Wind turbine operation characteristic considering wind-solar coordination control method in cold region

Abstract

The application relates to the technical field of energy scheduling, in particular to a cold region wind-solar coordinated control method considering the operation characteristics of wind turbine generators, which comprises the following steps: acquiring the operation parameters of each wind turbine generator and the meteorological parameters of the environment in real time, extracting multi-dimensional features reflecting the abnormal aerodynamic performance and mechanical load of the wind turbine generator, and identifying icing by using a classification model; quantifying the available output margin of each wind turbine generator, judging whether the wind turbine generator meets the trigger condition of state switching to execute a state switching strategy, and determining the upper limit of the available output of each wind turbine generator; collecting the irradiance of a photovoltaic array, the battery temperature and the snow accumulation parameters of the environment in real time, and determining the upper limit of the available output of the photovoltaic array; constructing an objective function and a constraint condition, and solving the optimal output of each wind turbine generator and the photovoltaic array to execute wind-solar coordinated control. The application improves the efficient utilization rate of cold region wind-solar energy and the safety and coordinated control ability of grid-connected output.

Description

Technical Field

[0001] This application relates to the field of energy dispatching technology, specifically to a method for coordinated control of wind and solar power in cold regions that takes into account the operating characteristics of wind turbine units. Background Technology

[0002] Cold regions possess abundant wind and solar energy resources. However, the extreme microclimate conditions in these regions have a complex impact on the actual operating characteristics of wind and solar power equipment. For wind turbines, the high air density caused by low temperatures drastically increases rotor thrust, forcing the turbines to actively adjust pitch to limit power and prevent overload. Simultaneously, blades are highly susceptible to icing, severely damaging their aerodynamic shape and causing mechanical imbalances, leading to a decrease in actual energy capture efficiency. For photovoltaic arrays, continuous snowfall can create shading on the photovoltaic panel surface, blocking solar radiation transmission. However, extremely low temperatures can effectively suppress carrier recombination within the photovoltaic cells, significantly improving photoelectric conversion efficiency, resulting in complex bidirectional coupling fluctuations in photovoltaic output.

[0003] Currently, conventional new energy dispatching methods all assume that each energy device participates in regulation according to its rated capacity, without considering the dynamic constraints of extreme weather in cold regions on the actual available output limit of the equipment. This results in dispatching commands often deviating from the actual output limit of the equipment. If the command is too high, it can easily cause serious mechanical damage to the icing overloaded units. If the command is too conservative, it will cause a large amount of wind and solar curtailment, which seriously affects the stability of high-precision dispatching and tracking and wind and solar coordinated control under the premise of ensuring the safe operation of equipment in cold regions. Summary of the Invention

[0004] To address the aforementioned technical issues, a coordinated control method for wind and solar power in cold regions, taking into account the operating characteristics of wind turbine units, is provided.

[0005] The solution to the technical problem presented in this application is to provide a coordinated control method for wind and solar power in cold regions that takes into account the operating characteristics of wind turbine units, including the following steps:

[0006] Real-time acquisition of operating parameters of each wind turbine and meteorological parameters of the surrounding environment; extraction of multi-dimensional features reflecting aerodynamic performance and mechanical load anomalies of the wind turbine; and use of classification model for icing identification to obtain the real-time icing status of each wind turbine.

[0007] The performance degradation of wind turbines under the influence of icing and density coupling with cold air is assessed, the available output margin of each wind turbine is quantified, and the wind turbines are judged to meet the trigger conditions for state switching based on the available output margin and the icing status. The state switching strategy is then executed. The target power is allocated to each wind turbine based on the final operating state to which the wind turbine is to be switched, combined with the grid dispatch plan. After the state switching is triggered, a transition period is preset to control the output limit of the wind turbines to smoothly transition from the output status before the trigger to the target power, so as to determine the available output limit of each wind turbine in real time.

[0008] The irradiance, battery temperature, and snow cover parameters of the photovoltaic array are collected in real time. The compensation effect of low temperature on photoelectric efficiency and the optical blocking effect of snow on radiation transmission are evaluated to reduce the photovoltaic power generation capacity and determine the upper limit of the usable output of the photovoltaic array.

[0009] An objective function is constructed to minimize the grid dispatch tracking deviation, energy storage output, and wind and solar curtailment. Constraints are established with the upper limit of available output as the boundary. The objective function is solved under the constraints to calculate the optimal output of each wind turbine and photovoltaic array, so as to perform wind and solar coordinated control.

[0010] Preferably, the operating parameters include active power, rotor speed, blade pitch angle, and triaxial vibration acceleration of the nacelle; the meteorological parameters include wind speed, air temperature, and air density; the extraction of multidimensional features reflecting aerodynamic performance and mechanical load anomalies of the wind turbine specifically includes extracting power deviation features, vibration deviation features, speed deviation features, and blade pitch angle features, and the extraction process is as follows:

[0011] Construct a fitting function of theoretical output power with respect to wind speed, substitute the real-time wind speed of each wind turbine into the fitting function, and output the theoretical output power corresponding to the wind speed; calculate the difference between the real-time active power and the theoretical output power of each wind turbine as the power deviation feature.

[0012] Using each moment as the endpoint, a time period of a preset length is extracted and defined as a local time window. The average rotational speed of the wind turbine at all moments within the local time window is calculated as the average rotational speed. The fundamental frequency of the wind turbine rotation is calculated using this average rotational speed as the 1P frequency, and three times this average rotational speed is calculated as the 3P frequency. The 1P frequency and the 3P frequency are collectively referred to as specific frequencies. Frequency domain analysis is performed on the triaxial acceleration at all moments within the local time window, and the corresponding spectrum diagrams for each axis are extracted. The difference between the amplitude at each specific frequency in the spectrum diagram of each axis and the corresponding preset reference amplitude is calculated as the amplitude deviation. The amplitude deviation of the three axes at all specific frequencies is used as the vibration deviation characteristic.

[0013] The difference between the real-time wind turbine speed and the preset reference speed is calculated as a speed deviation characteristic;

[0014] Each moment and several consecutive moments preceding it are defined as a local time period. The rate of change of the pitch angle between two adjacent moments within the local time period is calculated, and the average of the absolute values ​​of all rates of change within the local time period is used as the pitch angle characteristic.

[0015] Preferably, the step of using a classification model to identify icing to obtain the real-time icing status of each wind turbine includes:

[0016] The real-time power deviation characteristics, vibration deviation characteristics, speed deviation characteristics, and pitch angle characteristics of each wind turbine are combined into a multi-dimensional feature vector, which is then input into a pre-trained ensemble learning model for icing identification, and the posterior probability of each wind turbine belonging to each icing level is output.

[0017] The icing level corresponding to the maximum posterior probability is selected as the real-time icing status of the wind turbine; the icing level includes no icing, light icing, and severe icing.

[0018] Preferably, the quantification of the available output margin of each wind turbine includes:

[0019] Based on the real-time icing status of each wind turbine, a power reduction factor is pre-set for it; using the posterior probability of the icing level, this power reduction factor is corrected, resulting in the corrected power reduction factor. The calculation formula is: In the formula, The output reduction factor before correction. The preset first adjustment value, To preset the second adjustment value, Let be the confidence coefficient, where , Let be the posterior probability that the wind turbine unit is free from icing. This represents the posterior probability that the wind turbine is subject to light icing. Preset weights;

[0020] Calculate the ratio of the real-time air density of each wind turbine to the preset standard air density, and use this ratio as the density ratio. Based on this density ratio, determine the density correction coefficient of each wind turbine in real time using a pre-built piecewise function.

[0021] Obtain the rated power of each wind turbine, and multiply it by the corrected output reduction factor, density correction factor, and preset mechanical attenuation factor to obtain the usable output margin of each wind turbine.

[0022] Preferably, determining whether the wind turbine meets the triggering conditions for state switching in order to execute the state switching strategy includes:

[0023] The ratio of the available output margin to the rated power of each wind turbine is calculated in real time and used as the proportion of available output.

[0024] When each wind turbine is in normal operation, if there is no ice-free state for several consecutive moments and the available output ratio is less than the preset first threshold, the normal operation state will be switched to the derating operation state; otherwise, the normal operation state will be maintained.

[0025] When each wind turbine is in derating operation mode, if the icing status is non-icing for multiple consecutive moments and the available output ratio is greater than the preset second threshold, the derating operation mode is switched to normal operation mode; if the icing status is severe icing for multiple consecutive moments, or the ambient temperature is lower than the preset value, the derating operation mode is switched to shutdown recovery mode; otherwise, the derating operation mode is maintained; wherein, the preset second threshold is greater than the preset first threshold.

[0026] When each wind turbine is in the shutdown recovery state, if it receives a message that the de-icing of the turbine has been completed and the ambient temperature has risen above the preset value, it will switch from the shutdown recovery state to the derated operation state; otherwise, it will remain in the shutdown recovery state.

[0027] Preferably, the allocation of target power to each wind turbine includes:

[0028] The ratio between the available output ratio of each wind turbine and the sum of the available output ratios of all wind turbines is calculated in real time and used as the allocation weight.

[0029] When each wind turbine needs to switch to normal operation, the target power allocated to it is... for: In the formula, To assign weights; To receive the total power deficit from the grid's dispatch plan, This represents the available output margin. It is a minimum value function;

[0030] When each wind turbine needs to switch to derating operation, the target power allocated to it is... for: In the formula, This is a safety margin factor preset for low power.

[0031] When each wind turbine needs to switch to a shutdown recovery state, the target power allocated to it is... for: .

[0032] Preferably, determining the upper limit of available output for each wind turbine includes:

[0033] For any given moment, if it is before the state transition is triggered, its available output limit is the available output margin.

[0034] If any of the aforementioned moments falls within the transition period, a linear interpolation algorithm is used to interpolate between the available output margin at the moment before the state switch is triggered and the target power allocated in the new state, and the interpolation result is used as the upper limit of available output.

[0035] If any of the stated moments exceeds the end of the transition period, then the upper limit of its available output is the target power.

[0036] Preferably, the snow accumulation parameters include the snow thickness and snow density on the horizontal surface; determining the upper limit of the available output of the photovoltaic array specifically includes:

[0037] Real-time temperature efficiency correction factor for photovoltaic array The specific formula is as follows: In the formula, The preset power temperature coefficient, The temperature difference is the difference between the real-time cell temperature of the photovoltaic array and the preset standard temperature.

[0038] The ratio of real-time snow density to preset new snow density is used as the density conversion value; the product of real-time snow thickness, cosine of photovoltaic array installation tilt angle, and density conversion value is calculated as the real-time equivalent snow thickness; the product of this equivalent snow thickness and preset snow equivalent attenuation coefficient is negatively mapped as the real-time snow attenuation factor.

[0039] Calculate the ratio between the real-time irradiance of the photovoltaic array and the preset standard irradiance, and use it as the irradiance ratio; obtain the rated power of the photovoltaic array, and multiply it by the temperature efficiency correction coefficient, the snow accumulation attenuation factor, and the irradiance ratio, and use it as the upper limit of the available output of the photovoltaic array.

[0040] Preferably, the process of constructing the objective function is as follows:

[0041] objective function for: In the formula, To receive the total power deficit from the grid's dispatch plan, The rated power of the energy storage system. Total installed capacity for The actual total output at any given moment This indicates the total power of wind and solar power curtailment. , as well as This indicates the preset weighting coefficient. ;

[0042] in, , Indicates that the photovoltaic array is in The optimal output needs to be determined at all times. For energy storage systems in The optimal output needs to be determined at all times. For the first Each wind turbine unit The optimal output is to be determined at any given time, where N is the number of all wind turbine units;

[0043] , For photovoltaic arrays in The maximum available output at any given time. For the first Each wind turbine unit The maximum available output at any given time. This represents the function that takes the maximum value. This represents the function that takes the minimum value.

[0044] Preferably, the constraint condition is:

[0045] Wind turbine output constraints: If the first If a wind turbine is in a shutdown recovery state, then ;

[0046] Photovoltaic array output constraints: ;

[0047] Energy storage system operating constraints: ; ,in, This refers to the minimum allowable charging power limit for energy storage systems. This refers to the maximum allowable discharge power limit for energy storage systems. This represents the minimum state of charge of the battery in the energy storage system. This represents the maximum battery state of charge of the energy storage system. For energy storage systems in The state of charge at any given moment.

[0048] This application has at least the following beneficial effects:

[0049] This application identifies the icing state of wind turbines by extracting multidimensional features including power deviation, vibration deviation, speed deviation, and pitch angle. It can accurately capture latent features caused by icing from deep operational data, significantly improving the accuracy and robustness of icing identification. Furthermore, by quantifying the available output margin of each wind turbine, it creatively couples the aerodynamic energy capture attenuation caused by icing with the control limits induced by high-altitude air density, accurately quantifying the static theoretical power generation capacity of the turbines under harsh operating conditions. This provides an extremely reliable basis for subsequent triggering state switching strategies and establishing safe control boundaries. Based on fundamental data, the system determines whether wind turbines meet the trigger conditions for state switching and executes the state switching strategy. It utilizes available output margin and icing status as the basis for state switching, dynamically matching a tiered control and protection mode according to the actual performance degradation levels of the turbines. This allows the turbines to safely reduce load or shut down under extreme conditions and smoothly load when the environment improves, effectively avoiding the potential mechanical failures caused by forced full-load operation under damaged conditions. By determining the final operating state the wind turbines will switch to, and combining this with the grid dispatch plan, target power is allocated to each wind turbine, ensuring the macro-dispatch needs of the power grid are met. This system can precisely allocate power proportionally based on the actual operating capacity of each wind turbine, ensuring that turbines in regular operation undertake the main power generation tasks, turbines operating at reduced capacity have their output limited to mitigate risks, and turbines shut down can be safely decommissioned. This achieves differentiated optimal allocation of power distribution and reasonable load sharing within the wind farm. Furthermore, during the transition period after a state switch is triggered, linear interpolation smoothly transitions the upper limit of output from the pre-switch value to the target power, completely eliminating output jumps caused by sudden command changes when turbines cross operating state boundaries. This effectively prevents severe torque shocks to the wind turbine drivetrain components caused by abrupt power changes. The upper limit of the available output of the generator set is accurately calculated to determine the absolute power limit that the unit can safely generate under harsh operating conditions. This fundamentally eliminates mechanical overload damage caused by dispatching instructions blindly exceeding the actual load capacity of the equipment. It provides a dynamic safety boundary with transition smoothness constraints for subsequent optimized dispatching, greatly ensuring the continuity and stability of the active power output of the entire grid. By separating the optical blocking loss caused by snow accumulation and the photoelectric conversion gain brought by extreme cold through physical mechanisms and coupling them with dual factors, the upper limit of the available output of the photovoltaic array is determined, and the absolute power limit that the photovoltaic array can actually generate under complex and harsh weather coupling is accurately quantified.By constructing objective functions and constraints, the optimal output of each wind turbine and photovoltaic array is calculated to execute coordinated wind and solar control. The precisely derived upper limit of actual wind and solar output is used as a hard safety boundary. Under the premise of absolutely ensuring that the underlying equipment is not overloaded, the charging and discharging of the energy storage system is coordinated, perfectly realizing the maximization of wind and solar renewable energy absorption, minimization of energy storage losses, and high-precision tracking of grid commands. This improves the efficient utilization rate of wind and solar energy in cold regions and the safe and coordinated control capability of grid-connected output. Attached Figure Description

[0050] The following section provides a more detailed description of the cold-region wind-solar coordinated control method of this application, which takes into account the operating characteristics of wind turbine units.

[0051] Figure 1 A flowchart illustrating the steps of a cold-region wind-solar coordinated control method considering the operating characteristics of wind turbine units, as provided in an embodiment of this application. Detailed Implementation

[0052] The following, in conjunction with the accompanying drawings and embodiments, provides a more detailed description of the cold-region wind-solar coordinated control method that takes into account the operating characteristics of wind turbine units, as proposed in this application.

[0053] Please see Figure 1 The diagram illustrates a flowchart of a cold-region wind-solar coordinated control method considering the operating characteristics of wind turbine generators, according to an embodiment of this application. The method includes the following steps:

[0054] Step 1: Real-time acquisition of operating parameters of each wind turbine and meteorological parameters of the surrounding environment; real-time acquisition of irradiance, battery temperature and snow accumulation parameters of the photovoltaic array.

[0055] Step 2: Extract multi-dimensional features that reflect the aerodynamic performance and mechanical load anomalies of the wind turbine units, and use a classification model to identify icing in order to obtain the real-time icing status of each wind turbine unit.

[0056] Step 3: Assess the performance degradation of wind turbines under the influence of icing and cold air density coupling, quantify the available output margin of each wind turbine, and determine whether the wind turbine meets the triggering conditions for state switching based on the available output margin and icing status. Then, execute the state switching strategy. Based on the final operating state that the wind turbine is to switch to, and in conjunction with the grid dispatch plan, allocate target power to each wind turbine. After the state switching is triggered, a transition period is preset to control the output limit of the wind turbine to smoothly transition from the output status before the trigger to the target power, so as to determine the available output limit of each wind turbine in real time.

[0057] Step 4: Real-time collection of irradiance, battery temperature and snow accumulation parameters of the photovoltaic array, evaluation of the compensation effect of low battery temperature on photoelectric efficiency and the optical blocking effect of snow on radiation transmission to reduce the photovoltaic power generation capacity and determine the upper limit of the available output of the photovoltaic array.

[0058] Step 5: Construct an objective function that minimizes grid dispatch tracking deviation, energy storage output, and wind and solar curtailment. Establish constraints with the upper limit of available output as the boundary. Solve the objective function under the constraints to calculate the optimal output of each wind turbine and photovoltaic array, so as to perform wind and solar coordinated control.

[0059] In cold environments, wind turbine blades are prone to icing, causing power output and operating characteristics to deviate significantly from rated conditions. Simultaneously, photovoltaic modules also face the effects of snow cover and low temperatures, resulting in a substantial decrease in actual output. Conventional dispatching fails to detect the actual capacity reduction caused by these cold-region characteristics, easily leading to dispatch imbalances and excessive allocation commands. To accurately assess the actual available output of power generation equipment and achieve differentiated collaborative control, the operating parameters of the wind turbines and photovoltaic arrays are first collected in step 1, specifically:

[0060] Step 101: Real-time acquisition of operating parameters of each wind turbine, including active power, rotor speed, blade pitch angle, and triaxial vibration acceleration of the nacelle; and real-time acquisition of wind speed, temperature and air pressure of the environment where each wind turbine is located, and real-time deduction of air density based on the real-time temperature and air pressure.

[0061] The specific formula is as follows: ; In the formula, For real-time air density, For real-time air pressure, Standard atmospheric pressure This represents the real-time temperature; the standard atmospheric pressure is 101.325 kPa.

[0062] Step 102: Real-time acquisition of the irradiance of the photovoltaic array, the battery temperature, and the snow accumulation parameters of the surrounding environment; the snow accumulation parameters include the snow thickness and snow density on the horizontal surface.

[0063] It should be noted that snow thickness and density are obtained through weather station detection.

[0064] The data acquisition time interval is 1 second. As another implementation method, the implementer can set it according to the actual situation. The acquired data is time-aligned. The time alignment is a well-known technology and will not be described in detail here. After removing the data of wind turbines and photovoltaic arrays during shutdown or fault periods, the Laida criterion is used to detect outliers in the acquired data to remove abnormal data. The missing values ​​are filled using a linear interpolation algorithm. The Laida criterion and the linear interpolation algorithm are well-known technologies and will not be described in detail here.

[0065] Furthermore, in cold environments, wind turbine blades are prone to varying degrees of icing. Icing disrupts the aerodynamic shape of the blades, increases their mass, and leads to a significant decrease in aerodynamic efficiency and distortion of mechanical loads. Failure to accurately monitor the icing status of wind turbines in real time can result in severely distorted power output assessments at the wind farm level, potentially causing power imbalances in the grid, and exacerbating mechanical fatigue and even leading to turbine shutdowns and propeller failures. Therefore, real-time assessment of the icing status of each wind turbine is essential. When wind turbines experience icing, their operating parameters exhibit an abnormal coupling characteristic of high wind speed, low engine speed, low power output, and high vibration.

[0066] In step 2, the specific process is as follows:

[0067] First, icing reduces blade lift and significantly increases drag, resulting in a significantly lower output power of the wind turbine at the same wind speed compared to the normal, ic-free operating conditions. Furthermore, the power loss caused by icing intensifies as wind speed increases.

[0068] Step 201: Construct a fitting function of theoretical output power with respect to wind speed, substitute the real-time wind speed of each wind turbine into the fitting function, and output the theoretical output power corresponding to the wind speed; calculate the difference between the real-time active power and the theoretical output power of each wind turbine as the power deviation feature.

[0069] The process of constructing the fitting function of theoretical output power with respect to wind speed is as follows:

[0070] Active power and wind speed data were collected from historical periods when the air density was within a predetermined neighborhood of the standard air density and the wind turbines were in a non-icing state. These were used as power-wind speed sample data. Using active power as the dependent variable and wind speed as the independent variable, a five-parameter Logistic function was selected for fitting. An optimization algorithm was used to fit the power-wind speed sample data, and the optimal parameters of the five-parameter Logistic function were obtained, thus yielding the fitting function of theoretical output power with respect to wind speed. The Logistic function... The five parameters include the lower asymptote, upper asymptote, inflection point position, slope parameter, and asymmetry factor. The lower asymptote is set to range from -20 to 0; the upper asymptote is set to range from rated power to 1.05 times the rated power; the inflection point position is set to range from 0.5 to 2.0 times the rated wind speed; the slope parameter is set to range from 2 to 8; and the asymmetry factor is set to range from 1 to 1000. Within these five ranges, the optimal five parameters are obtained using an optimization algorithm. The Logistic function and the optimization algorithm are well-known techniques and will not be described in detail here.

[0071] In this embodiment, the standard air density is The preset neighborhood range is to As another implementation method, the implementer can set it according to the actual situation.

[0072] Secondly, the unevenness of blade icing leads to significant mass and aerodynamic imbalances. This imbalance effect is directly reflected in the vibration response of the wind turbine, manifested as an abnormal increase in the amplitude of specific frequency components. Specifically, icing causes uneven distribution of ice on the blade surface, resulting in asymmetrical changes in the blade's mass and airfoil. This causes periodic unbalanced loads during rotor rotation, forming distinct vibration characteristics at the first harmonic (1P) and third harmonic (3P) of the rotor's rotational frequency. The increase in amplitude at the 1P frequency mainly reflects mass imbalance; the uneven distribution of icing mass causes the rotor's center of mass to deviate from the center of rotation, generating a centrifugal force pulse with each rotation. The increase in amplitude at the 3P frequency mainly reflects aerodynamic imbalance; the different degrees of icing on the three blades result in inconsistent aerodynamic performance, and the aerodynamic impact intensity generated when each blade passes through the tower varies, forming a periodic fluctuation three times per rotation. Based on this analysis, the amplitudes of triaxial acceleration at frequencies of 1P and 3P are extracted by time-frequency domain transformation, and the increment of the amplitude relative to the un-iced state is evaluated. This quantitatively characterizes the degree of imbalance caused by icing and provides a vibration index reflecting the mechanical response characteristics for icing state identification.

[0073] Step 202: Using each moment as the endpoint, extract a time period of a preset length, defined as a local time window; calculate the average rotational speed of the wind turbine at all moments within the local time window, using this average rotational speed to calculate the fundamental frequency of the wind turbine rotation as the 1P frequency, and multiply it by three to obtain the 3P frequency; collectively refer to the 1P and 3P frequencies as specific frequencies; perform frequency domain analysis on the triaxial acceleration at all moments within the local time window, extracting the corresponding spectrum diagrams for each axis; calculate the difference between the amplitude at each specific frequency in the spectrum diagram of each axis and the corresponding preset reference amplitude, as the amplitude deviation; use the amplitude deviation of the three axes at all specific frequencies as vibration deviation characteristics;

[0074] Specifically, the duration of the local time window is 10 minutes. As other implementation methods, implementers can set it according to the actual situation. Secondly, the three-axis acceleration refers to the acceleration of the cabin in the X-axis, Y-axis and Z-axis. The Fast Fourier Transform is used to perform frequency domain analysis on the acceleration of each axis to obtain the spectrum diagram corresponding to each axis. The Fast Fourier Transform is a well-known technology and will not be described in detail here.

[0075] Secondly, the process of obtaining the preset reference amplitude is as follows:

[0076] The wind turbine rotor speed range is divided into multiple speed ranges. The speed range to which the average speed belongs is extracted. The triaxial acceleration of the wind turbine in the speed range and in the non-icing state during the historical period is collected in advance. Frequency domain analysis is performed on each of them to obtain the historical spectrum of each axis. The amplitude of each axis at each specific frequency is extracted from the historical spectrum of each axis as the reference amplitude.

[0077] In addition, the wind turbine speed operating range is based on the cut-in speed of the wind turbine as the lower limit and the rated speed as the upper limit. The wind turbine speed operating range is divided equally according to the preset equal interval step size, so as to obtain multiple speed ranges. Among them, the cut-in speed refers to the lowest wind turbine speed corresponding to the start of grid-connected power generation of the wind turbine. The preset equal interval step size is set to 1 rpm. As other implementation methods, implementers can set it according to the actual situation.

[0078] It should be noted that at each moment, two amplitude deviations corresponding to the 1P and 3P frequencies will be extracted for each axis. Thus, six amplitude deviations will be extracted for the three axes. These six amplitude deviations will be combined to form the vibration deviation characteristics at that moment.

[0079] Furthermore, icing reduces the rotational torque of wind turbine blades, resulting in a lower rotor speed under the same conditions compared to when there is no icing. By analyzing the real-time rotor speed deviation from the uniced state, the attenuation of the wind turbine's aerodynamic energy harvesting capability can be effectively characterized.

[0080] Step 203: Calculate the difference between the real-time wind turbine speed and the preset reference speed as the speed deviation feature;

[0081] The process of obtaining the preset reference speed is as follows:

[0082] Wind speed and rotor speed of wind turbines operating normally during historical periods in an ice-free state were collected as wind speed-rotor speed sample data. Using wind speed as the independent variable and rotor speed as the dependent variable, a polynomial fitting algorithm was used to fit the wind speed-rotor speed sample data to obtain a fitting function of the theoretical rotor speed with respect to wind speed. The real-time wind speed of each wind turbine was then substituted into the fitting function of the theoretical rotor speed with respect to wind speed to calculate and output the corresponding theoretical rotor speed, which was used as the preset benchmark speed. The polynomial fitting algorithm used was cubic polynomial fitting, a well-known technique that will not be elaborated upon here.

[0083] Finally, when icing occurs, the pitch system needs to adjust the blade angle more frequently to maintain stable unit operation, resulting in a significant increase in the fluctuation range and rate of pitch angle movement. By extracting the changes in pitch angle, the severity of the pitch system's actions can be reflected.

[0084] Step 204: Define each moment and the consecutive moments preceding it as a local time period, calculate the rate of change of the pitch angle between two adjacent moments within the local time period, and take the average of the absolute values ​​of all the rates of change within the local time period as the pitch angle feature.

[0085] Specifically, each moment and the 60 consecutive moments preceding it are defined as a local time period. As another implementation method, the implementer can set it according to the actual situation. The calculation process of the rate of change is a well-known technology and will not be described in detail here.

[0086] Step 205: The real-time power deviation features, vibration deviation features, speed deviation features, and pitch angle features of each wind turbine are combined into a multi-dimensional feature vector, which is then input into a pre-trained ensemble learning model for icing identification. The model outputs the posterior probability of each wind turbine belonging to each icing level. The icing level corresponding to the highest posterior probability is selected and marked as the real-time icing state of the wind turbine. The icing levels include no icing, light icing, and severe icing.

[0087] The process involves collecting operating parameters of wind turbines that were operating normally during historical periods. Following the above process, power deviation characteristics, vibration characteristics, speed deviation characteristics, and pitch angle characteristics are calculated to form multi-dimensional feature vectors for each moment in the historical period. Training samples are obtained from historical on-site inspection records and de-icing system action records. The training samples are manually labeled and divided into three levels: no icing (label 0), light icing (label 1), and severe icing (label 2), which serve as sample labels. The training samples from all moments in the historical period are combined into a training set. Before inputting the model, the multi-dimensional feature vectors are normalized using the maximum-minimum normalization method. Specifically, the maximum and minimum values ​​of all multi-dimensional feature vectors in the training set are selected in the same dimension. Using the maximum and minimum values, the maximum-minimum normalization method is used to normalize each dimension of the multi-dimensional feature vectors. Furthermore, when the normalized result is less than 0 or greater than 1, a truncation function is used to forcibly truncate it to 0 or 1 to prevent interference from abnormal extreme values. The maximum-minimum normalization method is a well-known technique and will not be elaborated here.

[0088] The ensemble learning model is trained by using the normalized multidimensional feature vectors from the training set as input and the corresponding sample labels as output. In this embodiment, the ensemble learning model adopts the Lightweight Gradient Boosting Tree (LightGBM) model, which is a well-known technique and will not be described in detail here.

[0089] The real-time normalized multidimensional feature vector is input into the trained ensemble learning model, which outputs the posterior probability of the wind turbine at each icing level. The icing level corresponding to the maximum posterior probability is selected to identify the icing level of the wind turbine.

[0090] Furthermore, the actual usable output of wind turbines in cold environments depends not only on conventional parameters such as wind speed and rated power, but also, crucially, on the dual constraints of icing conditions and changes in air density. Icing alters the aerodynamic shape of the blades and reduces the lift-to-drag ratio, significantly decreasing the wind energy capture efficiency of the turbines. Low temperatures in cold regions significantly increase air density; excessive air density increases theoretical wind energy but also drastically increases the thrust load on the rotor, forcing the turbines to actively limit power to avoid overload. In addition, the turbines' mechanical transmission systems also experience inherent efficiency losses at low temperatures. These factors interact, resulting in the maximum safe output of wind turbines being far lower than their rated power. If the grid dispatch cannot accurately perceive this true output boundary and blindly issues high-power commands, it will lead to frequent triggering of safety protection mechanisms and even structural fatigue damage; if the issued commands are too conservative, it will cause severe wind curtailment. Therefore, to ensure the mechanical safety of the turbines and achieve optimal wind energy dispatch across the entire field, it is necessary to calculate and dynamically update the true usable output margin of each wind turbine in real time, based on the identified icing conditions and the characteristics of air density changes. Therefore, in step 3, the specific process is as follows:

[0091] Step 301: Based on the real-time icing status of each wind turbine, a power reduction coefficient is pre-set for it; using the posterior probability of the icing level, this power reduction coefficient is corrected, and the corrected power reduction coefficient is... The calculation formula is: ; In the formula, The output reduction factor before correction. The preset first adjustment value, To preset the second adjustment value, Let be the confidence coefficient, where , Let be the posterior probability that the wind turbine unit is free from icing. This represents the posterior probability that the wind turbine is subject to light icing. Preset weights;

[0092] Specifically, the active power and wind speed of normally operating wind turbines during historical periods are collected at different times. Based on inspection and de-icing records, the actual icing level of the wind turbines at each time is manually marked. The wind speed at each time is substituted into a pre-constructed fitting function of theoretical output power with respect to wind speed to obtain the theoretical output power. The ratio of active power to theoretical output power at each time is calculated as the output reduction coefficient. The range of values ​​for all output reduction coefficients under the same icing level during historical periods is statistically analyzed as the reduction coefficient interval. An empirical value is selected within the reduction coefficient interval corresponding to each icing level to set the output reduction coefficient. In this embodiment, the reduction coefficient interval corresponding to no icing is: The reduction factor range corresponding to light icing is: The reduction factor range corresponding to severe icing is: Therefore, if the current icing level of the wind turbine is no icing, the output reduction factor is set to 1; if the current icing level of the wind turbine is light icing, the output reduction factor is set to 0.85; if the current icing level of the wind turbine is severe icing, the output reduction factor is set to 0.5. As other implementation methods, the implementer can set the factor according to the actual situation.

[0093] Secondly, the preset first adjustment value is set to 0.9, the preset second adjustment value is set to 0.1, and the preset weight is set to 0.5. The preset weight is used to reduce the impact of light icing on... The influence is to ensure the conservative principle of the revision: when At higher levels, the model tends to determine that the unit is in a normal or slightly icing state, in which case the output reduction coefficients before and after correction are relatively close; when When the value is low, the model is uncertain about the judgment result or tends to be in a state of severe icing. In this case, the corrected output reduction factor is reduced to 90% of the original output reduction factor, and a conservative correction is made.

[0094] It should be noted that the corrected output reduction factor is used to characterize the true retention ratio of the aerodynamic energy capture efficiency of the unit under icing conditions. The larger the value, the smaller the performance degradation of the wind turbine due to icing, and the closer its actual power output is to the theoretical normal value. Conversely, the smaller the value, the more severe the icing damages the aerodynamic performance of the blades, and the more the unit must reduce its output.

[0095] Step 302: Calculate the ratio of the real-time air density of each wind turbine to the preset standard air density as the density ratio. Based on this density ratio, determine the density correction coefficient of each wind turbine in real time using a pre-constructed piecewise function.

[0096] Among them, the basic formula based on wind power is: In the formula, The wind energy utilization coefficient, For swept area, air density, For wind speed; ideally, and If both remain constant, then the unit's output power is strictly proportional to the cube of the air density and the wind speed. In this case, the unit operates at a wind speed of... air density is Output power under certain conditions Compared with standard air density Standard wind speed Reference output power under certain conditions The ratio is: ;

[0097] It should be noted that standard wind speed refers to the minimum steady-state ambient wind speed at which a wind turbine reaches its rated output power under standard air density conditions, which is usually the rated wind speed indicated on the wind turbine's nameplate.

[0098] However, in cold environments, the Reynolds number on the blade surface changes under different air densities, leading to nonlinear shifts in airfoil lift-drag characteristics and aerodynamic efficiency. More importantly, the sharp increase in thrust caused by high air density triggers the active pitch limiting power protection of the main control system, resulting in a decrease in the actual wind energy utilization coefficient. The nonlinear decay occurs with increasing wind speed, which eliminates the need for real-time monitoring. To accurately assess power loss under the premise of [previous goal], the constant cube in the formula is discarded, and the exponential parameter to be fitted is used to equivalently absorb [the power loss]. The effects of nonlinear decay and Reynolds number variation. Considering the significant differences in the protection strategy response of the unit under high and low air density conditions, the expression for the piecewise function is: ; In the formula, This represents the density correction factor. The density ratio, i.e. , and For exponential parameters; The index parameter and The acquisition process is as follows: Collect the active power, wind speed, and air density of wind turbines that are ice-free and in normal grid-connected power generation status during historical periods, and calculate the density ratio at each time point. and wind speed ratio The ratio of active power to the unit's baseline power is used as the power ratio, and the data is divided into low-density and high-density sample sets according to the density ratio, i.e., the historical period. The time period is a high-density sample set, within the historical period The time frame is the low-density sample set; with power ratio as the dependent variable and density ratio and wind speed ratio as independent variables, nonlinear exponential fitting is performed on the low-density sample set and the high-density sample set respectively to solve for the exponential parameters. and .

[0099] The fitted result , Substituting the real-time density ratio and wind speed into the piecewise function, the output result is the density correction coefficient of the wind turbine.

[0100] Step 303: Obtain the rated power of each wind turbine, and multiply it by the corrected output reduction factor, density correction factor and preset mechanical attenuation factor to obtain the available output margin of each wind turbine; calculate the ratio of the available output margin to the rated power of each wind turbine in real time to obtain the available output percentage.

[0101] It should be noted that if the available output margin is greater than the rated power, the final available output margin of the wind turbine is directly set to equal its rated power.

[0102] Specifically, the mechanical attenuation coefficient is used to characterize the mechanical transmission efficiency loss of wind turbine units in cold environments. During operation, the aerodynamic energy captured by the wind turbine needs to be transmitted and converted through a series of mechanical and electrical components such as the main shaft, gearbox, and generator. This process inevitably involves friction and heat loss. Especially in extremely low-temperature environments, the viscosity of gearbox lubricating oil increases significantly, and the cold contraction of mechanical components and the increase in the coefficient of friction further exacerbate the mechanical resistance and power loss of the transmission chain. According to relevant testing standards for wind turbine units (such as NB / T 10992-2022) and combined with actual operating statistics of units in cold regions, the mechanical transmission efficiency typically fluctuates between 82% and 98% (i.e., a loss of 2% to 18%). Therefore, the preset range of the mechanical attenuation coefficient is set to 0.82~0.98. In this implementation, the preset mechanical attenuation coefficient is set to 0.87. As for other implementation methods, the implementer can set it according to the actual situation.

[0103] The rated power is set by referring to the wind turbine's manufacturer's datasheet. It should be noted that the available output ratio is equivalent to the product of the corrected output reduction factor, the density correction factor, and the preset mechanical attenuation factor. Therefore, the closer the available output ratio is to 1, the stronger the overall output capacity of the unit; the lower the value, the more severely the overall output capacity is limited.

[0104] In cold-region wind farms, the icing state of wind turbines is a dynamic evolutionary process, not a static event. The formation, intensification, maintenance, and shedding of blade icing are influenced by a combination of factors, including ambient temperature, humidity, and wind speed. Due to differences in micro-topography, installation orientation, and operating strategies, the rate and extent of icing evolution may vary significantly among individual turbines. If severely iced turbines continue to be operated under normal conditions, the blades will be subjected to excessive aerodynamic loads and unbalanced centrifugal forces, potentially leading to safety accidents such as blade structural damage, pitch bearing overload, or even breakage over time. When the icing level exceeds a safe threshold, the turbine must be disconnected from operation or its output limited to protect the equipment. If turbines that have returned to normal operation are still restricted to low-power operation, it results in a waste of clean energy; conversely, if turbines that are ready to output power fail to respond to dispatch requests in a timely manner, it weakens the wind farm's ability to track grid dispatch instructions and affects grid frequency stability. Therefore, in order to achieve the dual goals of safe operation of wind turbines and stable power distribution to the grid, it is necessary to formulate a stepped state switching strategy with hysteresis characteristics based on icing conditions and the proportion of available output. By setting multiple operating states, the unit can gradually reduce the load when performance deteriorates and smoothly recover when performance is restored.

[0105] Step 304: Based on the real-time available output percentage and icing status, execute the state switching strategy, specifically as follows:

[0106] When each wind turbine is in normal operation, if there is no ice-free state for several consecutive moments and the available output ratio is less than the preset first threshold, the normal operation state will be switched to the derating operation state; otherwise, the normal operation state will be maintained.

[0107] When each wind turbine is in derating operation mode, if the icing status is non-icing for multiple consecutive moments and the available output ratio is greater than the preset second threshold, the derating operation mode will be switched to normal operation mode; if the icing status is severe icing for multiple consecutive moments, or the ambient temperature is lower than the preset value, the derating operation mode will be switched to shutdown recovery mode; otherwise, the derating operation mode will be maintained; wherein, the preset second threshold is greater than the preset first threshold.

[0108] When each wind turbine is in the shutdown recovery state, if it receives a message that the de-icing of the turbine has been completed and the ambient temperature has risen above the preset value, it will switch from the shutdown recovery state to the derated operation state; otherwise, it will remain in the shutdown recovery state.

[0109] Specifically, the available output percentage of wind turbines that were in an ice-free state and normally connected to the grid during a historical period is obtained. The lower quartile and median of all available output percentages during the historical period are calculated. The lower quartile is used as a preset first threshold, and the median is used as a preset second threshold. The calculation of the lower quartile and median is a well-known technique and will not be elaborated here. Furthermore, when each wind turbine is in normal operation, if there is no ice-free state for 10 consecutive minutes and the available output percentage is less than the preset first threshold, the system switches from normal operation to derating operation. When each wind turbine is in derating operation, if there is no ice-free state for 15 consecutive minutes and the available output percentage is greater than the preset second threshold, the system switches from derating operation to normal operation. If there is severe icing for 20 consecutive minutes, or the ambient temperature is lower than a preset value, the system switches from derating operation to shutdown recovery. Other implementation methods can be set by the implementer according to the actual situation.

[0110] It should be noted that the state switching follows the principle of step-by-step switching, and switching is only allowed between adjacent states. That is, the normal operation state and the derated operation state can be switched to each other, and the derated operation state and the shutdown recovery state can be switched to each other. Direct switching between the normal operation state and the shutdown recovery state is prohibited to avoid step changes in the output command that could cause grid oscillations and mechanical shocks to the unit's drive chain.

[0111] Furthermore, after confirming that the operating status of each wind turbine needs to be switched, in order to avoid grid oscillation and mechanical shock to the turbine due to a step change in the output allocation command, a preset transition period is set, and the actual output pre-command of each wind turbine is smoothly transitioned through a linear interpolation algorithm.

[0112] Step 305: Based on the operating state that each wind turbine needs to switch into, allocate target power to it, specifically as follows:

[0113] The ratio between the available output ratio of each wind turbine and the sum of the available output ratios of all wind turbines is calculated in real time and used as the allocation weight.

[0114] When each wind turbine needs to switch to normal operation, the target power allocated to it is... for: In the formula, To assign weights; To receive the total power deficit from the grid's dispatch plan, This represents the available output margin. It is a minimum value function;

[0115] When each wind turbine needs to switch to derating operation, the target power allocated to it is... for: In the formula, The preset safety margin factor is a safety margin factor for low power conditions; the preset safety margin factor ranges from 0.9 to 0.95. In this embodiment, the value is set to 0.9. In other implementation methods, the implementer can set it according to the actual situation.

[0116] When each wind turbine needs to switch to a shutdown recovery state, the target power allocated to it is... for: ;

[0117] Step 306: When switching the triggered state, a transition period of preset duration is set, with the trigger time as the starting time; the real-time available output limit of each wind turbine is determined, specifically as follows:

[0118] For any given moment, if it is before the state transition is triggered, its available output limit is the available output margin.

[0119] If any of the aforementioned moments falls within the transition period, a linear interpolation algorithm is used to interpolate between the available power margin at the moment before the state transition is triggered and the target power allocated in the new state. The interpolation result is used as the upper limit of available power. The specific formula for this upper limit of available power is as follows: ; In the formula, This represents the upper limit of available output at each moment during the transition period. The available output margin at the moment preceding the state transition trigger. The target power allocated after the state transition. The time interval between each moment during the transition period and the starting moment. This refers to the total duration of the transition period;

[0120] If any of the aforementioned moments exceeds the end of the transition period, then the upper limit of its available output power is the target power;

[0121] The transition period is 5 minutes. As another implementation method, the implementer can set it according to the actual situation. Setting this transition period and performing linear interpolation can make the upper limit of the available output of the wind turbine smoothly transition from the previous state to the target power of the new state when the state changes. This avoids the step change of the control command, thereby preventing the serious mechanical impact on the transmission components such as the gearbox of the unit caused by the sudden change of the output command, and effectively maintaining the stability of the active power connected to the grid throughout the field.

[0122] Furthermore, in cold environments, the actual power generation capacity of photovoltaic arrays is severely constrained by both extreme low temperatures and snow cover. On the one hand, the photoelectric conversion efficiency of photovoltaic modules is highly sensitive to the cell operating temperature. Unlike normal temperature conditions, the ultra-low temperatures in extremely cold environments can significantly increase the open-circuit voltage of crystalline silicon modules, thereby bringing a reverse gain in power generation efficiency. Ignoring this low-temperature efficiency enhancement effect will lead to a conservative assessment of photovoltaic potential. On the other hand, continuous snowfall will cause snow layers of varying thicknesses and transmittance to form on the surface of photovoltaic modules, resulting in a sharp reduction in effective solar radiation reaching the cell surface. This optical shading effect caused by snow accumulation not only evolves dynamically with meteorological conditions but is also usually coupled with the low-temperature effect, making it a decisive factor in the precipitous drop in photovoltaic output in cold regions. Therefore, in order to accurately quantify the true dispatchable potential of photovoltaic arrays under harsh cold conditions, it is necessary to physically decouple the temperature effect from snow shading and accurately assess the real-time available power limit of the photovoltaic array. Therefore, in step 4, the specific process is as follows:

[0123] Step 401: Calculate the difference between the real-time cell temperature of the photovoltaic array and the preset standard temperature, as the temperature difference; then calculate the real-time temperature efficiency correction coefficient of the photovoltaic array. The specific formula is as follows: ; In the formula, The preset power temperature coefficient, For temperature difference;

[0124] For crystalline silicon modules, the power temperature coefficient is set to... And the preset standard temperature is set to As other implementation methods, implementers can set them according to the actual situation;

[0125] It should be noted that when the temperature difference is less than 0, the temperature efficiency correction factor is greater than 1, indicating that the photovoltaic module's photoelectric conversion efficiency is relatively improved due to the low temperature in cold regions; when the temperature difference is greater than 0, the temperature efficiency correction factor is less than 1, indicating that the heat loss caused by the high temperature leads to the relative decrease in the photovoltaic module's photoelectric conversion efficiency.

[0126] In cold environments, the impact of snow accumulation on photovoltaic (PV) array output is mainly reflected in its blocking of solar radiation. The thicker and denser the snow, the more significant the blocking effect on solar radiation. Conversely, in the early stages of snow accumulation, the snow thickness is thinner and the snow density is lower, resulting in a weaker blocking effect on solar radiation and a weaker impact on PV array output. Furthermore, considering that PV modules are not horizontal but installed at a certain angle, the actual thickness of snow accumulation on the tilted module surface is less than the thickness on a horizontal surface. Therefore, based on geometric projection relationships, and considering the installation tilt angle of the PV modules, snow thickness, and snow density, the equivalent snow thickness of the PV modules is calculated as follows:

[0127] Step 402: Obtain the snow thickness and snow density detected by the weather station; use the ratio of the real-time snow density to the preset new snow density as the density conversion value; calculate the product of the real-time snow thickness, the cosine of the installation tilt angle of the photovoltaic array, and the density conversion value as the real-time equivalent snow thickness; perform a negative mapping on the product of the equivalent snow thickness and the preset snow equivalent attenuation coefficient as the real-time snow attenuation factor.

[0128] The installation tilt angle is set by referring to the photovoltaic array design documents; the range of fresh snow density is... In this embodiment, the preset fresh snow density is set to As other implementation methods, implementers can set them according to the actual situation;

[0129] The attenuation effect of snow on solar radiation from photovoltaic arrays is not consistent across different spectral bands. Transmittance spans several orders of magnitude from the visible to the near-infrared band. When evaluating the overall shading effect of snow, the transmittance at each wavelength needs to be weighted and integrated according to the incident solar spectrum to obtain the equivalent extinction coefficient. Here, considering that in cold environments, snow accumulation on photovoltaic modules typically exists in three forms: fresh snow, compacted snow, and ice, leading to significant differences in their optical attenuation characteristics. In the case of fresh snow, the ice crystals are small and have extremely high porosity, resulting in a small scattering / absorption cross-section per unit volume and strong solar radiation penetration. The equivalent extinction coefficient obtained by weighted integration is the smallest. In the case of compacted snow, the snow particles are tightly bound together, porosity decreases, and the ice-air interface per unit volume increases dramatically, leading to a significant enhancement of optical scattering and a decrease in solar radiation penetration. The equivalent extinction coefficient obtained by weighted integration increases. In the case of ice-covered shells, there are very few internal bubbles. Although scattering is weakened, the absorption path becomes longer, and the smooth ice-covered surface causes radiation to undergo multiple total internal reflections within the ice layer, experiencing a longer effective optical path. This results in the most severe light energy dissipation, and the equivalent extinction coefficient obtained by weighted integration is the largest. Therefore, by determining the snow morphology range based on the variation law of snow density and setting monotonically increasing equivalent snow attenuation coefficients for fresh snow, compacted snow, and ice-covered snow morphologies, the true physical blocking effect of different phases of ice and snow evolution on the power generation potential of photovoltaic modules can be accurately quantified. Therefore, the process for obtaining the preset equivalent snow attenuation coefficient is as follows:

[0130] Based on real-time snow density, determine its snow cover type, specifically: when the snow density is less than or equal to... This is considered a new snow condition, when the snow density is greater than... And less than or equal to 0.4 This is a compacted snow condition; when the snow density is greater than 0.4... It is in a frozen state; therefore:

[0131] When the snowfall is in the fresh snow condition, the effective snow cover attenuation coefficient ranges from 0.015 to 0.025 cm. -1 In this embodiment, the equivalent snow attenuation coefficient is set to 0.02cm. -1 ;

[0132] When the snow is compacted, the effective snow cover attenuation coefficient ranges from 0.035 to 0.045 cm. -1 In this embodiment, the equivalent snow attenuation coefficient is set to 0.04cm. -1 ;

[0133] When the snow is in a frozen state, the effective snow attenuation coefficient ranges from 0.055 to 0.065 cm. -1 In this embodiment, the equivalent snow attenuation coefficient is set to 0.06cm. -1As another implementation method, the implementer can set it according to the actual situation;

[0134] According to the Bouger-Lambert-Beer law, when radiation passes through a homogeneous medium, the transmitted intensity decreases exponentially with the thickness of the medium. Therefore, an exponential function is used for negative mapping, the specific process of which is as follows:

[0135] The product of the equivalent snow thickness and the preset equivalent snow attenuation coefficient is denoted as... ,but The result is used as the result of the negative mapping, where, Represents an exponential function with the natural constant as the base;

[0136] It should be noted that the product of the real-time snow thickness and the cosine of the installation tilt angle of the photovoltaic array represents the thickness of the snow layer perpendicular to the module surface when the snow is uniformly distributed along the tilted surface. This calculation is based on the fact that the path length of solar radiation through the snow is equal to the snow thickness in the normal direction; the larger the tilt angle, the smaller the equivalent snow thickness. The density conversion value reflects the degree of compaction of the snow due to time, wind pressure compaction, or alternating freeze-thaw cycles. The larger this value, the more the current snow deviates from the loose, fresh snow morphology, the smaller the gaps between snow particles, and the higher the water content or ice content. By introducing the density conversion value to calculate the equivalent snow thickness, the diverse morphologies and densities of snow in nature are successfully unified and normalized to the density benchmark of fresh snow, eliminating calculation errors caused by differences in physical density at different stages of snowfall evolution. Therefore, the equivalent snow thickness reflects… The normal optical transmission path length after snowfall actually accumulates and compacts on a tilted photovoltaic panel. The larger this value, the thicker or denser the actual snow layer blocking the photovoltaic panel surface, and the longer the physical distance required for sunlight to penetrate the snow layer to reach the solar cells. The snow attenuation factor reflects the proportion of solar radiation transmitted through the snow layer. The smaller this value, the thicker and denser the snow layer (such as in an ice shell state), the stronger the scattering and absorption blocking effect on sunlight, and the lower the proportion of effective solar radiation that ultimately penetrates the snow to reach the photovoltaic panel, resulting in a severe limitation on the actual usable power generation potential of the photovoltaic array. The larger this value, the thinner the snow on the photovoltaic panel surface and the better the light transmittance (such as a small amount of fresh snow), the weaker the obstruction of sunlight. In this case, most sunlight can penetrate the snow to reach the solar cells, the negative impact of environmental shading on the photovoltaic array is small, and its power generation potential is highly preserved.

[0137] Step 403: Calculate the ratio between the real-time irradiance of the photovoltaic array and the preset standard irradiance, as the irradiance ratio; obtain the rated power of the photovoltaic array, and multiply it by the temperature efficiency correction coefficient, the snow accumulation attenuation factor, and the irradiance ratio, as the real-time available output limit of the photovoltaic array.

[0138] Specifically, this refers to the preset standard irradiance setting. As another implementation method, the implementer can set it according to the actual situation; the rated power can be obtained by looking up the nominal parameters on the photovoltaic array's factory nameplate or the fixed configuration table of the equipment management system.

[0139] It should be noted that the snow accumulation alarm threshold is set to 0.9. When the snow accumulation attenuation factor remains below 0.9, it indicates severe snow cover on the photovoltaic array, triggering a snow melting suggestion. Conversely, when the actual irradiance is below the minimum irradiance, a blackout condition is triggered, causing the photovoltaic modules to cease power output and directly setting the available power output limit to 0. The minimum irradiance is set to... As another implementation method, the implementer can set it according to the actual situation.

[0140] It should be noted that a higher irradiance ratio indicates stronger solar radiation and more energy that the photovoltaic array can capture; a higher upper limit of available output indicates sufficient sunlight, significant efficiency gains from low temperatures, and minimal snow accumulation on the panels, meaning the photovoltaic array possesses extremely high actual power generation capacity.

[0141] Furthermore, based on the real-time icing status of each wind turbine, the available output limits of the wind turbines and photovoltaic arrays, and the real-time status of the energy storage system, the power generation equipment across the entire site is coordinated and scheduled. Specifically, in step 5:

[0142] Step 501: Construct an objective function with the objectives of minimizing grid dispatch tracking deviation, minimizing energy storage output, and minimizing wind and solar curtailment. Specifically, it is expressed as: ;

[0143] In the formula, To receive the total power deficit from the grid's dispatch plan, The rated power of the energy storage system. Total installed capacity for The actual total output at any given moment This indicates the total power of wind and solar power curtailment. , as well as This indicates the preset weighting coefficient. ; This represents a function that takes the minimum value.

[0144] in, , Indicates that the photovoltaic array is in The optimal output needs to be determined at all times. For energy storage systems in The optimal output needs to be determined at all times. For the first Each wind turbine unit The optimal output is to be determined at any given time, where N is the number of all wind turbine units;

[0145] , For photovoltaic arrays in The maximum available output at any given time. For the first Each wind turbine unit The maximum available output at any given time. This represents the function that takes the maximum value.

[0146] It should be noted that, A positive value indicates energy storage discharge, and a negative value indicates energy storage charging; secondly, in this embodiment, Set to 0.5. Set to 0.3. The value is set to 0.2. As another implementation method, the implementer can set it according to the actual situation.

[0147] Step 502, the constraint conditions are:

[0148] Wind turbine output constraints: If the first If a wind turbine is in a shutdown recovery state, then ;

[0149] Photovoltaic array output constraints: ;

[0150] Energy storage system operating constraints: ; ; in, This refers to the minimum allowable charging power limit for energy storage systems. This refers to the maximum allowable discharge power limit for energy storage systems. This represents the minimum state of charge of the battery in the energy storage system. This represents the maximum battery state of charge of the energy storage system. For energy storage systems in The state of charge at any given moment;

[0151] In this embodiment, The value is 20%. If the value is 100%, the output will stop when the battery state of charge of the energy storage system is less than 20%.

[0152] Step 503: Based on the constructed objective function and constraints, a multi-objective optimization algorithm is used to minimize the solution, and the optimal output of each wind turbine, the optimal output of the photovoltaic array, and the optimal output of the energy storage system are calculated.

[0153] Specifically, the multi-objective optimization algorithm adopts the NSGA-II algorithm, which is a well-known technology and will not be described in detail here;

[0154] Based on the optimal output obtained from the solution, the coordination and decision-making center generates differentiated power control commands and issues them to the underlying equipment for execution, in order to coordinate and control the cold-region landscape. Specifically:

[0155] For wind turbines in normal operating condition, issue their corresponding optimal output and require the units to respond normally.

[0156] For wind turbines operating under reduced power rating, their corresponding optimal output is issued as the power-limited operating threshold, requiring the turbines to maintain within this safe power range through active pitch control.

[0157] For wind turbines in the shutdown recovery state, issue shutdown or standby protection commands and prohibit forced grid connection and power output;

[0158] For photovoltaic arrays, their corresponding optimal output is issued as a reference value for active power to control their adjusted output;

[0159] For the energy storage system, its corresponding optimal output is issued as a reference value for charging and discharging power. When the command is positive, the energy storage system is controlled to discharge to make up for the power gap caused by the limited power generation due to icing in cold regions. When the command is negative, the energy storage system is controlled to charge to smooth out the fluctuations and overflows of wind and solar power output across the entire field, and finally achieve stable and coordinated power scheduling of wind, solar and energy storage systems in cold regions.

Claims

1. A method for coordinated wind and solar control in cold regions considering the operating characteristics of wind turbine units, characterized in that, The method includes the following steps: Real-time acquisition of operating parameters of each wind turbine and meteorological parameters of the surrounding environment; extraction of multi-dimensional features reflecting aerodynamic performance and mechanical load anomalies of the wind turbine; and use of classification model for icing identification to obtain the real-time icing status of each wind turbine. The performance degradation of wind turbines under the influence of icing and density coupling with cold air is assessed, the available output margin of each wind turbine is quantified, and the wind turbines are judged to meet the trigger conditions for state switching based on the available output margin and the icing status. The state switching strategy is then executed. The target power is allocated to each wind turbine based on the final operating state to which the wind turbine is to be switched, combined with the grid dispatch plan. After the state switching is triggered, a transition period is preset to control the output limit of the wind turbines to smoothly transition from the output status before the trigger to the target power, so as to determine the available output limit of each wind turbine in real time. The irradiance, battery temperature, and snow cover parameters of the photovoltaic array are collected in real time. The compensation effect of low temperature on photoelectric efficiency and the optical blocking effect of snow on radiation transmission are evaluated to reduce the photovoltaic power generation capacity and determine the upper limit of the usable output of the photovoltaic array. An objective function is constructed to minimize the grid dispatch tracking deviation, energy storage output, and wind and solar curtailment. Constraints are established with the upper limit of available output as the boundary. The objective function is solved under the constraints to calculate the optimal output of each wind turbine and photovoltaic array, so as to perform wind and solar coordinated control.

2. The cold-region wind-solar coordinated control method considering the operating characteristics of wind turbine units as described in claim 1, characterized in that, The operating parameters include active power, rotor speed, blade pitch angle, and triaxial vibration acceleration of the nacelle; the meteorological parameters include wind speed, air temperature, and air density; the extraction of multidimensional features reflecting aerodynamic performance and mechanical load anomalies of the wind turbine specifically includes extracting power deviation features, vibration deviation features, speed deviation features, and blade pitch angle features, and the extraction process is as follows: Construct a fitting function of theoretical output power with respect to wind speed, substitute the real-time wind speed of each wind turbine into the fitting function, and output the theoretical output power corresponding to the wind speed; calculate the difference between the real-time active power and the theoretical output power of each wind turbine as the power deviation feature. Using each moment as the endpoint, a time period of a preset length is extracted and defined as a local time window. The average rotational speed of the wind turbine at all moments within the local time window is calculated as the average rotational speed. The fundamental frequency of the wind turbine rotation is calculated using this average rotational speed as the 1P frequency, and three times this average rotational speed is calculated as the 3P frequency. The 1P frequency and the 3P frequency are collectively referred to as specific frequencies. Frequency domain analysis is performed on the triaxial acceleration at all moments within the local time window, and the corresponding spectrum diagrams for each axis are extracted. The difference between the amplitude at each specific frequency in the spectrum diagram of each axis and the corresponding preset reference amplitude is calculated as the amplitude deviation. The amplitude deviation of the three axes at all specific frequencies is used as the vibration deviation characteristic. The difference between the real-time wind turbine speed and the preset reference speed is calculated as a speed deviation characteristic; Each moment and several consecutive moments preceding it are defined as a local time period. The rate of change of the pitch angle between two adjacent moments within the local time period is calculated, and the average of the absolute values ​​of all rates of change within the local time period is used as the pitch angle characteristic.

3. The cold-region wind-solar coordinated control method considering the operating characteristics of wind turbine units as described in claim 2, characterized in that, The method of using a classification model for icing identification to obtain the real-time icing status of each wind turbine includes: The real-time power deviation characteristics, vibration deviation characteristics, speed deviation characteristics, and pitch angle characteristics of each wind turbine are combined into a multi-dimensional feature vector, which is then input into a pre-trained ensemble learning model for icing identification, and the posterior probability of each wind turbine belonging to each icing level is output. The icing level corresponding to the maximum posterior probability is selected as the real-time icing status of the wind turbine; the icing level includes no icing, light icing, and severe icing.

4. The cold-region wind-solar coordinated control method considering the operating characteristics of wind turbine units as described in claim 3, characterized in that, The quantification of the available output margin of each wind turbine includes: Based on the real-time icing status of each wind turbine, a power reduction factor is pre-set for it; using the posterior probability of the icing level, this power reduction factor is corrected, resulting in the corrected power reduction factor. The calculation formula is: In the formula, The output reduction factor before correction. The preset first adjustment value, To preset the second adjustment value, Let be the confidence coefficient, where , Let be the posterior probability that the wind turbine unit is free from icing. This represents the posterior probability that the wind turbine is subject to light icing. Preset weights; Calculate the ratio of the real-time air density of each wind turbine to the preset standard air density, and use this ratio as the density ratio. Based on this density ratio, use a pre-built piecewise function to determine the density correction coefficient of each wind turbine. Obtain the rated power of each wind turbine, and multiply it by the corrected output reduction factor, density correction factor, and preset mechanical attenuation factor to obtain the usable output margin of each wind turbine.

5. The cold-region wind-solar coordinated control method considering the operating characteristics of wind turbine units as described in claim 3, characterized in that, The step of determining whether the wind turbine meets the triggering conditions for state switching in order to execute the state switching strategy includes: The ratio of the available output margin to the rated power of each wind turbine is calculated in real time and used as the proportion of available output. When each wind turbine is in normal operation, if there is no ice-free state for several consecutive moments and the available output ratio is less than the preset first threshold, the normal operation state will be switched to the derating operation state; otherwise, the normal operation state will be maintained. When each wind turbine is in derating operation mode, if the icing status is non-icing for multiple consecutive moments and the available output ratio is greater than the preset second threshold, the derating operation mode will be switched to normal operation mode; if the icing status is severe icing for multiple consecutive moments, or the ambient temperature is lower than the preset value, the derating operation mode will be switched to shutdown recovery mode; otherwise, the derating operation mode will be maintained; wherein, the preset second threshold is greater than the preset first threshold. When each wind turbine is in the shutdown recovery state, if it receives a message that the de-icing of the turbine has been completed and the ambient temperature has risen above the preset value, it will switch from the shutdown recovery state to the derated operation state; otherwise, it will remain in the shutdown recovery state.

6. The cold-region wind-solar coordinated control method considering the operating characteristics of wind turbine units as described in claim 5, characterized in that, The allocation of target power to each wind turbine includes: The ratio between the available output ratio of each wind turbine and the sum of the available output ratios of all wind turbines is calculated in real time and used as the allocation weight. When each wind turbine needs to switch to normal operation, the target power allocated to it is... for: In the formula, To assign weights; To receive the total power deficit from the grid's dispatch plan, This represents the available output margin. It is a minimum value function; When each wind turbine needs to switch to derating operation, the target power allocated to it is... for: In the formula, This is a safety margin factor preset for low power. When each wind turbine needs to switch to a shutdown recovery state, the target power allocated to it is... for: .

7. The cold-region wind-solar coordinated control method considering the operating characteristics of wind turbine units as described in claim 1, characterized in that, Determining the upper limit of available output for each wind turbine includes: For any given moment, if it is before the state transition is triggered, its available output limit is the available output margin. If any of the aforementioned moments falls within the transition period, a linear interpolation algorithm is used to interpolate between the available output margin at the moment before the state switch is triggered and the target power allocated in the new state, and the interpolation result is used as the upper limit of available output. If any of the stated moments exceeds the end of the transition period, then the upper limit of its available output is the target power.

8. The cold-region wind-solar coordinated control method considering the operating characteristics of wind turbine units as described in claim 1, characterized in that, The snow accumulation parameters include the snow thickness and snow density on the horizontal surface; determining the upper limit of the available output of the photovoltaic array specifically includes: Real-time temperature efficiency correction factor for photovoltaic array The specific formula is as follows: In the formula, The preset power temperature coefficient, The temperature difference is the difference between the real-time cell temperature of the photovoltaic array and the preset standard temperature. The ratio of real-time snow density to preset new snow density is used as the density conversion value; the product of real-time snow thickness, cosine of photovoltaic array installation tilt angle, and density conversion value is calculated as the real-time equivalent snow thickness; the product of this equivalent snow thickness and preset snow equivalent attenuation coefficient is negatively mapped as the real-time snow attenuation factor. Calculate the ratio between the real-time irradiance of the photovoltaic array and the preset standard irradiance, and use it as the irradiance ratio; obtain the rated power of the photovoltaic array, and multiply it by the temperature efficiency correction coefficient, the snow accumulation attenuation factor, and the irradiance ratio, and use it as the upper limit of the available output of the photovoltaic array.

9. The cold-region wind-solar coordinated control method considering the operating characteristics of wind turbine units as described in claim 1, characterized in that, The process of constructing the objective function is as follows: objective function for: In the formula, To receive the total power deficit from the grid's dispatch plan, The rated power of the energy storage system. Total installed capacity for The actual total output at any given moment This represents the total power of wind and solar power curtailed. , as well as This indicates the preset weighting coefficient. ; in, , Indicates that the photovoltaic array is in The optimal output needs to be determined at all times. For energy storage systems in The optimal output needs to be determined at all times. For the first Each wind turbine unit The optimal output is to be determined at any given time, where N is the number of all wind turbine units; , For photovoltaic arrays in The maximum available output at any given time. For the first Each wind turbine unit The maximum available output at any given time. This represents the function that takes the maximum value. This represents the function that takes the minimum value.

10. The cold-region wind-solar coordinated control method considering the operating characteristics of wind turbine units as described in claim 9, characterized in that, The constraints are as follows: Wind turbine output constraints: If the first If a wind turbine is in a shutdown recovery state, then ; Photovoltaic array output constraints: ; Energy storage system operating constraints: ; ,in, This refers to the minimum allowable charging power limit for energy storage systems. This refers to the maximum allowable discharge power limit for energy storage systems. This represents the minimum state of charge of the battery in the energy storage system. This represents the maximum battery state of charge of the energy storage system. For energy storage systems in The state of charge at any given moment.

Images