Active power optimization and solution method in multi-fan scene

Abstract

The invention provides a method for performing active power optimization in a multi-fan scene. A noise and delay processing mechanism and a model prediction control thought are combined, an NSGA-II algorithm is adopted to optimize a target function containing noise and delay, and it is ensured that a power distribution scheme has the anti-interference capability. The method comprises the following specific steps: firstly, collecting wind power plant operation data (such as wind speed and wind direction), and establishing a mathematical model for converting wind energy into electric energy; thirdly, establishing a fan spindle torque model and a tower thrust model, and analyzing the stress conditions of the fan spindle torque model and the tower thrust model; then, fatigue damage of a fan material is calculated, a noise measurement model is established to simulate random noise in sensor data, and a communication delay model is established to simulate delay in data transmission; and performing multi-objective optimization by using an NSGA-II algorithm, adjusting operation parameters of the fan, constructing a robust optimization model, and enhancing the anti-interference capability of the system. And finally, comparing an optimization result with an ideal condition without noise and delay, and evaluating an effect.

Description

Technical Field

[0001] The present invention belongs to the fields of wind farm control and optimization, communication technology and delay processing, sensor technology and measurement noise processing, robust optimization and control theory, multi-objective optimization algorithm, data analysis and modeling, experimental design and result analysis, and power system automation, and in particular relates to an active power optimization and solution method in a multi-wind turbine scenario. Background Art

[0002] With the expansion of the scale of wind farms in my country, the mechanical structure of wind turbines has become increasingly complex, and the flexibility of components has increased, making wind turbines more susceptible to fatigue damage during long-term operation. Existing wind farm power optimization solution methods have multiple defects: the simplified model leads to low accuracy of power output and cannot fully consider the mutual influence between wind turbines; the fatigue damage monitoring capability is insufficient, and there is a lack of prediction and maintenance of the long-term health status of wind turbines; the real-time and adaptability are poor, making it difficult to cope with complex and changeable wind farm environments; there are limitations in the collaborative optimization of multiple wind turbines, and the coupling relationship between wind turbines cannot be effectively handled. These defects make it difficult for existing methods to simultaneously achieve efficient power output and equipment health management of wind turbine groups, especially in large-scale wind farms, which limits the improvement of overall performance. Therefore, how to reduce the cumulative fatigue damage of wind turbine groups by optimizing their operation strategies has become a key challenge in the field of wind power generation. Summary of the invention

[0003] The purpose of the present invention is to provide an active power optimization and solution method in a multi-wind turbine scenario, propose a noise and delay processing mechanism, use the idea of ​​model predictive control, use historical data for prediction in the absence of delay data, and solve the technical problems existing in the background technology by optimizing the objective function containing noise and delay based on the optimization framework of the NSGA-II algorithm.

[0004] In order to solve the above technical problems, the specific technical solutions of the present invention are as follows:

[0005] A method for optimizing and solving active power in a multi-wind turbine scenario includes the following steps:

[0006] S1. Collect various index data of wind speed, wind direction and wind rotor radius of wind turbines of N wind turbines;

[0007] S2. Establish a wind energy conversion model, a main shaft torque and tower thrust model, and a material fatigue damage calculation model based on various wind turbine indicator data;

[0008] S3. By analyzing the collected data of various indicators of the wind turbine, a measurement noise model is established to filter out noise interference; considering the communication delay in the actual operating environment, a communication delay model is constructed to analyze its potential impact on system performance and ensure the robustness of the model;

[0009] S4. Establish a robust optimization model for the noise model and communication delay model based on the NSGA-II algorithm;

[0010] S5. Compare the optimization results of the fan output power and communication delay time of the model with noise and communication delay with the ideal case without noise and delay.

[0011] Furthermore, a wind energy conversion model is constructed, including the following steps:

[0012] The model is mainly based on the relationship between wind speed and energy conversion efficiency, and uses mathematical formulas to describe the process of converting wind energy into mechanical energy to ensure that the model can reflect the actual operating status.

[0013] S2.1. Establish the wind energy conversion model of the i-th wind turbine through the wind energy capture formula:

[0014] P i =0.5ρS i V i 3 (1)

[0015] Among them, P i is the wind power of the i-th wind turbine, ρ is the air density, which is 1.225 kg / m 3 , S i =πr i 2 is the swept area of ​​the wind wheel, r i is the radius of the rotor of the i-th wind turbine, V i is the wind speed of the i-th wind turbine;

[0016] Mechanical power output of the i-th fan The relationship between the captured wind energy and the power factor C 1 It represents the efficiency of wind energy captured by the i-th wind turbine:

[0017]

[0018] Among them, the power coefficient C 1 It is a nonlinear coefficient, which is affected by factors such as wind speed and wind rotor speed. When the wind speed is low, C 1 As the wind speed increases; when the wind speed exceeds the rated wind speed, the fan power reaches saturation, C 1 Then it declines;

[0019] According to the law of conservation of energy, the output power of wind energy is related to the main shaft torque and angular velocity as follows:

[0020]

[0021] in, The i-th fan is the main shaft torque, ω i is the angular velocity of the i-th wind turbine rotor; calculate the main shaft torque:

[0022]

[0023] Tower thrust Affected by wind speed and aerodynamic forces, proportional to the square of wind speed:

[0024]

[0025] Among them, C 2 is the thrust coefficient, which indicates the force of the fan at different wind speeds; S i =πr i 2 is the swept area of ​​the wind wheel, r i is the radius of the rotor of the i-th wind turbine, V i is the wind speed of the i-th wind turbine;

[0026] S2.2. Establish the i-th wind turbine main shaft torque and tower thrust model:

[0027] Model the main shaft torque and tower thrust of the wind turbine; establish the functional relationship between the main shaft torque and wind speed and load by analyzing the mechanical characteristics of the wind turbine under different working conditions;

[0028] The model is divided into three areas according to wind speed: low wind speed, rated wind speed, and high wind speed; the low wind speed range is V<11.2m / s. In this range, the wind speed output power is low, the mechanical parts are subjected to small forces, and fatigue damage comes from high cycle times; the rated wind speed range is V≈11.2m / s. At the rated wind speed, the wind speed output power reaches the maximum, and fatigue damage accumulates evenly; the high wind speed range is V>11.2m / s. At high wind speeds, the wind turbine performs power reduction, the main shaft torque is reduced, but the tower thrust increases, and fatigue damage accumulation accelerates;

[0029] The main shaft torque and tower thrust model is expressed by a piecewise function:

[0030] Calculation of the main shaft torque:

[0031]

[0032] Tower thrust calculation:

[0033]

[0034] in, is the main shaft torque of the i-th fan, is the tower thrust of the i-th wind turbine, P i is the wind power of the i-th wind turbine, C 1is the power coefficient, ρ is the air density, S i =πr i 2 is the swept area of ​​the wind wheel, V i is the wind speed of the i-th fan, ω i is the angular velocity of the i-th wind turbine rotor;

[0035] S2.3. Calculate the fatigue damage of fan materials:

[0036] Combined with the actual operating environment, a material fatigue damage model is established: the fatigue life of the wind turbine structure is analyzed using the stress-strain relationship and rain flow counting method, and the damage and failure mode of the material during long-term operation are evaluated to provide a basis for maintenance decisions;

[0037] Using the Palmgren-Miner linear cumulative damage theory:

[0038]

[0039] Where D is the cumulative fatigue damage, n i is the stress level The number of cycles under i is the material at stress level F i The maximum number of cycles under the condition is usually obtained through the SN curve.

[0040] Furthermore, S3 specifically includes the following steps:

[0041] S3.1. Establish measurement noise model:

[0042] The measured value y at time t is set to satisfy the following formula:

[0043] y t =x t +N(0,(σx t ) 2 ) (9)

[0044] Where σ is the standard deviation proportionality factor of the noise, x t is the observed value at the current moment;

[0045] S3.2. Establish a communication delay model:

[0046] Setting the delay τ t In [0,τ max ]The interval follows a uniform distribution:

[0047] τ t ~U(0,τ max )(10)

[0048] The observation value that the controller can obtain at time t is:

[0049]

[0050] Among them, x t is the observed value at the current moment, is the observed value after the delay.

[0051] The above model assumes the worst case scenario, that is, the actual value measured by the sensor has a certain degree of delay, and the controller can only use the delayed data for optimal control.

[0052] Further, S4 specifically includes the following steps:

[0053] S4.1. For optimized active power distribution, the goal is to minimize the overall fatigue damage of all wind turbines while considering the optimization goal of wind turbine power distribution. Fatigue damage includes fatigue damage of the main shaft and fatigue damage of the tower. The objective function is defined as minimizing the average fatigue damage of each component and adding the optimization goal of wind turbine power distribution;

[0054] Spindle fatigue damage The formula derivation is as follows:

[0055] Fatigue damage of the main shaft of the i-th fan The fatigue damage of the spindle is related to the cyclic stress of the spindle torque; when set to a short time, the fatigue damage is proportional to the square of the spindle torque:

[0056]

[0057] Among them, k 1 is a coefficient related to material properties and stress cycle characteristics. Since the material and structural parameters are consistent, k 1 Same for all fans; is the main shaft torque of the i-th fan;

[0058] Tower fatigue damage of the i-th wind turbine The fatigue damage of the tower is related to the tower thrust, and the fatigue damage is set to be proportional to the square of the tower thrust:

[0059]

[0060] Among them, k 2 Indicates the damage coefficient of the tower

[0061] Main shaft fatigue damage objective function: Minimize the average value of fatigue damage of all wind turbine main shafts

[0062]

[0063] Tower fatigue damage objective function: Minimize the average fatigue damage of all wind turbine towers

[0064]

[0065] Wind turbine power allocation optimization objective function: Minimize the variance of active power allocation of all wind turbines

[0066]

[0067] The goal is to minimize the fatigue damage of the wind turbine and the variance of power distribution. The comprehensive objective function is as follows:

[0068]

[0069] Where N is the number of fans, is the fatigue damage of the main shaft of the i-th fan, is the fatigue damage of the tower of the i-th wind turbine, is the output mechanical power of the i-th fan, is the average output mechanical power of the i-th wind turbine, α, β, γ are weight coefficients;

[0070] S4.2. Constraints:

[0071]

[0072] in, is the output mechanical power of the i-th wind turbine tower, and its value is a fixed value;

[0073] S4.3, Measurement Noise Robust Optimization:

[0074] For each wind speed and power reference value, a random noise term is added:

[0075]

[0076] in, is the measured wind speed of the ith wind turbine in time period w, is the actual wind speed of the ith wind turbine in time period w, is the power reference value measured by the ith wind turbine in the w time period, is the actual power reference value, and is the noise term of the wind speed and power reference value of the i-th wind turbine, satisfying:

[0077]

[0078] S4.4. Communication delay optimization modeling:

[0079] In practical applications, there is a delay in the data transmission process, that is, at each moment, the data used by the present invention is not collected in real time, but comes from the measurement result of the previous moment. In order to simulate the delay of sensor data transmission in practical applications, the present invention introduces a communication delay model to simulate actual data transmission, and at the same time sets the maximum delay time of the system to 10 seconds in combination with actual engineering applications;

[0080] For the data at each moment, consider the data of the previous normal communication moment:

[0081]

[0082] in, is the wind speed data of the i-th wind turbine at the previous moment, is the power reference value data of the i-th wind turbine at the current moment, It is the power reference value data of the previous moment;

[0083] S4.5. Objective function of the robust optimization model:

[0084]

[0085] Constraints:

[0086]

[0087] Furthermore, S5 specifically includes the following steps:

[0088] Based on the solution of the robust optimization model, the weight coefficients α = 0.3, β = 0.3, γ = 0.4 are obtained, so the comprehensive objective function is as follows:

[0089]

[0090] The active power optimization and solution method in a multi-wind turbine scenario of the present invention has the following advantages: by introducing the model predictive control idea based on noise and delay processing mechanism, combined with the NSGA-II multi-objective optimization algorithm, many technical problems in wind farm power optimization are solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] Figure 1 The present invention is a flow chart of a method for optimizing and solving active power in a multi-wind turbine scenario.

[0092] Figure 2(a) is the variance result diagram before and after the optimization of communication delay time, and Figure 2(b) is the power distribution diagram of 100 wind turbines at time step 68.

[0093] Figure 3(a) shows the optimization time for each time step, Figure 3(b) shows the reference power variance before and after optimization, and Figure 3(c) shows the reduction ratio result.

[0094] Figure 4(a) is the variance result diagram before and after noise suppression optimization, and Figure 4(b) is the power distribution diagram of 100 wind turbines at time step 70.

[0095] Figure 5 This is a diagram showing the comprehensive inhibition effect. DETAILED DESCRIPTION

[0096] In order to better understand the purpose, structure and function of the present invention, the following further describes in detail the active power optimization and solution method in a multi-wind turbine scenario of the present invention in conjunction with the accompanying drawings.

[0097] S1: Collect wind speed, wind direction, and wind rotor radius data of N wind turbines;

[0098] First, the operating data of the fan under different wind speed and load conditions are collected. The data is cleaned and normalized to eliminate noise and outliers, thus providing accurate basic data for the subsequent model construction.

[0099] S2: Based on the various index data of wind turbines, a wind energy conversion model, a main shaft torque and tower thrust model are established, and a calculation model for material fatigue damage is established.

[0100] S2.1. Establish the wind energy conversion model of the i-th wind turbine through the wind energy capture formula:

[0101] P i =0.5ρS i V i 3 (1)

[0102] Among them, P i is the wind power of the i-th wind turbine, ρ is the air density, which is 1.225 kg / m 3 , S i =πr i 2 is the swept area of ​​the wind wheel, r i is the radius of the rotor of the i-th wind turbine, V i is the wind speed of the i-th wind turbine;

[0103] Mechanical power output of the i-th fan The relationship between the captured wind energy and the power factor C 1 It represents the efficiency of wind energy captured by the i-th wind turbine:

[0104]

[0105] Among them, the power coefficient C 1 is a nonlinear coefficient; when the wind speed is low, C 1As the wind speed increases; when the wind speed exceeds the rated wind speed, the fan power reaches saturation, C 1 Then it declines;

[0106] According to the law of conservation of energy, the output power of wind energy is related to the main shaft torque and angular velocity as follows:

[0107]

[0108] in, The i-th fan is the main shaft torque, ω i is the angular velocity of the i-th wind turbine rotor; calculate the main shaft torque:

[0109]

[0110] Tower thrust Affected by wind speed and aerodynamic forces, proportional to the square of wind speed:

[0111]

[0112] Among them, C 2 is the thrust coefficient, which indicates the force of the fan at different wind speeds; S i =πr i 2 is the swept area of ​​the wind wheel, r i is the radius of the rotor of the i-th wind turbine, V i is the wind speed of the i-th wind turbine;

[0113] S2.2. Establish the i-th wind turbine main shaft torque and tower thrust model:

[0114] Model the main shaft torque and tower thrust of the wind turbine; establish the functional relationship between the main shaft torque and wind speed and load by analyzing the mechanical characteristics of the wind turbine under different working conditions;

[0115] The model is divided into three areas according to wind speed: low wind speed, rated wind speed, and high wind speed; the low wind speed range is V<11.2m / s. In this range, the wind speed output power is low, the mechanical parts are subjected to small forces, and fatigue damage comes from high cycle times; the rated wind speed range is V≈11.2m / s. At the rated wind speed, the wind speed output power reaches the maximum, and fatigue damage accumulates evenly; the high wind speed range is V>11.2m / s. At high wind speeds, the wind turbine performs power reduction, the main shaft torque is reduced, but the tower thrust increases, and fatigue damage accumulation accelerates;

[0116] The main shaft torque and tower thrust model is expressed by a piecewise function:

[0117] Calculation of the main shaft torque:

[0118]

[0119] Tower thrust calculation:

[0120]

[0121] in, is the main shaft torque of the i-th fan, is the tower thrust of the i-th wind turbine, P i is the wind power of the i-th wind turbine, C 1 is the power coefficient, ρ is the air density, S i =πr i 2 is the swept area of ​​the wind wheel, V i is the wind speed of the i-th fan, ω i is the angular velocity of the i-th wind turbine rotor;

[0122] S2.3. Calculate the fatigue damage of fan materials:

[0123] Establish material fatigue damage model; use stress-strain relationship and rain flow counting method to analyze the fatigue life of wind turbine structure, and evaluate the damage and failure mode of materials during long-term operation;

[0124] Using the Palmgren-Miner linear cumulative damage theory:

[0125]

[0126] Where D is the cumulative fatigue damage, n i is the stress level The number of cycles under i is the material at stress level F i The maximum number of cycles under .

[0127] S3: By analyzing the collected data of various indicators of the fan, a measurement noise model is established to filter out noise interference; considering the communication delay in the actual operating environment, a communication delay model is constructed to analyze its potential impact on system performance and ensure the robustness of the model.

[0128] S3.1. Establish measurement noise model:

[0129] The measured value y at time t is set to satisfy the following formula:

[0130] y t =x t +N(0,(σx t ) 2 ) (9)

[0131] Where σ is the standard deviation proportionality factor of the noise, x t is the observed value at the current moment;

[0132] S3.2. Establish a communication delay model:

[0133] Setting the delay τ t In [0,τ max ]The interval follows a uniform distribution:

[0134] τ t ~U(0,τ max )(10)

[0135] The observation value that the controller can obtain at time t is:

[0136] x t =x t-τt (11)

[0137] Among them, x t is the observed value at the current moment, x t-τt is the observed value after the delay.

[0138] S4: Establish a robust optimization model for the noise model and communication delay model based on the NSGA-II algorithm.

[0139] Establishment of robust optimization model

[0140] S4.1. Define the objective function as minimizing the average fatigue damage of each component and add the optimization objective of wind turbine power distribution;

[0141] Spindle fatigue damage The formula derivation is as follows:

[0142] Fatigue damage of the main shaft of the i-th fan In a short setting, fatigue damage is proportional to the square of the spindle torque:

[0143]

[0144] Among them, k 1 is a coefficient related to material properties and stress cycle characteristics. Since the material and structural parameters are consistent, k 1 Same for all fans; is the main shaft torque of the i-th fan;

[0145] Tower fatigue damage of the i-th wind turbine Set the fatigue damage to be proportional to the square of the tower thrust:

[0146]

[0147] Among them, k 2 Indicates the damage coefficient of the tower

[0148] Main shaft fatigue damage objective function: Minimize the average value of fatigue damage of all wind turbine main shafts

[0149]

[0150] Tower fatigue damage objective function: Minimize the average fatigue damage of all wind turbine towers

[0151]

[0152] Wind turbine power allocation optimization objective function: Minimize the variance of active power allocation of all wind turbines

[0153]

[0154] The goal is to minimize the fatigue damage of the wind turbine and the variance of power distribution. The comprehensive objective function is as follows:

[0155]

[0156] Where N is the number of fans, is the fatigue damage of the main shaft of the i-th fan, is the fatigue damage of the tower of the i-th wind turbine, is the output mechanical power of the i-th fan, is the average output mechanical power of the i-th wind turbine, α, β, γ are weight coefficients;

[0157] S4.2. Constraints:

[0158]

[0159] in, is the output mechanical power of the i-th wind turbine tower, and its value is a fixed value;

[0160] S4.3, Measurement Noise Robust Optimization:

[0161] For each wind speed and power reference value, a random noise term is added:

[0162]

[0163] in, is the measured wind speed of the ith wind turbine in time period w, is the actual wind speed of the ith wind turbine in time period w, is the power reference value measured by the ith wind turbine in the w time period, is the actual power reference value, and is the noise term of the wind speed and power reference value of the i-th wind turbine, satisfying:

[0164]

[0165] S4.4. Communication delay optimization modeling:

[0166] In practical applications, there is a delay in the data transmission process, that is, at each moment, the data used by the present invention is not collected in real time, but comes from the measurement result of the previous moment. In order to simulate the delay of sensor data transmission in practical applications, the present invention introduces a communication delay model to simulate actual data transmission, and at the same time sets the maximum delay time of the system to 10 seconds in combination with actual engineering applications;

[0167] For the data at each moment, consider the data of the previous normal communication moment:

[0168]

[0169] in, is the wind speed data of the i-th wind turbine at the previous moment, is the power reference value data of the i-th wind turbine at the current moment, It is the power reference value data of the previous moment;

[0170] S4.5. Objective function of the robust optimization model:

[0171]

[0172] Constraints:

[0173]

[0174] In step S5: the optimized results are compared and analyzed with the ideal operation without noise and delay. Through this comparison, the actual application effect of the model is evaluated, and the effectiveness and feasibility of the optimization model under real conditions are further confirmed.

[0175] Based on the solution of the model, the weight coefficients α = 0.3, β = 0.3, γ = 0.4 are obtained, so the comprehensive objective function is as follows:

[0176]

[0177] Next, the optimization results of the fan output power and communication delay time of the noise and communication delay models are compared with the ideal situation without noise and delay.

[0178] (1) Variance before and after delay suppression optimization:

[0179] Figure 2(a) shows the delay suppression - variance before and after optimization. The horizontal axis represents the time step, ranging from 50 to 70. The vertical axis represents the variance of the reference power. The blue curve (before optimization) represents the variance of the reference power before optimization at each time step. The red curve (after optimization) represents the variance of the reference power after optimization at each time step. From the fluctuation point of view, the standard deviation before optimization is almost zero between time steps 50 and 70, indicating that the power distribution before optimization is very stable. The standard deviation after optimization fluctuates significantly between time steps 50 and 70. From the trend analysis point of view, the standard deviation of power distribution before optimization has no obvious trend and remains almost zero, showing a high degree of uniformity. Although the standard deviation fluctuates after optimization, the overall level is high, indicating that the optimized power distribution is more concentrated on certain wind turbines.

[0180] Figure 2(b) shows the power distribution at time step 68. The horizontal axis represents the wind turbine index, ranging from 1 to 100, and the vertical axis represents the power. At time step 68, the power distribution diagram shows the power output of each wind turbine. The power output of most wind turbines is concentrated between 4.0 and 4.2 MW, indicating that the power distribution is relatively uniform. Several wind turbines (such as those with indices around 10, 30, and 40) show obvious power peaks, reaching more than 4.4 MW. These wind turbines may have been assigned higher power outputs based on the results of the optimization algorithm to meet the overall optimization goal.

[0181] Based on the above analysis of the data in the figure, although the standard deviation of the optimized power distribution is higher, the fluctuation is smaller and the distribution is more stable. This optimization helps to improve the overall operating efficiency and reliability of the wind farm. At the same time, by concentrating power distribution on more efficient wind turbines, it maximizes the grid dispatching needs and reduces the fatigue damage of wind turbines.

[0182] (2) Optimization time, reference power variance before and after optimization, and reduction ratio:

[0183] Figure 3(a) shows the time spent on optimization at each time step (50 to 70). The optimization time remains between 3.5 seconds and 4.5 seconds in most time steps, with some fluctuations. These fluctuations may be caused by the data characteristics at different time steps or the randomness of the optimization algorithm.

[0184] Figure 3(b) shows the reference power variance before and after optimization. The reference power variance before optimization (blue circle) remains almost unchanged, indicating that the initial reference power allocation does not differ much between time steps. The reference power variance after optimization (red cross) changes significantly at each time step, and the variance value is significantly higher than before optimization.

[0185] Figure 3(c) shows the reduction ratio. It can be seen that the variance reduction (blue circle) is negative in most time steps, which means that the variance of the reference power increases after optimization. The reduction ratios of shaft fatigue damage reduction (yellow cross) and tower fatigue damage reduction (orange square) are both positive, and the tower fatigue damage reduction ratio is significantly higher than the shaft fatigue damage reduction ratio, indicating that the optimization is mainly focused on reducing the fatigue damage of the tower.

[0186] (3) Variance before and after noise suppression optimization:

[0187] Figure 4(a) shows that the power variance before optimization has no obvious trend, showing great instability. This shows that before optimization, the power distribution of each wind turbine is relatively uniform, but there are certain fluctuations. Figure 4(b) shows that although the power variance after optimization also fluctuates, the overall level is high and the fluctuation amplitude is relatively small, showing a certain stability. This shows that the optimized power distribution is more concentrated on certain wind turbines, resulting in an increase in power variance. By comparing the two figures, it can be seen that the power variance after optimization is much higher than that before optimization as a whole, because the optimized power distribution is more concentrated on certain wind turbines, resulting in an increase in variance. The power distribution variance before optimization fluctuates greatly, while the fluctuation after optimization is relatively small, indicating that the optimized distribution is more stable. The goal of optimization is to reduce certain specific indicators (such as fatigue damage), rather than simply reducing the variance. Therefore, although the variance of the optimized power distribution is higher, the fluctuation amplitude is smaller, indicating that the optimized power distribution is more stable.

[0188] (4) Reduction ratio of shaft fatigue damage and tower fatigue damage:

[0189] Figure 5 The reduction ratio of shaft fatigue damage (blue bar) and tower fatigue damage (orange bar) in each time step is shown. It can be seen that the reduction ratio of tower fatigue damage is significantly higher than that of shaft fatigue damage in most time steps, further verifying that the optimization is mainly focused on reducing tower fatigue damage. In some time steps (such as steps 52, 54, and 66), the reduction ratio of tower fatigue damage is significantly higher than that of other time steps, indicating that the optimization effect is more significant in these time steps.

[0190] In the multi-wind turbine scenario of an actual wind farm, the present invention considers the noise interference of sensor measurement data and the impact of communication delay on data transmission. In order to improve the robustness of the model, disturbances are introduced into the measurement data to simulate noise, and when data is delayed or missing, model predictive control technology is used to use historical data to predict power allocation, thereby ensuring the continuity of optimal scheduling among multiple wind turbines. In the optimization framework based on the NSGA-II algorithm, multi-objective optimization problems with noise and delay are solved to ensure that the power allocation scheme in the multi-wind turbine scenario can effectively cope with external interference, maximize power generation efficiency and reduce equipment losses.

[0191] It is to be understood that the present invention is described by some embodiments, and it is known to those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the scope of protection of the present invention.

Claims

1. A method for optimizing and solving active power in a multi-wind turbine scenario, characterized in that: The following steps are involved: S1. Collect various index data of wind speed, wind direction and wind rotor radius of wind turbines of N wind turbines; S2. Establish a wind energy conversion model, a main shaft torque and tower thrust model, and a material fatigue damage calculation model based on various wind turbine indicator data; S3. By analyzing the collected data of various indicators of the wind turbine, a measurement noise model is established to filter out noise interference; considering the communication delay in the actual operating environment, a communication delay model is constructed to analyze its potential impact on system performance and ensure the robustness of the model; S4. Establish a robust optimization model for the noise model and communication delay model based on the NSGA-II algorithm; S5. Compare the optimization results of the fan output power and communication delay time of the model with noise and communication delay with the ideal case without noise and delay.

2. The method for optimizing and solving active power in a multi-wind turbine scenario according to claim 1 is characterized in that: Building a wind energy conversion model includes the following steps: S2.

1. Establish the wind energy conversion model of the i-th wind turbine through the wind energy capture formula: Among them, P i is the wind power of the i-th wind turbine, ρ is the air density, is the swept area of ​​the wind wheel, r i is the radius of the rotor of the i-th wind turbine, V i is the wind speed of the i-th wind turbine; Mechanical power output of the i-th fan The relationship between the captured wind energy and the power factor C1 is expressed as follows: Among them, the power coefficient C1 is a nonlinear coefficient; when the wind speed is low, C1 increases with the wind speed; when the wind speed exceeds the rated wind speed, the wind turbine power reaches saturation and C1 decreases accordingly; According to the law of conservation of energy, the output power of wind energy is related to the main shaft torque and angular velocity as follows: in, The i-th fan is the main shaft torque, ω i is the angular velocity of the i-th wind turbine rotor; calculate the main shaft torque: Tower thrust Affected by wind speed and aerodynamic forces, proportional to the square of wind speed: Among them, C2 is the thrust coefficient, which indicates the force of the wind turbine at different wind speeds; is the swept area of ​​the wind wheel, r i is the radius of the rotor of the i-th wind turbine, V i is the wind speed of the i-th wind turbine; S2.

2. Establish the i-th wind turbine main shaft torque and tower thrust model: Model the main shaft torque and tower thrust of the wind turbine; establish the functional relationship between the main shaft torque and wind speed and load by analyzing the mechanical characteristics of the wind turbine under different working conditions; The model is divided into three areas according to wind speed: low wind speed, rated wind speed, and high wind speed; the low wind speed range is V<11.2m / s. In this range, the wind speed output power is low, the mechanical parts are subjected to small forces, and fatigue damage comes from high cycle times; the rated wind speed range is V≈11.2m / s. At the rated wind speed, the wind speed output power reaches the maximum, and fatigue damage accumulates evenly; the high wind speed range is V>11.2m / s. At high wind speeds, the wind turbine performs power reduction, the main shaft torque is reduced, but the tower thrust increases, and fatigue damage accumulation accelerates; The main shaft torque and tower thrust model is expressed by a piecewise function: Calculation of the main shaft torque: Tower thrust calculation: in, is the main shaft torque of the i-th fan, is the tower thrust of the i-th wind turbine, P i is the wind power of the i-th wind turbine, C1 is the power coefficient, ρ is the air density, is the swept area of ​​the wind wheel, V i is the wind speed of the i-th fan, ω i is the angular velocity of the i-th wind turbine rotor; S2.

3. Calculation of fan material fatigue damage: Establish material fatigue damage model: Use stress-strain relationship and rain flow counting method to analyze the fatigue life of wind turbine structure and evaluate the damage and failure mode of materials during long-term operation; Using the Palmgren-Miner linear cumulative damage theory: Where D is the cumulative fatigue damage, n i is the stress level The number of cycles under i is the material at stress level F i The maximum number of cycles under .

3. The method for optimizing and solving active power in a multi-wind turbine scenario according to claim 2 is characterized in that: S3 specifically includes the following steps: S3.

1. Establish measurement noise model: The measured value y at time t is set to satisfy the following formula: y t =x t +N(0,(σx t ) 2 ) (9) Where σ is the standard deviation proportionality factor of the noise, x t is the observed value at the current moment; S3.

2. Establish a communication delay model: Setting the delay τ t In [0,τ max ]The interval follows a uniform distribution: t t ~U(0,τ max (10) The observation value that the controller can obtain at time t is: Among them, x t is the observed value at the current moment, is the observed value after the delay.

4. According to the method for optimizing and solving active power in a multi-wind turbine scenario according to claim 3, S4 specifically comprises the following steps: S4.

1. Define the objective function as minimizing the average fatigue damage of each component and add the optimization objective of wind turbine power distribution; Spindle fatigue damage The formula derivation is as follows: Fatigue damage of the main shaft of the i-th fan In a short setting, fatigue damage is proportional to the square of the spindle torque: Where k1 is a coefficient related to material properties and stress cycle characteristics. Since the material and structural parameters are consistent, k1 is the same for all fans. is the main shaft torque of the i-th fan; Tower fatigue damage of the i-th wind turbine Set the fatigue damage to be proportional to the square of the tower thrust: Among them, k2 represents the damage coefficient corresponding to the tower Main shaft fatigue damage objective function: Minimize the average value of fatigue damage of all wind turbine main shafts Tower fatigue damage objective function: Minimize the average fatigue damage of all wind turbine towers Wind turbine power allocation optimization objective function: Minimize the variance of active power allocation of all wind turbines The goal is to minimize the fatigue damage of the wind turbine and the variance of power distribution. The comprehensive objective function is as follows: Where N is the number of fans, is the fatigue damage of the main shaft of the i-th fan, is the fatigue damage of the tower of the i-th wind turbine, is the output mechanical power of the i-th fan, is the average output mechanical power of the i-th wind turbine, α, β, γ are weight coefficients; S4.

2. Constraints: in, is the output mechanical power of the i-th wind turbine tower, and its value is a fixed value; S4.3, Measurement Noise Robust Optimization: For each wind speed and power reference value, a random noise term is added: in, is the measured wind speed of the ith wind turbine in time period w, is the actual wind speed of the ith wind turbine in time period w, is the power reference value measured by the ith wind turbine in the w time period, is the actual power reference value, and is the noise term of the wind speed and power reference value of the i-th wind turbine, satisfying: S4.

4. Communication delay optimization modeling: A communication delay model is introduced to simulate actual data transmission; For the data at each moment, consider the data of the previous normal communication moment: in, is the wind speed data of the i-th wind turbine at the previous moment, is the power reference value data of the i-th wind turbine at the current moment, It is the power reference value data of the previous moment; S4.

5. Objective function of the robust optimization model: Constraints:

5. The method for optimizing and solving active power in a multi-wind turbine scenario according to claim 4 is characterized in that: S5 specifically includes the following steps: Based on the solution of the robust optimization model, the weight coefficients α = 0.3, β = 0.3, γ = 0.4 are obtained, so the comprehensive objective function is as follows:

Images