence and a standardized wind direction prediction sequence; constructing a continuously differentiable energy capture efficiency surface according to the standardized wind speed prediction sequence and the standardized wind direction prediction sequence as input sources, and analyzing the yaw angle instruction sequence and the pitch angle instruction sequence along the gradient direction of the efficiency surface; generating a wind turbine power generation efficiency evaluation report based on the yaw angle instruction sequence and the pitch angle instruction sequence, and optimizing according to the wind turbine power generation efficiency evaluation report.

[0008] Optionally, in the first implementation manner of the first aspect of the present application, the method comprises: collecting an original wind speed sequence in real time through a wind speed sensor and collecting an original wind direction sequence in real time through a wind direction sensor; segmenting and intercepting the original wind speed sequence based on a preset time window, calculating the statistical variance value of each segment of wind speed data, dividing the statistical variance value by the average wind speed value of the segment to generate a dynamic turbulence intensity factor sequence; calling a preset rated wind speed parameter to perform division calculation on the original wind speed sequence element by element to generate a basic standardized wind speed sequence, and fusing the dynamic turbulence intensity factor sequence as an additional dimension with the basic standardized wind speed sequence to obtain an enhanced standardized wind speed sequence; identifying the angle jump event of adjacent data points in the original wind direction sequence, injecting an angle compensation amount at the jump point to minimize the adjacent angle difference to generate a continuous wind direction sequence; calling a preset circular angle constant to perform division calculation on the continuous wind direction sequence element by element to generate a basic standardized wind direction value, calculating the difference value of adjacent timestamp standardized wind direction values, and binding the wind direction change amount as an additional dimension with the basic standardized wind direction value to obtain a standardized wind direction sequence.

[0009] Optionally, in the second implementation manner of the first aspect of the present application, the method comprises: extracting a basic standardized wind speed value in the enhanced standardized wind speed sequence, extracting a wind direction change amount in the standardized wind direction sequence containing the change amount feature, multiplying the standardized wind speed value of the historical time window by the sine value of the wind direction change amount at the corresponding moment to generate a composite feature sequence; constructing a three-dimensional phase space vector to obtain a three-dimensional phase space point set containing a time marker according to the time stamp; setting a dynamic sliding window to retain the latest N three-dimensional coordinate points, performing Delaunay subdivision calculation on the coordinate point set in the window to generate a dynamically updated tetrahedral mesh topology; inputting the three-dimensional coordinate point at the current moment into the tetrahedral mesh topology, traversing the tetrahedral unit to locate a target simplex containing the point, recording the vertex index set of the target simplex to obtain a simplex vertex index set to which the current point belongs; extracting the historical standardized wind speed value at the corresponding moment according to the vertex index set, synchronously extracting the historical standardized wind direction value at the corresponding moment to generate a simplex vertex historical data packet.

[0010] Optionally, in the third implementation manner of the first aspect of the present application, the method comprises: receiving a simplex vertex index set, retrieving a standardized wind speed value at a corresponding time stamp from a historical database according to the index, synchronously retrieving a standardized wind direction value at the corresponding time stamp to obtain a simplex vertex wind speed data set and a simplex vertex wind direction data set; calculating a long memory feature parameter based on the time distribution of the vertex data, generating a fractional differential weight coefficient sequence according to the long memory feature parameter, and constructing a fractional differential operator template; performing non-iterative convolution calculation according to the simplex vertex wind speed data set to generate a standardized wind speed prediction sequence, performing non-iterative convolution calculation according to the simplex vertex wind direction data set to generate a standardized wind direction prediction sequence, to obtain a primary wind speed prediction sequence and a primary wind direction prediction sequence; extracting a dynamic turbulence intensity factor sequence, generating a correction coefficient matrix according to the current turbulence intensity value, and applying the correction coefficient to the primary wind speed prediction sequence to obtain a corrected standardized wind speed prediction sequence.

[0011] Optionally, in the fourth implementation manner of the first aspect of the present application, the method comprises: preloading a pitch angle-tip speed ratio aerodynamic efficiency coefficient table, calculating a real-time tip speed ratio according to the current standardized wind speed prediction value, and interpolating the dynamic aerodynamic efficiency coefficient sequence in the coefficient table; constructing an energy surface three-dimensional, and fusing the three-dimensional data into a time-varying continuous energy surface according to the time stamp; calculating the partial derivative of the energy surface in the yaw angle dimension, calculating the partial derivative of the energy surface in the pitch angle dimension, and generating a gradient optimization direction vector sequence by combining the sign function with the square root of the gradient modulus; performing proportional integration along the gradient optimization direction vector to output a yaw angle increment instruction sequence and a pitch angle increment instruction sequence, and superimposing the initial attitude angle to generate a yaw angle absolute instruction sequence and a pitch angle absolute instruction sequence; encapsulating the absolute control instruction into a transmission data packet according to the time stamp, writing into a first-in-first-out instruction buffer, and obtaining a control instruction data packet.

[0012] Optionally, in the fifth implementation manner of the first aspect of the present application, the energy capture efficiency value of the energy curve is set as : ; wherein, is a normalized wind speed prediction value; is a current yaw angle of the wind turbine; is a normalized wind direction prediction value; is a decay coefficient; is a dynamic aerodynamic efficiency coefficient.

[0013] Optionally, in the sixth implementation manner of the first aspect of the present application, the method comprises: calling a yaw angle absolute instruction sequence and a pitch angle absolute instruction sequence to drive the yaw system and the pitch system of the wind turbine to perform actions, synchronously collecting actual output power sequences, actual wind speed sequences and actual wind direction sequences within a preset time period after the actions are performed to generate a unit response data set; taking the corrected normalized wind speed prediction sequence as an input wind speed and taking the normalized wind direction prediction sequence as an input wind direction, in combination with the currently performed yaw angle absolute instruction sequence and the pitch angle absolute instruction sequence, querying a preloaded pitch angle-tip speed ratio aerodynamic efficiency coefficient table to obtain a theoretical output power sequence; aligning the actual output power sequence and the theoretical output power sequence according to time stamps, calculating power capture efficiency values element by element, associating yaw angle instruction values, pitch angle instruction values, corrected normalized wind speed prediction values and normalized wind direction prediction values at corresponding time points to generate a multi-dimensional efficiency evaluation matrix; setting an efficiency threshold interval, scanning the multi-dimensional efficiency evaluation matrix, identifying a continuous time period in which the power capture efficiency values are continuously below a lower threshold, extracting the normalized wind speed prediction sequence, the normalized wind direction prediction sequence, the yaw angle instruction sequence and the pitch angle instruction sequence corresponding to the time period to generate a low-efficiency operation warning data packet; in response to the low-efficiency operation warning data packet, extracting actual wind speed sequences, actual wind direction sequences and actual output power sequences within the warning time period, inversely deducing actual aerodynamic efficiency coefficients based on the relationship between the actual wind speed and the tip speed ratio, comparing the theoretical values of corresponding working points in the preloaded aerodynamic efficiency coefficient table, calculating calibration deviations and updating the coefficient table to generate a calibrated aerodynamic efficiency coefficient table; analyzing the yaw angle instruction distribution and the pitch angle instruction distribution in the high-efficiency operation area of the multi-dimensional efficiency evaluation matrix, counting high-frequency instruction intervals, updating the boundary values of the high-frequency instruction intervals as the yaw angle control threshold and the pitch angle control threshold to generate an optimized control threshold parameter set. The multi-dimensional efficiency evaluation matrix, the low-efficiency operation warning data packet, the calibrated aerodynamic efficiency coefficient table and the optimized control threshold parameter set are packaged to generate a wind turbine wind power efficiency evaluation report.

[0014] Optionally, in the seventh implementation method of the first aspect of the present invention, it also includes sensor abnormality prediction inversion verification: receiving the wind turbine power output measurement value in real time, reversely calculating the theoretical wind speed and direction range based on the current control instruction, and generating a sensor abnormality confidence factor when the measured sensor data continues to exceed the theoretical range; starting when the abnormality confidence factor exceeds the threshold, extracting the time and space nearest neighbor point set from the dynamic triangular mesh topology, calling the historical standardized data at the corresponding moment to replace the real-time sensor input, and obtaining an abnormal working condition replacement data packet; inputting the abnormal working condition replacement data packet into the fractional-order differential joint prediction module, skipping the optimization calculation and directly calling the historical optimal instruction template to generate a fault-tolerant control instruction stream; monitoring the power recovery slope after the fault-tolerant control instruction is executed, switching back to the normal control mode when the slope meets the standard, and recording the abnormal event characteristics to the fault knowledge base.

[0015] The second aspect of the present invention provides a wind turbine generator system power generation efficiency optimization system based on wind direction and wind speed monitoring, the wind turbine generator system power generation efficiency optimization system based on wind direction and wind speed monitoring includes: an acquisition module for acquiring an original wind speed sequence and an original wind direction sequence, calling a rated wind speed parameter to perform a division calculation on the original wind speed sequence to generate a standardized wind speed sequence, calling a circular angle constant to perform a division calculation on the original wind direction sequence to generate a standardized wind direction sequence; a construction module for constructing a three-dimensional phase space vector with the standardized wind speed sequence and the standardized wind direction sequence as input sources, and generating a dynamic triangular mesh topology structure based on a real-time updated historical data stream; a generation module for generating a dynamic triangular mesh topology structure based on the dynamic triangular mesh The simplex to which the current phase space vector belongs is located in the grid topology structure, the historical standardized wind speed and wind direction data of all vertices of the simplex are extracted, a closed-form calculation is performed on the vertex data based on a fractional-order differential operator, and a standardized wind speed prediction sequence and a standardized wind direction prediction sequence are output; an analysis module is used to construct a continuously differentiable energy capture efficiency surface with the standardized wind speed prediction sequence and the standardized wind direction prediction sequence as input sources, and the yaw angle command sequence and the pitch angle command sequence are obtained by analysis along the gradient direction of the efficiency surface; an allocation module is used to generate a wind turbine wind power efficiency evaluation report based on the yaw angle command sequence and the pitch angle command sequence, and optimize the wind turbine according to the wind power efficiency evaluation report.

[0016] The mechanism of the present invention is as follows: by constructing a phase space topological structure that integrates spatiotemporal dynamic characteristics, combining fractional-order differential prediction with gradient optimization control, precise coordinated control of wind turbine posture under complex wind field conditions is achieved.

[0017] Beneficial effects: generate enhanced standardized wind speed sequence and standardized wind direction sequence with varying amount of features, deeply mine data hidden information, provide rich and valuable data for accurate prediction and control, improve data utilization efficiency and application value, better adapt to nonlinear, non-stationary and turbulence affected wind conditions through the introduction of dynamic turbulence intensity factor and the generation of continuous wind direction sequence, provide reliable data support for stable operation and efficient power generation of wind turbine generators; The closed-form calculation of the vertex data with fractional differential operators better captures the nonlinear and long memory characteristics of the data through its memory and genetic characteristics, making the prediction sequence more accurate and providing accurate input for optimal control, and the results are corrected by the dynamic turbulence intensity factor, considering the influence of turbulence on wind speed prediction, making the prediction more realistic and enhancing the practicality and adaptability of the prediction model under different wind conditions; An energy capture efficiency surface is constructed to comprehensively reflect the energy capture efficiency of the unit by considering multiple factors, and the gradient analytical instruction sequence along the surface is realized to achieve global optimal control, so that the unit can approach the optimal operating state under various wind conditions, improve power generation efficiency, and real-time generation of yaw angle and pitch angle instruction sequence to make the unit respond to wind condition changes in time and adjust the posture to improve the adaptability to complex wind conditions and improve power generation efficiency and operating stability; Synchronous data streams are generated by extracting timestamps, the instructions are converted into relative increments and dead zones are compensated, S-shaped speed planning curves are output to generate smooth driving pulses, ensuring that the instructions are real-time, accurate and reliable, so that the motor runs smoothly, reduces loss and vibration, and improves service life and efficiency; An abnormal sensor prediction inversion verification mechanism is set up, and when an abnormality is detected, historical data is used to replace real-time input to generate fault-tolerant instruction stream, ensuring stable operation of the system, avoiding control failure and power generation efficiency due to sensor problems, and improving system reliability and robustness; After fault-tolerant control, monitor the power recovery slope, switch back to normal mode after reaching the standard and record the abnormal event characteristics to the fault knowledge base, provide reference for system optimization and maintenance, help researchers improve the system and improve performance and reliability, and realize long-term stable and efficient operation. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 An embodiment of the wind turbine generator power generation efficiency optimization method based on wind direction and wind speed monitoring in the embodiment of the application is shown in the figure; Figure 2 Another embodiment of the wind turbine generator power generation efficiency optimization method based on wind direction and wind speed monitoring in the embodiment of the application is shown in the figure; Figure 3 An embodiment of the wind turbine generator power generation efficiency optimization system based on wind direction and wind speed monitoring in the embodiment of the application is shown in the figure; Figure 4An embodiment of the wind direction and wind speed monitoring based wind turbine power generation efficiency optimization device in the embodiments of the present application is shown. DETAILED DESCRIPTION

[0019] The wind direction and wind speed monitoring based wind turbine power generation efficiency optimization method and system in the embodiments of the present application are used to realize long-term stable and efficient operation of the wind turbine. The terms “first”, “second”, “third”, “fourth” and the like (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms “include” or “have” and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0020] For the sake of understanding, the specific process of the embodiments of the present application is described below. Please refer to Figure 1 An embodiment of the wind direction and wind speed monitoring based wind turbine power generation efficiency optimization method in the embodiments of the present application includes: 101, real-time sensing data acquisition and standardized calculation, obtaining an original wind speed sequence through a wind speed sensor and obtaining an original wind direction sequence through a wind direction sensor; calling a rated wind speed parameter to perform division calculation on the original wind speed sequence to generate a standardized wind speed sequence; calling a circumferential angle constant to perform division calculation on the original wind direction sequence to generate a standardized wind direction sequence; It can be understood that the execution subject of the present application can be a wind direction and wind speed monitoring based wind turbine power generation efficiency optimization system, and can also be a terminal or a server, which is not limited here. The embodiments of the present application take the server as the execution subject for example.

[0021] It should be noted that the real-time monitoring of a certain wind farm on May 15, 2025 from 10:00 to 10:05 is taken as an example to specifically explain the implementation process of step 101. Sensor data acquisition, original wind speed sequence: the ultrasonic wind speed sensor (range 0-60 m / s, accuracy ±0.3 m / s) deployed at the top of the wind turbine collects data at a frequency of 1 Hz, and obtains 300 original wind speed values (unit: m / s) within 5 minutes: ; Original wind direction sequence: synchronous acquisition of ultrasonic wind direction data (range 0-360°, accuracy ±3°): ; Wind speed standardization calculation, call rated wind speed parameter: the rated wind speed of this fan is 15 m / s (set according to fan model).

[0022] Perform division calculation: divide each original wind speed value by rated wind speed, generate dimensionless standardized wind speed sequence: example: ; Output sequence: ; Wind direction standardization calculation, call circumferential angle constant: use circumferential angle constant 360° (fixed parameter).

[0023] Perform division calculation: divide each original wind direction value by 360°, generate standardized wind direction sequence in [0, 1) interval: example: ; Output sequence: .

[0024] 102、Dynamic phase space topology construction calculation, take the standardized wind speed sequence and the standardized wind direction sequence as input sources; construct a three-dimensional phase space vector: the first dimension is the current standardized wind speed value, the second dimension is the product of the historical standardized wind speed value and the sine of the wind direction change, and the third dimension is the current standardized wind direction value; based on the real-time updated historical data stream, dynamically generate a triangular mesh topology structure through the Delaunay subdivision algorithm; It should be noted that the following takes the real-time monitoring data of a certain wind farm on May 15, 2025, 10:00-10:05 as an example: Input data preparation, standardized wind speed sequence (from step 101): (300 values in total, sampling frequency 1 Hz); Standardized wind direction sequence (from step 101): (unit: circumferential proportion, range [0, 1)); Three-dimensional phase space vector construction, generate a three-dimensional vector for each real-time data at a time: First dimension (current wind speed): directly take the standardized wind speed value at the current time: at 10:00:02, ; Second dimension (product of historical wind speed and sine of wind direction change): take the standardized wind speed at the previous time (historical wind speed); calculate the wind direction change (unit radian) between the current and the previous time: ; Calculate the product: : at 10:00:02, , ; Third dimension (current wind direction): directly take the current standardized wind direction value. At 10:00:02, ; The generated three-dimensional vector example is shown in Table 1: Table 1 Three-dimensional vector

[0025] 5 minutes, 300 vectors are generated, forming a dynamic point set.

[0026] Delaunay triangulation is dynamically constructed, and the initial grid is generated: taking the first vector (0.613, 0, 0.425) as the starting point, an initial tetrahedral grid covering the entire data range is generated (three-dimensional Delaunay requires tetrahedron, but the principle is the same as triangulation).

[0027] Point-by-point insertion and update: new point insertion: insert the new vector (0.633, 0.023, 0.431 at 10:00:02) into the current grid. Local reconstruction: find the tetrahedron containing the new vector; connect the new point with the tetrahedron vertices, and split into 4 small tetrahedrons; check the circumscribed sphere property: if the circumscribed sphere of a tetrahedron contains other points, perform the Flip operation to ensure that no point is inside the circumscribed sphere. Dynamic maintenance: every time a new vector is added (0.1 seconds each time), repeat the above process to update the grid topology in real time.

[0028] Output structure: finally generate an unstructured grid composed of triangular facets (three-dimensional partition surface), each triangular facet corresponds to a local region of phase space: at 10:00:05, the grid contains 297 triangular facets, covering all 300 vector points.

[0029] 103、Fractional differential joint prediction calculation, locate the current phase space vector in the dynamic triangular grid topology structure; extract the historical standardized wind speed and wind direction data of all vertices of the simplex; based on the fractional differential operator, perform closed-form calculation on the vertex data, output the standardized wind speed prediction sequence and the standardized wind direction prediction sequence; It should be noted that the following is an example of real-time data of a certain wind farm on May 15, 2025, 10:00-10:05: Simplex positioning and vertex data extraction, the standardized phase space vector at 10:00:03 is (the second dimension is calculated by the previous time sequence: ).

[0030] Simplex positioning: in the dynamic Delaunay triangular grid (containing 297 tetrahedrons) generated in step 102, find the tetrahedron containing V through the point positioning algorithm.

[0031] Assume the positioning result is a tetrahedron T, whose four vertices correspond to historical time data: vertex 1: vector (0.613, 0, 0.425) at 10:00:00; vertex 2: vector (0.633, 0.023, 0.431) at 10:00:01; vertex 3: vector (0.673, 0.031, 0.439) at 10:00:02; vertex 4: vector (0.580, 0.018, 0.422) at 9:59:58; Vertex data extraction: Extract the historical standardized wind speed sequence and wind direction sequence corresponding to the four vertices (length: the latest 5 sampling points): wind speed sequence: vertex 1 = [0.58, 0.60, 0.61, 0.62, 0.613], vertex 2 = [0.61, 0.62, 0.63, 0.63, 0.633], etc. Wind direction sequence: vertex 1 = [0.42, 0.421, 0.423, 0.424, 0.425], vertex 2 = [0.428, 0.429, 0.430, 0.430, 0.431], etc.

[0032] Fractional differential closed-form calculation based on Grünwald-Letnikov fractional differential definition (order ), joint prediction is performed on the tetrahedron vertex data: Estimation calculation: Taking wind speed prediction as an example, weighted sum of vertex 1 sequence: , the weight is generated by the fractional differential operator (negatively related to the time distance, recent data weight is higher): vertex 1 wind speed prediction value = 0.621, vertex 2 = 0.642, vertex 3 = 0.681, vertex 4 = 0.598.

[0033] Correction value fusion: The estimated values of the tetrahedron vertices are fused by weighting the barycentric coordinates of V in the tetrahedron: assume the barycentric coordinates of V are (0.2, 0.3, 0.4, 0.1); ; similarly, the wind direction prediction value is 0.436 (standardized value) through angle cycle processing.

[0034] Prediction sequence output, repeat the above process to generate the standardized prediction sequence of the next 5 seconds as shown in Table 2: Table 2 Standardized prediction sequence

[0035] Fractional differential captures the long-term memory characteristics of historical data through non-integer order (0 ) is more suitable for the nonlinear changes of wind speed / wind direction than integer order models.

[0036] 104. Continuous energy surface gradient optimization calculation, with the normalized wind speed prediction sequence and the normalized wind direction prediction sequence as input sources; construct a continuous differentiable energy capture efficiency surface, which is defined by three products: (a) the cube of the normalized wind speed prediction value; (b) an exponential decay function with the yaw angle deviation from the predicted wind direction as the variable; (c) the aerodynamic efficiency coefficient table related to the pitch angle; Calculate the yaw angle command sequence and the pitch angle command sequence along the gradient direction of the efficiency surface; It should be noted that the following is an example of real-time data of a certain wind farm on May 15, 2025, from 10:00 to 10:05: Input data preparation, normalized wind speed prediction sequence (from step 103): (dimensionless, normalized based on the rated wind speed of 15 m / s); normalized wind direction prediction sequence (from step 103): (circumferential proportion, range [0, 1), corresponding to true angles 157.7°, 158.8°); Energy capture efficiency surface construction, for each future time, construct a surface function, take the 10:00:04 time as an example: wind speed cube term (a): take the normalized wind speed prediction value 0.662, calculate the cube: ; Physical meaning: wind energy power is proportional to the cube of wind speed, reflecting the basic energy potential.

[0037] Yaw deviation decay term (b): define an exponential decay function: ; : current yaw angle of the wind turbine (assuming 30°); : predicted wind direction normalized value 0.438 True angle 157.7°; deviation angle: (normalized to the range of 0°-180°); Take the decay coefficient k = 0.02 (empirical parameter), calculate: ; Effect: the larger the yaw deviation, the more significant the energy capture decay.

[0038] Pitch angle efficiency coefficient (c): look up the aerodynamic efficiency coefficient table (example fragment) as shown in Table 3: Table 3 Look up aerodynamic efficiency coefficient

[0039] Assuming the current pitch angle is 5°, the coefficient value is 0.92 according to the table; Final efficiency value calculation: ; Gradient optimization calculation, analytically solve the optimal control command along the gradient direction of the efficiency surface: Yaw angle optimization: calculate the partial derivative of efficiency with respect to yaw angle , adjust in the direction of gradient ascent: current ; command value: (reduce wind direction deviation); Pitch angle optimization: calculate the partial derivative of efficiency with respect to pitch angle , adjust in the direction of gradient ascent: current ; command value: (fine-tune to reduce aerodynamic loss); Gradient calculation needs to ensure that the surface is differentiable, and the exponential decay function and table lookup interpolation (linear interpolation) need to meet this condition. Output command sequence, repeat the above process to generate future 5-second commands as shown in Table 4:

[0040] Table 4 Future 5-second commands

[0041] 105、Based on the yaw angle command sequence and the pitch angle command sequence, generate a wind turbine wind power efficiency evaluation report, and optimize according to the wind turbine wind power efficiency evaluation report.

[0042] It should be noted that the real-time monitoring data of a certain wind farm on May 15, 2025, 10:00-10:05, based on the yaw angle command sequence (10:00:04 is 31.2°, 10:00:05 is 32.1°) and the pitch angle command sequence (10:00:04 is 4.8°, 10:00:05 is 4.5°) generated in step 104, generate an efficiency evaluation report and perform optimization.

[0043] Real-time control command execution: wind turbine control system responds to command sequence, dynamically adjusts yaw system (yaw angle from 30°→31.2° at 10:00:04) and variable pitch system (pitch angle from 5°→4.8°).

[0044] 10:00:05 continue to execute the command: yaw angle 32.1°, pitch angle 4.5°.

[0045] Efficiency evaluation index calculation (take 10:00:04 as an example): pre-execution efficiency: based on step 104, the original efficiency value is: ; Post-execution efficiency: the actual capture efficiency after adjustment is improved to 0.236 (measured value).

[0046] Improvement rate: ; The evaluation report is generated as shown in Table 5: ​Table 5 evaluation report

[0047] In the embodiments of the present application, the original wind speed and wind direction data are converted into dimensionless or standardized sequences in a specific interval by division operation with the rated wind speed and the constant of the circumferential angle. This processing method eliminates the influence of the data dimension and range difference, makes the data of different sources and scales comparable, provides a unified and standardized data basis for subsequent complex calculation and analysis, and improves the accuracy and stability of the model; three-dimensional phase space vectors are constructed and Delaunay subdivision algorithm is used to dynamically generate triangular mesh topology. This innovative method can capture the dynamic characteristics of wind speed and wind direction in real time, and convert complex multivariate time series data into intuitive topological structure. Through this structure, the relationship between wind speed and wind direction at different times can be more accurately described, providing more rich information for subsequent prediction and optimization, which helps to improve the accuracy of prediction and the effect of optimization; the history data is calculated by using fractional differential operator to realize the joint prediction of standardized wind speed and wind direction. Fractional differential has the characteristic of non-integer order, which can capture the long-term memory characteristics of historical data, and compared with traditional integer order model, it is more suitable for nonlinear changes of wind speed and wind direction. Through joint prediction, the mutual influence between wind speed and wind direction is considered, which improves the accuracy and reliability of prediction, and provides more accurate basis for subsequent optimization control.

[0048] Please refer to Figure 2 Another embodiment of the wind turbine power generation efficiency optimization method based on wind direction and wind speed monitoring in the embodiments of the present application includes: 201, real-time sensing data acquisition and standardized calculation, obtaining the original wind speed sequence through the wind speed sensor, and obtaining the original wind direction sequence through the wind direction sensor; calling the rated wind speed parameter to perform division calculation on the original wind speed sequence to generate a standardized wind speed sequence; calling the constant of the circumferential angle to perform division calculation on the original wind direction sequence to generate a standardized wind direction sequence; In particular, the original sensor data stream capture, real-time acquisition of the original wind speed sequence by the wind speed sensor; real-time acquisition of the original wind direction sequence by the wind direction sensor; product: time-synchronized original wind speed sequence and original wind direction sequence; dynamic turbulence intensity factor calculation, segmenting and intercepting the original wind speed sequence based on a preset time window; calculating the statistical variance value of each segment of wind speed data; dividing the statistical variance value by the average wind speed value of the segment to generate a dynamic turbulence intensity factor sequence; product: dynamic turbulence intensity factor sequence; enhanced standardized wind speed sequence generation, calling a preset rated wind speed parameter; performing element-by-element division calculation on the original wind speed sequence to generate a basic standardized wind speed sequence; fusing the dynamic turbulence intensity factor sequence as an additional dimension with the basic standardized wind speed sequence; product: enhanced standardized wind speed sequence containing turbulence intensity characteristics; wind direction continuity correction, identifying angle jump events of adjacent data points in the original wind direction sequence; injecting an angle compensation amount at the jump point to minimize the adjacent angle difference; generating a physically continuous corrected wind direction sequence; product: continuous wind direction sequence; composite standardized wind direction sequence generation, calling a preset circular angle constant; performing element-by-element division calculation on the continuous wind direction sequence to generate a basic standardized wind direction value; calculating the difference value of adjacent timestamp standardized wind direction values; binding the wind direction change amount as an additional dimension with the basic standardized wind direction value; product: standardized wind direction sequence containing change amount characteristics.

[0049] It should be noted that the data is derived from the actual monitoring (sampling frequency 1 Hz) of a certain 2MW wind turbine within 60 seconds: Original sensor data stream capture, wind speed sensor acquires original wind speed sequence (unit: m / s): (T0 to T5 seconds); wind direction sensor acquires original wind direction sequence (unit: °): ; product: time-synchronized original data (each 1 second a pair of wind speed-wind direction).

[0050] Dynamic turbulence intensity factor calculation, preset time window: 30 seconds (example takes the first 6 data points). Calculation process: segment wind speed average: m / s; statistical variance: ; turbulence intensity factor: ; product: dynamic turbulence intensity factor sequence [0.316] (output one factor every 30 seconds).

[0051] Enhanced standardized wind speed sequence generation, preset rated wind speed parameter: 15 m / s (wind turbine rated wind speed). Basic standardized wind speed: original wind speed ÷ rated wind speed; ; fuse turbulence intensity: generate a two-dimensional sequence with factor 0.316 as an additional dimension: ; Product: Enhanced normalized wind speed sequence (including turbulence characteristics). Wind direction continuity correction, problem: The original wind direction sequence has an angle jump at T2 seconds ( , actually only changes by 4°, but crosses the 360° boundary). Correction method: Detect the jump point (at time T2): Injection compensation amount: The wind direction at point T2 is increased by 360° and corrected to 362° (to minimize the difference between adjacent points: Only 4° difference). Corrected sequence: Product: A physically continuous sequence of corrected wind directions.

[0052] Composite standardized wind direction sequence generation, circular angle constant: 360°. Basic standardized wind direction: Corrected wind direction ÷ 360°: ; Wind direction change: Calculate the difference in standardized wind direction at adjacent moments: ; Bind additional dimensions: Generate a two-dimensional sequence: ; Product: standardized wind direction sequence with variation characteristics.

[0053] The data flow is shown in Table 6: Table 6 Data flow

[0054] A turbulence intensity factor (0.316) has been incorporated into the normalized wind speed series to form an enhanced feature.

[0055] 202. Dynamic phase space topology construction calculation, using the standardized wind speed sequence and the standardized wind direction sequence as input sources; constructing a three-dimensional phase space vector: the first dimension is the current standardized wind speed value, the second dimension is the sine product of the historical standardized wind speed value and the wind direction change, and the third dimension is the current standardized wind direction value; based on the real-time updated historical data stream, dynamically generating a triangular mesh topology structure using a Delaunay subdivision algorithm; Specifically, the phase space dimension compression calculation extracts the basic standardized wind speed value in the enhanced standardized wind speed sequence; the wind direction change amount in the standardized wind direction sequence containing the change amount feature is extracted; the standardized wind speed value of the historical time window is multiplied by the sine value of the wind direction change amount at the corresponding moment to generate a composite feature value; the product is a composite feature sequence; three-dimensional phase space vector construction, the first dimension is assigned the basic standardized wind speed value at the current moment; the second dimension is assigned the composite feature value at the delayed moment; the third dimension is assigned the basic standardized wind direction value at the current moment; the three-dimensional coordinate point set is assembled according to the time stamp; the product is a three-dimensional phase space point set containing a time marker; dynamic triangular mesh topology generation, set the dynamic sliding window to retain the latest N three-dimensional coordinate points; perform Delaunay subdivision calculation on the coordinate point set in the window; generate a mesh topology structure composed of tetrahedral units; the product is a dynamically updated tetrahedral mesh topology; real-time position simplex positioning, input the three-dimensional coordinate point at the current moment into the tetrahedral mesh topology; traverse the tetrahedral unit to locate the target simplex containing the point; record the vertex index set of the target simplex; the product is the simplex vertex index set to which the current point belongs; historical data association binding, extract the historical standardized wind speed value at the corresponding moment according to the vertex index set; synchronously extract the historical standardized wind direction value at the corresponding moment; generate a historical data set bound to the simplex vertex; the product is a simplex vertex historical data package.

[0056] It should be noted that the following is based on a 2MW wind turbine 60 seconds within the standardized wind speed sequence [0.55, 0.57, 0.82, 0.78, 0.65, 0.56] (T0-T5 seconds) and the standardized wind direction sequence [0.986, 0.994, 1.006, 0.997, 0.992, 0.983], to illustrate the implementation process of dynamic phase space topology construction: Phase space dimension compression calculation, input: standardized wind speed sequence: ; Standardized wind direction change sequence (adjacent wind direction difference): (T1-T5); Composite feature value generation: historical time window is 1 second (delayed 1 second), calculation: ; T2 moment: ; T3 moment: ; Product: composite feature sequence (T1-T5); Three-dimensional phase space vector construction, coordinate point assembly (time stamp T2 as an example): first dimension (current wind speed): T2 wind speed value 0.57; second dimension (composite feature value delayed 1 second): T1 composite feature 0.0066; third dimension (current wind direction): T2 wind direction value 0.994; three-dimensional coordinate point: (0.57, 0.0066, 0.994); All coordinate point sets (T2-T5) are shown in Table 7: Table 7 All coordinate point set (T2-T5)

[0057] Dynamic triangulation mesh topology generation, dynamic sliding window: keep the latest 4 points (T2-T5); Delaunay triangulation: input 4 three-dimensional points into the triangulation algorithm to generate a tetrahedral mesh. Topology structure: T2, T3, T4, T5 as the vertices of a tetrahedral unit, vertex index set .

[0058] Real-time position simplex body positioning, new point positioning (T6 time new point (0.56, -0.0058, 0.983)): add the new point to the window, remove the earliest point T2, and update the window to the T3-T6 point set. After re-triangulation, the new point is located in the tetrahedron Product: target simplex body vertex index set .

[0059] Historical data association and binding, vertex data extraction (take the simplex body positioned at T6 as an example) as shown in Table 8: Table 8 Vertex data

[0060] Product: simplex body vertex historical data package (binding wind speed, wind direction time series).

[0061] 203, Fractional differential joint prediction calculation, locate the simplex body to which the current phase space vector belongs in the dynamic triangulation mesh topology structure; extract the historical normalized wind speed and wind direction data of all vertices of the simplex body; based on the fractional differential operator, perform closed-form calculation on the vertex data, and output the normalized wind speed prediction sequence and the normalized wind direction prediction sequence; In particular, the simplex vertex data extraction receives a simplex vertex index set; the corresponding timestamp standardized wind speed value is called from the historical database according to the index; the corresponding timestamp standardized wind direction value is synchronously called; the product: the simplex vertex wind speed data set, the simplex vertex wind direction data set; the fractional differential operator construction, the long memory characteristic parameter is calculated based on the time distribution of the vertex data; the fractional differential weight coefficient sequence is generated according to the long memory characteristic parameter; the differential operator template containing the time decay characteristic is constructed; the product: the fractional differential operator template; the closed-form prediction calculation is executed, the simplex vertex wind speed data set is input into the differential operator template; the non-iterative convolution calculation is executed to generate a standardized wind speed prediction sequence; the simplex vertex wind direction data set is input into the same operator template; the non-iterative convolution calculation is executed to generate a standardized wind direction prediction sequence; the product: the primary wind speed prediction sequence, the primary wind direction prediction sequence; the turbulence intensity dynamic correction extracts a dynamic turbulence intensity factor sequence; a correction coefficient matrix is generated according to the current turbulence intensity value; the primary wind speed prediction sequence is applied with the correction coefficient; the product: the corrected standardized wind speed prediction sequence; the prediction result integration output aligns the corrected standardized wind speed prediction sequence with the primary wind direction prediction sequence according to the timestamp; it is packaged into a joint prediction data package; the product: the standardized wind speed and wind direction joint prediction package.

[0062] It should be noted that the following is based on the results of the dynamic phase space topology construction of a certain 2MW wind turbine within 60 seconds (simplex vertex index set corresponding to timestamps T3-T6), which illustrates the implementation process of the fractional differential joint prediction calculation steps, and the specific data is as follows: Simplex vertex data extraction, input: simplex vertex index set ; The historical data calling is shown in Table 9: Table 9 Historical data

[0063] Product: wind speed data set ; wind direction data set ; Fractional differential operator construction, long memory characteristic analysis: according to the time distribution of the vertex data (T3-T6 interval 1 second), the time decay weight is calculated. It is assumed that the influence of historical data on the current prediction decays with distance, and the decay coefficient is set to (T6 time weight is the highest, T3 is the lowest).

[0064] Differential operator template generation: normalize the decay coefficient to the weight sequence , construct the convolution template. Product: fractional differential operator template ; Closed-form prediction calculation execution, wind speed prediction calculation: wind speed data set and operator template convolution: ; output primary wind speed prediction at T7: 0.74; Wind direction prediction calculation: wind direction dataset convolved with same template: ; output primary wind direction prediction at T7: 0.998; Product: primary prediction sequence (T7 time instant); Turbulence intensity dynamic correction, input: current turbulence intensity factor (value 0.15 at T6 time instant); Correction factor generation: turbulence intensity positively correlates with wind speed fluctuation, set correction function as . ; wind speed prediction correction: corrected T7 wind speed prediction: 0.64; product: corrected wind speed prediction sequence [0.64] (T7 time instant); Prediction result integration output, data alignment: corrected wind speed 0.64 and primary wind direction 0.998 at T7 time instant are bound by time stamp.

[0065] Joint prediction package generation as table 10: Table 10 joint prediction package

[0066] 204、Continuous energy surface gradient optimization calculation, taking the normalized wind speed prediction sequence and the normalized wind direction prediction sequence as input sources; construct a continuous differentiable energy capture efficiency surface, which is defined by three products: (a) the cube of the normalized wind speed prediction value; (b) an exponential decay function with the yaw angle and the deviation of the predicted wind direction as variables; (c) an aerodynamic efficiency coefficient table related to the pitch angle; Analytically calculate the yaw angle command sequence and the pitch angle command sequence along the gradient direction of the efficiency surface; Specifically, the aerodynamic efficiency coefficient is dynamically mapped, a preloaded pitch angle-tip speed ratio aerodynamic efficiency coefficient table is obtained; a real-time tip speed ratio is calculated according to a current standardized wind speed prediction value; a dynamic aerodynamic efficiency coefficient sequence is obtained by interpolation in the coefficient table; and a product is the dynamic aerodynamic efficiency coefficient sequence; a three-dimensional energy surface is constructed, a first dimension is a cubic operation result of a standardized wind speed prediction sequence; a second dimension is a negative exponential function value with a yaw angle and a predicted wind direction deviation as independent variables; a third dimension is the dynamic aerodynamic efficiency coefficient sequence; the three-dimensional data is fused into a continuous energy surface according to a time stamp; a product is a time-varying continuous energy surface; a sign function gradient is analyzed, a partial derivative of the energy surface in the yaw angle dimension is calculated; a partial derivative of the energy surface in the pitch angle dimension is calculated; an optimization direction vector is generated by combining a sign function and a gradient modulus square root; a product is a gradient optimization direction vector sequence; a continuous control instruction is generated, a proportional integral is performed along the gradient optimization direction vector; a yaw angle increment instruction sequence and a pitch angle increment instruction sequence are output; an initial attitude angle is superimposed to generate absolute control instructions; products are a yaw angle absolute instruction sequence and a pitch angle absolute instruction sequence; instruction flow buffering output is performed, the absolute control instructions are packaged into transmission data packets according to the time stamp; and the transmission data packets are written into a first-in-first-out instruction buffer; and a product is a time-aligned control instruction data packet.

[0067] It should be noted that the following is based on a 2MW wind turbine 60-second standardized wind speed prediction sequence [0.85, 0.82, 0.78] (T0-T2 seconds) and a standardized wind direction prediction sequence [0.98, 0.95, 0.93] (T0-T2 seconds), to illustrate the implementation process of the continuous energy surface gradient optimization step: The aerodynamic efficiency coefficient is dynamically mapped, and a preloaded aerodynamic efficiency coefficient table (pitch angle vs. tip speed ratio) is as shown in Table 11: Table 11 Aerodynamic efficiency coefficient

[0068] Real-time tip speed ratio calculation: ; T0 moment: standardized wind speed prediction value m / s; assuming impeller radius 40 m, rotating speed ; Interpolation to obtain aerodynamic efficiency coefficient: interpolation at pitch angle 5°, tip speed ratio 0.94 Dynamic aerodynamic efficiency coefficient 0.32; product: dynamic aerodynamic efficiency coefficient sequence (T0-T2); Three-dimensional construction of energy surface, set the energy capture efficiency value of the energy surface to : ; Among them, is the standardized wind speed prediction value; Current yaw angle of wind turbine (sensor real-time measurement, standardized: 350° / 360°); Standardized wind direction prediction value; Decay coefficient (preset constant, 100 in the case, calibrated by wind turbine aerodynamic characteristics); Dynamic aerodynamic efficiency coefficient; First dimension (wind speed cubed): T0 moment: ; Second dimension (yaw deviation index decay): Current yaw angle of wind turbine: 350° (standardized value ); Predicted wind direction deviation: ; Negative exponential function: ; Third dimension (aerodynamic efficiency coefficient): 0.32; Three-dimensional energy value: ; All energy surface points (T0-T2) are shown in Table 12: Table 12 All energy surface points (T0-T2)

[0069] Symbolic function gradient analysis, yaw angle dimension gradient: Calculate the partial derivative of the energy surface at T0 point with respect to the yaw angle: Fine-tune the yaw angle (+0.1°) → New deviation | (350.1 / 360)-0.98|=0.0078 → New index value e^(-0.78)≈0.458 → New energy value 0.614×0.458×0.32≈0.090; ; Pitch angle dimension gradient: Fine-tune the pitch angle (+0.1°) → New aerodynamic coefficient (interpolation table) 0.325 → New energy value 0.614×0.449×0.325≈0.089; Partial derivative=(0.089-0.088) / 0.1=0.01; Gradient optimization direction vector: Symbolic function combined gradient modulus square root: [sign(0.02)×√(0.02),sign(0.01)×√(0.01)]≈[0.141,0.100]; Product: Gradient vector sequence [(0.141,0.100)] (T0 moment); Continuous control instruction generation, proportional integral generation of incremental instruction: Yaw angle increment: ; Pitch angle increment: ; Superimpose initial attitude angle: Current yaw angle 350° → Absolute instruction 350.0705°; Current pitch angle 5° → Absolute instruction 5.050°; Product as shown in Table 13: Table 13 Product

[0070] Instruction stream buffer output, instruction encapsulation: encapsulate the absolute instructions at T0-T2 time stamp as table 14: Table 14 Instruction stream buffer output

[0071] Write buffer: store in first-in first-out queue in time sequence, wait for execution.

[0072] 205, generate a wind turbine wind power efficiency evaluation report based on the yaw angle instruction sequence and the pitch angle instruction sequence, and optimize according to the wind turbine wind power efficiency evaluation report.

[0073] Specifically, the yaw angle absolute instruction sequence and the pitch angle absolute instruction sequence in the control instruction data packet are called to drive the yaw system and the pitch system of the wind turbine generator to execute actions, and the actual output power sequence, the actual wind speed sequence and the actual wind direction sequence within a preset time period after the execution of the actions are synchronously collected to generate a unit response data set; the corrected normalized wind speed prediction sequence is taken as the input wind speed, the normalized wind direction prediction sequence is taken as the input wind direction, the preloaded pitch angle-tip speed ratio aerodynamic efficiency coefficient table is queried in combination with the currently executed yaw angle absolute instruction sequence and the pitch angle absolute instruction sequence, and the theoretical output power sequence is calculated; the actual output power sequence and the theoretical output power sequence are aligned by time stamp, the power capture efficiency value (actual power / theoretical power) is calculated element by element, the yaw angle instruction value, the pitch angle instruction value, the corrected normalized wind speed prediction value and the normalized wind direction prediction value at the corresponding time are associated, and a multi-dimensional efficiency evaluation matrix is generated; an efficiency threshold interval is set, the multi-dimensional efficiency evaluation matrix is scanned, a continuous time period in which the power capture efficiency value continuously falls below the lower threshold value is identified, the normalized wind speed prediction sequence, the normalized wind direction prediction sequence, the yaw angle instruction sequence and the pitch angle instruction sequence corresponding to the time period are extracted, and a low-efficiency operation warning data packet is generated; in response to the low-efficiency operation warning data packet, the actual wind speed sequence, the actual wind direction sequence and the actual output power sequence within the warning period are extracted, the actual aerodynamic efficiency coefficient is back calculated based on the relationship between the actual wind speed and the tip speed ratio, the theoretical value of the corresponding working condition point in the preloaded aerodynamic efficiency coefficient table is compared, the calibration deviation is calculated and the coefficient table is updated, and a calibrated aerodynamic efficiency coefficient table is generated; the yaw angle instruction distribution and the pitch angle instruction distribution in the high-efficiency operation area (efficiency value≥upper threshold value) in the multi-dimensional efficiency evaluation matrix are analyzed, the high-frequency instruction interval is counted, the boundary values of the high-frequency instruction interval are updated as the yaw angle control threshold value and the pitch angle control threshold value, and an optimized control threshold parameter set is generated. Encapsulate the multi-dimensional efficiency evaluation matrix, the low-efficiency operation warning data packet, the calibrated aerodynamic efficiency coefficient table and the optimized control threshold parameter set to generate a wind turbine wind power efficiency evaluation report, feed back the calibrated aerodynamic efficiency coefficient table to the aerodynamic efficiency coefficient calling link, and feed back the optimized control threshold parameter set to the gradient optimization direction vector generation link; Note that the yaw angle and pitch angle instruction sequence (60s window) generated by the 2MW wind turbine in step 204 is synchronized with the actual running data, and the control effect is evaluated and the parameters are optimized.

[0074] Unit response data collection, execute instructions: yaw angle instruction: [350.07°, 350.13°, 350.20°] (T0-T2 seconds); pitch angle instruction: [5.05°, 5.09°, 5.14°]; Actual output: actual power: [820kW, 795kW, 770kW]; actual wind speed: [8.5m / s, 8.2m / s, 7.8m / s]; actual wind direction: [352°, 349°, 351°]; Theoretical power calculation, input prediction data: normalized wind speed prediction: [0.85, 0.82, 0.78]; normalized wind direction prediction: [0.98, 0.95, 0.93]; Query aerodynamic efficiency table (pitch angle vs tip speed ratio): T0 moment: pitch angle 5.05°, tip speed ratio 5.2 → interpolated efficiency coefficient 0.315; theoretical power formula: theoretical power = (wind speed 3 ) × aerodynamic coefficient × attenuation function; output theoretical power sequence: [865kW, 830kW, 798kW]; Efficiency evaluation matrix generation, capture efficiency value: actual power / theoretical power → [94.8%, 95.8%, 96.5%]; Correlation parameter matrix as shown in Table 15: Table 15 Correlation parameter matrix

[0075] Low efficiency alarm and parameter calibration, alarm trigger: when the efficiency is continuously <95% (such as T0), extract the corresponding period data.

[0076] Back-propagating actual aerodynamic efficiency: T0 actual power 820kW → back-propagating aerodynamic coefficient 0.302 (original theoretical value 0.315); calibration deviation: -4.1% → update aerodynamic table corresponding position coefficient to 0.302.

[0077] High efficiency interval optimization, statistics of high efficiency area (efficiency ≥96%) instruction distribution: yaw angle high frequency interval: 350.13°-350.20°; pitch angle high frequency interval: 5.09°-5.14°; Update control threshold: feedback the above interval boundaries to the gradient optimization module to constrain subsequent instruction generation.

[0078] Output evaluation report and feedback, report content: multi-dimensional efficiency matrix; low efficiency warning package: T0 period data + calibration deviation; optimization control threshold: yaw angle [350.13°, 350.20°], pitch angle [5.09°, 5.14°]; After calibration, the aerodynamic table is pushed to the aerodynamic efficiency calling module of step 204; the optimization threshold is pushed to the gradient direction vector generation link to narrow the search range.

[0079] 206, also including sensor anomaly prediction inversion verification: Real-time receive wind turbine power output measurement value; reverse calculate the theoretical wind speed and direction interval based on the current control instruction; generate an abnormal flag when the measured sensor data continuously exceeds the theoretical interval; product: sensor anomaly confidence factor; Emergency call of phase space historical data, start when the anomaly confidence factor exceeds the threshold; extract the spatiotemporal nearest neighbor point set from the dynamic triangular mesh topology; call the historical standardized data at the corresponding time to replace the real-time sensor input; product: abnormal condition replacement data package; Control instruction flow reconstruction, input the replacement data package into the fractional order differential joint prediction module; skip the optimization calculation and directly call the historical optimal instruction template; generate a fault-tolerant control instruction sequence; product: fault-tolerant control instruction flow; Perform state closed-loop feedback, monitor the power recovery slope after executing the fault-tolerant control instruction; switch back to normal control mode when the slope meets the standard; record the abnormal event characteristics to the fault knowledge base; product: control mode switching signal, fault feature archive.

[0080] It should be noted that the following is based on the running scene of a 2MW wind turbine within 60 seconds (sampling frequency 1Hz), which illustrates the implementation process of the sensor anomaly prediction inversion verification step (step 206). Assuming that after the current control instruction is generated, the actual power output of the wind turbine continuously deviates from the theoretical expectation, the specific implementation is as follows: Theoretical interval inversion and abnormal flag, input: T0-T5 second control instruction: yaw angle instruction sequence [350.5°, 351.2°, 352.0°], pitch angle instruction sequence [5.1°, 5.3°, 5.6°]; measured power sequence: [1.52, 1.48, 1.25, 1.30, 1.42, 1.38] MW; Theoretical interval calculation (take T2 time as an example): according to the aerodynamic efficiency coefficient table and the instruction, the theoretical wind speed should be 8.0-8.5m / s (standardized wind speed 0.53-0.57), and the theoretical wind direction should be 352-356° (standardized wind direction 0.978-0.989).

[0081] Anomaly detection: T2 measured data: wind speed 7.2 m / s (normalized 0.48), wind direction 358° (normalized 0.994), both out of theoretical interval. 3 consecutive overruns (T2-T4) → trigger anomaly flag, generate confidence factor 0.92 (threshold = 0.9).

[0082] Historical data substitution and fault-tolerant control, phase space emergency call: extract spatiotemporal nearest neighbor points from dynamic topology (T1, T3, T4 time data), form a replacement data package as shown in Table 16: Table 16 Replacement data package

[0083] Fault-tolerant instruction generation: input the replacement data into the prediction module, skip the optimization calculation, and directly call the historical optimal instruction template (T3 time instruction: yaw angle 352.0°, pitch angle 5.6°). Output fault-tolerant instruction sequence: [352.0°, 5.6°] (T6 time).

[0084] Closed-loop feedback and mode switching, power recovery monitoring: after executing the fault-tolerant instruction, T6-T8 seconds power output: [1.55, 1.58, 1.60] MW → recovery slope 0.8% / s (calculation: (1.60-1.55) / 1.55x100% ÷ 3 seconds). Slope exceeds threshold 0.5% / s → trigger switching signal, return to normal control mode.

[0085] Fault feature recording: anomaly event features: sustained low wind speed + reverse jump, record to knowledge base (marked as "wind direction sensor zero drift fault"). Data flow example as shown in Table 17: Table 17 Data flow example

[0086] In the embodiment of the present application, through multi-dimensional data standardization and enhancement processing, dynamic phase space topology construction, fractional differential joint prediction and continuous energy surface gradient optimization, etc., accurate prediction of wind speed and direction and optimal adjustment of fan posture are realized, so that the fan is always in the best power generation state, and the power generation efficiency is improved; the sensor anomaly prediction inversion verification mechanism can monitor sensor data anomalies in real time, enable fault tolerance control in time, avoid fan operation anomalies caused by sensor failure, and enhance the stability and reliability of the system; the dynamic phase space topology construction and continuous energy surface gradient optimization method fully utilizes historical data and real-time data, mines the internal relationship between data, improves the utilization value of data, and provides more accurate basis for fan control; the sensor anomaly prediction inversion verification mechanism can timely discover sensor failure and record abnormal event features to the fault knowledge base, providing reference for operation and maintenance personnel, helping to develop maintenance plans in advance and reduce operation and maintenance costs; various factors such as wind speed, wind direction and turbulence intensity are considered comprehensively, and the system can adapt to complex and variable wind conditions, and realize efficient operation of the fan under different wind conditions.

[0087] The wind turbine power generation efficiency optimization method based on wind direction and speed monitoring in the embodiment of the present application is described above, and the wind turbine power generation efficiency optimization system based on wind direction and speed monitoring in the embodiment of the present application is described below, please refer to Figure 3 The wind turbine power generation efficiency optimization system based on wind direction and speed monitoring in the embodiment of the present application includes: an acquisition module 301, configured to acquire an original wind speed sequence and an original wind direction sequence, call a rated wind speed parameter to perform division calculation on the original wind speed sequence to generate a standardized wind speed sequence, and call a circular angle constant to perform division calculation on the original wind direction sequence to generate a standardized wind direction sequence; a construction module 302, configured to take the standardized wind speed sequence and the standardized wind direction sequence as input sources, construct a three-dimensional phase space vector, and generate a dynamic triangular mesh topology structure based on real-time updated historical data flow; a generation module 303, configured to locate a current phase space vector in the dynamic triangular mesh topology structure, extract historical standardized wind speed and wind direction data of all vertices of the simplex body, perform closed calculation on the vertex data based on a fractional differential operator, and output a standardized wind speed prediction sequence and a standardized wind direction prediction sequence; an analysis module 304, configured to take the standardized wind speed prediction sequence and the standardized wind direction prediction sequence as input sources, construct a continuously differentiable energy capture efficiency surface, and analyze a yaw angle instruction sequence and a pitch angle instruction sequence along the gradient direction of the efficiency surface; and a distribution module 305, configured to generate a fan wind power efficiency evaluation report based on the yaw angle instruction sequence and the pitch angle instruction sequence, and optimize according to the fan wind power efficiency evaluation report.

[0088] In the embodiment of the present invention, the acquisition module standardizes the original wind speed and wind direction sequences, eliminates dimensional differences, makes different data comparable and unified, provides a more accurate and reasonable data basis for subsequent analysis and processing, and helps to improve the accuracy of overall prediction and optimization of the system; the construction module generates a dynamic triangular mesh topology structure based on the real-time updated historical data stream, which can reflect the changing characteristics and trends of wind speed and wind direction in real time. Compared with the static model, it can better adapt to the complex and changeable wind field environment, and improve the system's adaptability and processing efficiency to different wind conditions; the generation module uses fractional differential operators to perform closed-form calculations on vertex data, and the fractional The integer-order differential operator has memory and genetic characteristics, which can better capture the historical information and dynamic change characteristics of the data. Compared with the traditional integer-order differential operator, it can output a more accurate standardized wind speed and wind direction prediction sequence, providing a reliable basis for subsequent optimization; the analytical module constructs a continuously differentiable energy capture efficiency surface, and obtains the yaw angle and pitch angle command sequence by analyzing along the gradient direction of the efficiency surface. It can find the optimal direction of energy capture efficiency, make the wind turbine attitude adjustment more accurate, effectively improve the power generation efficiency of the wind turbine, and ensure that the wind turbine can respond quickly and accurately according to the real-time wind conditions, achieve efficient operation, and improve power generation and energy utilization.

[0089] above Figure 3 The wind turbine generation efficiency optimization system based on wind direction and wind speed monitoring in an embodiment of the present invention is described in detail from the perspective of modular functional entities. The wind turbine generation efficiency optimization device based on wind direction and wind speed monitoring in an embodiment of the present invention is described in detail from the perspective of hardware processing.

[0090] Figure 4 This is a schematic diagram of the structure of a device for optimizing wind turbine power generation efficiency based on wind direction and speed monitoring, provided by an embodiment of the present invention. The device 400 for optimizing wind turbine power generation efficiency based on wind direction and speed monitoring may vary significantly depending on configuration or performance. The device may include one or more central processing units (CPUs) 410 (e.g., one or more processors), a memory 420, and one or more storage media 430 (e.g., one or more mass storage devices) storing application programs 433 or data 432. The memory 420 and storage media 430 may be either transient or persistent storage. The program stored in the storage medium 430 may include one or more modules (not shown), each of which may include a series of instructions for operating the device 400 for optimizing wind turbine power generation efficiency based on wind direction and speed monitoring. Furthermore, the processor 410 may be configured to communicate with the storage medium 430 to execute the series of instructions stored in the storage medium 430 on the device 400 for optimizing wind turbine power generation efficiency based on wind direction and speed monitoring.

[0091] The wind direction and wind speed monitoring based wind turbine power generation efficiency optimization device 400 can also include one or more power supplies 440, one or more wired or wireless network interfaces 450, one or more input / output interfaces 460, and / or one or more operating systems 431, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art can understand that, Figure 4 The illustrated wind direction and wind speed monitoring based wind turbine power generation efficiency optimization device structure does not constitute a limitation to the wind direction and wind speed monitoring based wind turbine power generation efficiency optimization device, and can include more or fewer components than illustrated, or combine certain components, or different component arrangements.

[0092] The present application also provides a wind direction and wind speed monitoring based wind turbine power generation efficiency optimization device, which includes a memory and a processor, the memory stores computer readable instructions, and the computer readable instructions are executed by the processor to make the processor execute the steps of the wind direction and wind speed monitoring based wind turbine power generation efficiency optimization method in each of the above embodiments.

[0093] The present application also provides a computer readable storage medium, which can be a non-volatile computer readable storage medium, or a volatile computer readable storage medium, and the computer readable storage medium stores instructions, and the instructions make the computer execute the steps of the wind direction and wind speed monitoring based wind turbine power generation efficiency optimization method when the instructions are run on the computer.

[0094] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0095] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the entire or part of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0096] The above-described embodiments are merely used to illustrate the technical solutions of the present application, rather than limit the same; even though the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some of the technical features; and these modifications or replacements do not cause the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A wind turbine power generation efficiency optimization method based on wind direction and wind speed monitoring, characterized in that, The wind turbine power generation efficiency optimization method based on wind direction and wind speed monitoring comprises: Obtain the original wind speed sequence and the original wind direction sequence, call the rated wind speed parameter to process the original wind speed sequence to generate a standardized wind speed sequence, and call the circumferential angle constant to process the original wind direction sequence to generate a standardized wind direction sequence; Take the standardized wind speed sequence and the standardized wind direction sequence as input sources to construct a three-dimensional phase space vector, generate a dynamic triangular mesh topology based on real-time updated historical data flow, and output a standardized wind speed prediction sequence and a standardized wind direction prediction sequence. In the dynamic triangular mesh topology, locate the current phase space vector belonging to a simplex, extract the historical standardized wind speed and wind direction data of all vertices of the simplex, perform closed calculation on the vertex data, and output the standardized wind speed prediction sequence and the standardized wind direction prediction sequence. According to the standardized wind speed prediction sequence and the standardized wind direction prediction sequence as input sources, construct a continuously differentiable energy capture efficiency surface, and analytically obtain a yaw angle instruction sequence and a pitch angle instruction sequence along the gradient direction of the efficiency surface. Based on the yaw angle instruction sequence and the pitch angle instruction sequence, generate a wind turbine power generation efficiency evaluation report, and optimize according to the wind turbine power generation efficiency evaluation report.

2. The wind direction and speed based monitoring method for wind turbine generator efficiency optimization as claimed in claim 1, wherein, It comprises: Real-time acquisition of the original wind speed sequence by the wind speed sensor, and real-time acquisition of the original wind direction sequence by the wind direction sensor; Based on the preset time window, the original wind speed sequence is segmented and intercepted, the statistical variance value of each segment of wind speed data is calculated, the statistical variance value is divided by the average wind speed value of the segment, and a dynamic turbulence intensity factor sequence is generated; Call the preset rated wind speed parameter to perform division calculation on the original wind speed sequence element by element to generate a basic standardized wind speed sequence, and fuse the dynamic turbulence intensity factor sequence as an additional dimension with the basic standardized wind speed sequence to obtain an enhanced standardized wind speed sequence; Identify the angle jump event of adjacent data points in the original wind direction sequence, inject an angle compensation amount at the jump point to minimize the adjacent angle difference, and generate a continuous wind direction sequence; Call the preset circumferential angle constant to perform division calculation on the continuous wind direction sequence element by element to generate a basic standardized wind direction value, calculate the difference value of adjacent timestamp standardized wind direction values, and bind the wind direction change amount as an additional dimension with the basic standardized wind direction value to obtain a standardized wind direction sequence.

3. The wind direction and speed based monitoring method for wind turbine generator efficiency optimization of claim 2, wherein, It comprises: Extract the basic standardized wind speed value in the enhanced standardized wind speed sequence, extract the wind direction change amount in the standardized wind direction sequence, multiply the standardized wind speed value of the historical time window with the sine value of the corresponding time wind direction change amount, and generate a composite feature sequence; Construct a three-dimensional phase space vector, and assemble a three-dimensional phase space point set with time markers according to the timestamp; Set a dynamic sliding window to retain the latest N three-dimensional coordinate points, perform Delaunay subdivision calculation on the coordinate point set in the window to generate a dynamically updated tetrahedral mesh topology; Input the current three-dimensional coordinate point into the tetrahedral mesh topology, traverse the tetrahedral unit to locate the target simplex containing the point, record the vertex index set of the target simplex, and obtain the vertex index set of the simplex to which the current point belongs; According to the vertex index set, extract the historical standardized wind speed value at the corresponding time, synchronously extract the historical standardized wind direction value at the corresponding time, and generate a simplex vertex historical data packet.

4. The wind direction and speed based monitoring method for wind turbine generator efficiency optimization of claim 3, wherein, It comprises: Receiving a simplex vertex index set, retrieving the corresponding timestamp standardized wind speed value from the history database according to the index, synchronously retrieving the corresponding timestamp standardized wind direction value, obtaining a simplex vertex wind speed data set and a simplex vertex wind direction data set; Calculating a long memory characteristic parameter based on the time distribution of vertex data, generating a fractional differential weight coefficient sequence according to the long memory characteristic parameter, and constructing a fractional differential operator template; Performing non-iterative convolution calculation based on the simplex vertex wind speed data set to generate a standardized wind speed prediction sequence, performing non-iterative convolution calculation based on the simplex vertex wind direction data set to generate a standardized wind direction prediction sequence, and obtaining a primary wind speed prediction sequence and a primary wind direction prediction sequence; Extracting a dynamic turbulence intensity factor sequence, generating a correction coefficient matrix according to the current turbulence intensity value, and applying the correction coefficient to the primary wind speed prediction sequence to obtain a corrected standardized wind speed prediction sequence.

5. The wind direction and speed based monitoring method for wind turbine generator efficiency optimization as claimed in claim 4 wherein, It includes: Preloading a pitch angle-tip speed ratio aerodynamic efficiency coefficient table, calculating a real-time tip speed ratio according to the current standardized wind speed prediction value, and interpolating the coefficient table to obtain a dynamic aerodynamic efficiency coefficient sequence; Constructing an energy surface three-dimensional, and fusing the three-dimensional data into a time-varying continuous energy surface according to the timestamp; Calculating the partial derivative of the energy surface in the yaw angle dimension and the partial derivative of the energy surface in the pitch angle dimension, and generating a gradient optimization direction vector sequence by combining the sign function with the square root of the gradient modulus; Proportional integration along the gradient optimization direction vector outputs a yaw angle increment instruction sequence and a pitch angle increment instruction sequence, and superimposes the initial attitude angle to generate a yaw angle absolute instruction sequence and a pitch angle absolute instruction sequence; Encapsulating the absolute control instruction into a transmission data packet according to the timestamp, writing it into a first-in-first-out instruction buffer, and obtaining a control instruction data packet.

6. The wind direction and speed based monitoring method for wind turbine generator efficiency optimization of claim 5, wherein, The energy capture efficiency value of the energy surface is set as : ； wherein, is a normalized wind speed prediction; is a current wind turbine yaw angle; is a normalized wind direction prediction; is a decay coefficient; is a dynamic aerodynamic efficiency coefficient.

7. The wind direction and speed based monitoring method for wind turbine generator efficiency optimization of claim 5, wherein, It includes: Calling the yaw angle absolute instruction sequence and the pitch angle absolute instruction sequence to drive the yaw system and the variable pitch system of the wind turbine to execute actions, synchronously collecting actual output power sequences, actual wind speed sequences and actual wind direction sequences within a preset time period after the actions are executed, and generating a unit response data set; Taking the corrected standardized wind speed prediction sequence as the input wind speed and the standardized wind direction prediction sequence as the input wind direction, querying the preloaded pitch angle-tip speed ratio aerodynamic efficiency coefficient table combined with the currently executed yaw angle absolute instruction sequence and pitch angle absolute instruction sequence, and obtaining a theoretical output power sequence; Aligning the actual output power sequence and the theoretical output power sequence according to the timestamp, calculating the power capture efficiency value element by element, associating the yaw angle instruction value, the pitch angle instruction value, the corrected standardized wind speed prediction value and the standardized wind direction prediction value at the corresponding time, and generating a multi-dimensional efficiency evaluation matrix; Setting an efficiency threshold interval, scanning the multi-dimensional efficiency evaluation matrix, identifying a continuous time period in which the power capture efficiency value is continuously below the lower threshold, extracting the standardized wind speed prediction sequence, the standardized wind direction prediction sequence, the yaw angle instruction sequence and the pitch angle instruction sequence corresponding to the time period, and generating a low-efficiency operation warning data packet; In response to the low-efficiency operation alarm data packet, actual wind speed sequence, actual wind direction sequence and actual output power sequence in the alarm period are extracted, actual aerodynamic efficiency coefficient is back calculated based on the relationship between actual wind speed and tip speed ratio, and the calibrated deviation is calculated by comparing the theoretical value of the corresponding working condition point in the pre-loaded aerodynamic efficiency coefficient table, and the coefficient table is updated to generate the calibrated aerodynamic efficiency coefficient table; The yaw angle instruction distribution and the pitch angle instruction distribution of the high-efficiency operation region in the multi-dimensional efficiency evaluation matrix are analyzed, the high-frequency instruction interval is counted, the boundary values of the high-frequency instruction interval are updated as the yaw angle control threshold and the pitch angle control threshold, and the optimized control threshold parameter set is generated; The multi-dimensional efficiency evaluation matrix, the low-efficiency operation alarm data packet, the calibrated aerodynamic efficiency coefficient table and the optimized control threshold parameter set are packaged to generate a wind turbine wind power efficiency evaluation report.

8. The wind direction and speed based monitoring method for wind turbine generator efficiency optimization of claim 7, wherein, It also includes sensor anomaly prediction inversion verification: Real-time wind turbine power output measurement values are received, and the theoretical wind speed and wind direction interval is calculated reversely based on the current control instruction. When the measured sensor data continuously exceeds the theoretical interval, a sensor anomaly confidence factor is generated; When the anomaly confidence factor exceeds the threshold, the spatiotemporal nearest neighbor point set is extracted from the dynamic triangular grid topology, the historical standardized data at the corresponding moment is called to replace the real-time sensor input to obtain an abnormal condition substitution data packet; The abnormal condition substitution data packet is input into the fractional order differential joint prediction module, the historical optimal instruction template is directly called by skipping the optimization calculation to generate a fault-tolerant control instruction stream; The power recovery slope after the execution of the fault-tolerant control instruction is monitored, and when the slope meets the standard, the normal control mode is switched back, and the abnormal event characteristics are recorded to the fault knowledge base.

9. A wind turbine generator power generation efficiency optimization system based on wind direction and wind speed monitoring, characterized in that, The wind turbine generator efficiency optimization system based on wind direction and wind speed monitoring includes: An acquisition module is configured to acquire an original wind speed sequence and an original wind direction sequence, perform division calculation on the original wind speed sequence by calling a rated wind speed parameter to generate a standardized wind speed sequence, and perform division calculation on the original wind direction sequence by calling a circular angle constant to generate a standardized wind direction sequence; A construction module is configured to construct a three-dimensional phase space vector by taking the standardized wind speed sequence and the standardized wind direction sequence as input sources, generate a dynamic triangular grid topology structure based on real-time updated historical data flow; A generation module is configured to locate a current phase space vector in the dynamic triangular grid topology structure, extract historical standardized wind speed and wind direction data of all vertices of the simplex body to which the current phase space vector belongs, and perform closed calculation on the vertex data based on a fractional order differential operator to output a standardized wind speed prediction sequence and a standardized wind direction prediction sequence; An analysis module is configured to construct a continuously differentiable energy capture efficiency surface by taking the standardized wind speed prediction sequence and the standardized wind direction prediction sequence as input sources, and analyze a yaw angle instruction sequence and a pitch angle instruction sequence along the gradient direction of the efficiency surface; A distribution module is configured to generate a wind turbine wind power efficiency evaluation report based on the yaw angle instruction sequence and the pitch angle instruction sequence, and perform optimization according to the wind turbine wind power efficiency evaluation report.

Images