A control method and device of a photovoltaic tracking support and a storage medium

Abstract

This disclosure provides a control method, device, and storage medium for a photovoltaic (PV) tracking bracket, comprising: predicting the electricity price within a first preset time range based on day-ahead electricity price data to determine a first electricity price curve; performing multiple iterations using the NSGA-II algorithm to obtain a new population satisfying a first constraint condition in each iteration, inputting it into a first objective function to obtain an income index, and determining the individual with the highest income index in the final generation as a second turning curve; predicting the electricity price within a second preset time range based on real-time electricity price data to determine a second electricity price curve; and fine-tuning the second turning curve within a second constraint condition based on the second electricity price curve to control the rotation of the PV tracking bracket. This disclosure, by integrating electricity price prediction and irradiance prediction, using the NSGA-II algorithm to select the turning angle of the PV tracking bracket, and combining real-time electricity price data to control the PV tracking bracket, reduces maintenance costs and maximizes power plant revenue.

Description

Technical Field

[0001] This disclosure relates to the field of photovoltaic power generation technology, and in particular to a control method, device and storage medium for a photovoltaic tracking bracket. Background Technology

[0002] With the market-oriented development of the photovoltaic industry, the all-electricity market has brought numerous challenges to photovoltaic power plants. Due to frequent market price fluctuations, the returns of photovoltaic power plants are no longer stable. Traditional photovoltaic tracking bracket control methods have exposed many problems, leading to significant risks of reduced profitability and increased curtailment rates. Traditional photovoltaic tracking bracket control methods are often based on fixed algorithms and preset programs, with the primary goal of maximizing solar power generation by maximizing solar module sunlight absorption. However, in market trading, simply pursuing power generation is no longer the optimal solution. Frequent market price fluctuations mean that when market electricity prices are low, even if power generation is increased, this excess electricity will not generate revenue and may instead cause damage to photovoltaic modules, increase maintenance costs, and lead to a decrease in overall profitability. Due to the lack of a dynamic response mechanism to market demand and grid absorption capacity, if photovoltaic power plants continue to focus on increasing power generation when grid absorption capacity is limited, electricity exceeding the grid's absorption capacity will be discarded, resulting in an increased curtailment rate. Summary of the Invention

[0003] This disclosure provides a control method, device, and storage medium for a photovoltaic tracking bracket, which addresses the problems of reduced profitability and increased curtailment rate in existing photovoltaic power plants aimed at increasing power generation.

[0004] In view of the above problems, firstly, the present disclosure provides a control method for a photovoltaic tracking bracket, comprising:

[0005] Acquire meteorological observation data, day-ahead electricity price data, and real-time electricity price data;

[0006] Based on the meteorological observation data, predict the irradiance within a first preset time range in the future, and determine the first irradiance curve that changes irradiance over time;

[0007] Based on the current day electricity price data, predict the electricity price within a first preset time range in the future, and determine a first electricity price curve that changes with time;

[0008] The first rotation curve of the photovoltaic tracking bracket is randomly generated as an individual. Multiple individuals are combined into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the population is screened to obtain a new population that meets the first constraint condition. The new population, the first irradiance curve and the first electricity price curve are input into the first objective function to obtain the income index. The individual with the highest income index in the final generation is determined as the second rotation curve.

[0009] Based on the real-time electricity price data, predict the electricity price within a second preset time range in the future, and determine a second electricity price curve that changes with time;

[0010] The second turning angle curve is fine-tuned within the second constraint conditions based on the second electricity price curve, and the rotation of the photovoltaic tracking bracket is controlled according to the fine-tuned second turning angle curve.

[0011] In conjunction with the first aspect, in one possible implementation, the first angle curve of the randomly generated photovoltaic tracking bracket changing with time is used as an individual. Multiple individuals are combined into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the current population is screened to obtain a new population that meets the first constraint condition. The new population, the first irradiance curve, and the first electricity price curve are input into a first objective function to obtain an income index. The individual with the highest income index in the final generation is determined as the second angle curve, including:

[0012] The first rotation angle curve, which is randomly generated and varies with time, is taken as an individual. Multiple individuals are formed into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the following operation is performed until the preset number of iterations is reached and the iteration stops. The individual with the highest income index in the final generation is determined as the second rotation angle curve:

[0013] Each individual in the population is input into the first constraint for screening, resulting in a new population that satisfies the first constraint.

[0014] Each individual in the new population, the first irradiance curve, and the first electricity price curve are input into the first objective function to obtain the income indicator corresponding to each individual;

[0015] The individuals in the new population are sorted according to the size of the income index to obtain a preset number of optimal individuals. The population composed of the preset number of optimal individuals is used as the parent population.

[0016] Perform a crossover operation on the individuals in the parent population to generate intermediate individuals;

[0017] The intermediate individuals are mutated to generate offspring individuals;

[0018] The combination of individuals from the parent population and the offspring individuals is determined as a new population, and the next iteration begins.

[0019] In conjunction with the first aspect, in one possible implementation, the first constraint condition includes at least one of the following: a constraint condition on the rotational speed of the photovoltaic tracking bracket and a constraint condition on the operating temperature of the photovoltaic tracking bracket.

[0020] The first objective function includes: the difference between the first power generation revenue and the first maintenance cost;

[0021] The first revenue from electricity generation is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price;

[0022] The first maintenance cost is related to the rotation angle of the photovoltaic tracking bracket.

[0023] In conjunction with the first aspect, in one possible implementation, the meteorological observation data includes cloud data monitored in real time by meteorological radar;

[0024] The second electricity price curve is fine-tuned within the second constraint condition for the second turning angle curve, and the rotation of the photovoltaic tracking bracket is controlled according to the fine-tuned second turning angle curve, including:

[0025] The first irradiance curve is corrected based on the cloud data to determine the second irradiance curve within a future second preset time range;

[0026] Based on the second turning curve, a third turning curve of the photovoltaic tracking bracket changing with time is randomly generated within the second constraint condition as an individual. Multiple individuals are combined into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the current population is screened to obtain a new population that meets the second constraint condition. The new population, the second irradiance curve, and the second electricity price curve are input into the second objective function to obtain the income index. The individual with the highest income index in the final generation is determined as the fourth turning curve.

[0027] The rotation of the photovoltaic tracking bracket is controlled according to the fourth turning angle curve;

[0028] The second constraint includes: the constraint on the range of rotation angle of the photovoltaic tracking bracket;

[0029] The second objective function includes: the difference between the second power generation revenue and the second maintenance cost;

[0030] The second revenue from electricity generation is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price;

[0031] The second maintenance cost is related to the rotation angle of the photovoltaic tracking bracket.

[0032] In conjunction with the first aspect, in one possible implementation, the meteorological observation data includes cloud data monitored in real time by meteorological radar;

[0033] When the price change within a second preset time range, as represented by the second electricity price curve, exceeds a preset threshold, the step of fine-tuning the second turning curve within the second constraint based on the second electricity price curve, and controlling the rotation of the photovoltaic tracking bracket based on the fine-tuned second turning curve, includes:

[0034] The first irradiance curve is corrected based on the cloud data to determine the third irradiance curve within a future second preset time range;

[0035] Based on the second turning curve, a fifth turning curve of the photovoltaic tracking bracket changing with time is randomly generated within the second constraint condition as an individual. Multiple individuals are combined into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the current population is screened to obtain a new population that meets the third constraint condition. The new population, the third irradiance curve and the second electricity price curve are input into the third objective function to obtain the income index. The individual with the highest income index in the final generation is determined as the sixth turning curve.

[0036] The rotation of the photovoltaic tracking bracket is controlled according to the sixth turning angle curve;

[0037] The second constraint includes: the constraint on the range of rotation angle of the photovoltaic tracking bracket;

[0038] The third objective function includes: the difference between the third power generation revenue and the third maintenance cost plus the sum of shadow revenue;

[0039] The third type of electricity revenue is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price;

[0040] The third maintenance cost is related to the rotation angle of the photovoltaic tracking bracket;

[0041] The shadow revenue is related to the preset shadow coefficient, the rotation angle of the photovoltaic tracking bracket, the irradiance, and the electricity price.

[0042] In conjunction with the first aspect, in one possible implementation, the meteorological observation data includes: numerical weather prediction data and sky imager data;

[0043] The step of predicting the irradiance within a first preset time range based on the meteorological observation data and determining the first irradiance curve as a function of time includes:

[0044] Based on the numerical weather forecast data and sky imager data, predict the irradiance within a first preset time range in the future, and determine the first irradiance curve that changes irradiance over time.

[0045] In conjunction with the first aspect, in one possible implementation, the step of predicting future electricity prices within a first preset time range based on the current-day electricity price data and determining a first electricity price curve that changes over time includes:

[0046] The day-ahead electricity price data is input into the ARIMA model to determine the first residual sequence;

[0047] The first residual sequence is input into the LSTM model to predict the electricity price within a first preset time range in the future, and the first electricity price curve is determined as the electricity price changes over time.

[0048] The step of predicting the electricity price within a second preset time range based on the real-time electricity price data and determining the second electricity price curve that changes over time includes:

[0049] The real-time electricity price data is input into the ARIMA model to determine the second residual sequence;

[0050] The second residual sequence is input into the LSTM model to predict the electricity price within a second preset time range in the future, thereby determining the second electricity price curve that changes with time.

[0051] In conjunction with the first aspect, in one possible implementation, the time period of the change in the rotation angle of the photovoltaic tracking bracket over time in the first rotation curve is the same as the time period for the release of the real-time electricity price data.

[0052] The second rotation curve is used to control the rotation of the photovoltaic tracking bracket within a first preset time range in the future.

[0053] Secondly, a control device for a photovoltaic tracking bracket is provided, comprising:

[0054] The data acquisition module is used to acquire meteorological observation data, day-ahead electricity price data, and real-time electricity price data;

[0055] The day-ahead forecasting module is used to predict the irradiance within a first preset time range based on the meteorological observation data, and determine the first irradiance curve as the irradiance changes over time; predict the electricity price within the first preset time range based on the day-ahead electricity price data, and determine the first electricity price curve as the electricity price changes over time; randomly generate the first angle curve as an individual for the photovoltaic tracking bracket, and form an initial population from multiple individuals. The NSGA-II algorithm is used for multiple iterations. In each iteration, the current population is screened to obtain a new population that meets the first constraint condition. The new population, the first irradiance curve, and the first electricity price curve are input into the first objective function to obtain the income index. The individual with the highest income index in the final generation is determined as the second angle curve.

[0056] The real-time correction module is used to predict the electricity price within a second preset time range based on the real-time electricity price data, determine the second electricity price curve that changes with time, fine-tune the second turning curve within the second constraint conditions based on the second electricity price curve, and control the rotation of the photovoltaic tracking bracket based on the fine-tuned second turning curve.

[0057] Thirdly, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, performs the steps of the control method for a photovoltaic tracking bracket as described in the first aspect or in any possible embodiment of the first aspect.

[0058] The beneficial effects of the embodiments disclosed herein include:

[0059] The control method, device, and storage medium for photovoltaic tracking brackets disclosed herein include: acquiring meteorological observation data, day-ahead electricity price data, and real-time electricity price data; predicting irradiance within a first preset time range based on meteorological observation data, and determining a first irradiance curve that varies with time; predicting electricity price within the first preset time range based on day-ahead electricity price data, and determining a first electricity price curve that varies with time; randomly generating a first angle curve that varies with time as an individual for the photovoltaic tracking bracket, forming an initial population from multiple individuals, performing multiple iterations using the NSGA-II algorithm, filtering the current population in each iteration to obtain a new population that satisfies a first constraint condition, inputting the new population, the first irradiance curve, and the first electricity price curve into a first objective function to obtain an income index, and determining the individual with the highest income index in the final generation as a second angle curve; predicting electricity price within a second preset time range based on real-time electricity price data, and determining a second electricity price curve that varies with time; fine-tuning the second angle curve within a second constraint condition based on the second electricity price curve, and controlling the rotation of the photovoltaic tracking bracket based on the fine-tuned second angle curve. The photovoltaic tracking bracket control method provided in this disclosure, compared with the prior art, integrates electricity price prediction and irradiance prediction, uses the NSGA-II algorithm to select the photovoltaic tracking bracket rotation angle that maximizes power generation revenue, reduces maintenance costs, and combines real-time electricity price data to fine-tune the rotation angle, thereby controlling the photovoltaic tracking bracket, maximizing power plant revenue, and reducing curtailment rate. Attached Figure Description

[0060] Figure 1 A flowchart of a control method for a photovoltaic tracking bracket provided in an embodiment of this disclosure;

[0061] Figure 2 A schematic diagram of a control method for a photovoltaic tracking bracket provided in an embodiment of this disclosure;

[0062] Figure 3 This is a structural diagram of the control device for a photovoltaic tracking bracket provided in an embodiment of the present disclosure. Detailed Implementation

[0063] This disclosure provides a control method, apparatus, and storage medium for a photovoltaic tracking bracket. Preferred embodiments of this disclosure are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustrative and explanatory purposes only and are not intended to limit the scope of this disclosure. Furthermore, the embodiments and features described herein can be combined with each other unless otherwise specified.

[0064] This disclosure provides a control method for a photovoltaic tracking bracket, such as... Figure 1 As shown, it includes the following steps:

[0065] S101. Obtain meteorological observation data, day-ahead electricity price data, and real-time electricity price data;

[0066] S102. Based on meteorological observation data, predict the irradiance within the first preset time range in the future, and determine the first irradiance curve that changes with time.

[0067] S103. Based on the current electricity price data, predict the electricity price within the first preset time range in the future, and determine the first electricity price curve that changes with time;

[0068] S104. Randomly generate the first turning angle curve of the photovoltaic tracking bracket changing with time as an individual, and form an initial population from multiple individuals. Use the NSGA-II algorithm to perform multiple iterations. In each iteration, the current population is screened to obtain a new population that meets the first constraint condition. Input the new population, the first irradiance curve and the first electricity price curve into the first objective function to obtain the income index. The individual with the highest income index in the final generation is determined as the second turning angle curve.

[0069] S105. Based on real-time electricity price data, predict the electricity price within a second preset time range in the future, and determine the second electricity price curve that changes with time;

[0070] S106. Fine-tune the second turning angle curve within the second constraint conditions according to the second electricity price curve, and control the rotation of the photovoltaic tracking bracket according to the fine-tuned second turning angle curve.

[0071] This disclosure pertains to the field of photovoltaic (PV) power generation technology. To promote the market-oriented development of the PV industry and gradually reduce its reliance on government subsidies, full-electricity market trading has become a crucial measure. Under this model, PV power plants no longer rely solely on fixed feed-in tariffs but directly participate in electricity market competition, trading electricity with various power users and electricity retailers. This shift marks a significant transition in the PV industry from a policy-driven to a market-driven model. However, full-electricity market trading also presents numerous challenges for PV power plants. On one hand, market prices fluctuate frequently, making the returns of PV power plants unstable. Factors such as market supply and demand, energy policy adjustments, and the randomness of new energy generation all influence electricity prices. For example, during periods of high new energy generation, the electricity market may experience oversupply, leading to a significant drop in electricity prices and impacting power plant revenue. On the other hand, PV power plants participating in market trading often lack robust market analysis capabilities, further reducing their profitability. In the context of full-electricity market trading, traditional PV tracking bracket control methods reveal numerous problems, significantly increasing the risk of reduced returns and higher curtailment rates for PV power plants. Traditional photovoltaic (PV) tracking system control methods are often based on fixed algorithms and preset programs, primarily aiming to maximize the amount of sunlight received by PV modules to increase power generation. For example, astronomical algorithms and light-sensor tracking aim to maximize instantaneous power generation, without considering the impact of electricity price fluctuations on profitability. However, in market transactions, simply pursuing maximum power generation is no longer the optimal solution. Market prices fluctuate frequently, and when market electricity prices are low, even if traditional control methods increase power generation, this excess electricity may not bring substantial profits. Instead, the cost of power generation, such as wear and tear on the PV tracking system due to frequent angle adjustments, may reduce the overall rate of return. For instance, during periods of high renewable energy generation, many PV power plants operate at full capacity simultaneously, leading to an oversupply in the electricity market and a sharp drop in electricity prices. In such situations, the excess electricity generated under traditional control methods may be sold at low prices, or even face a lack of buyers, undoubtedly eroding the profit margins of the power plants. From the perspective of curtailment rates, traditional PV tracking system control methods lack a dynamic response mechanism to market demand and grid absorption capacity. When the grid's absorption capacity is limited, if a photovoltaic (PV) power plant continues to generate electricity at its maximum capacity, the excess electricity is discarded, leading to a high curtailment rate. Under a full-market electricity trading model, PV power plants need to trade electricity with various power users and electricity sales companies. If they cannot adjust their power generation plans in a timely manner according to market demand and grid dispatch instructions, a mismatch between power generation and market demand can easily occur. For example, when the grid load is low, even with good sunlight conditions, PV power plants may need to limit power generation to avoid curtailment. Traditional PV tracking bracket control methods are difficult to flexibly adjust in real time to these complex and ever-changing market factors, resulting in a persistently high curtailment rate for PV power plants participating in market transactions, further reducing the actual revenue of the power plants.

[0072] In this embodiment of the disclosure, meteorological observation data is acquired, which may include at least one of the following: numerical weather prediction data, sky imager data, and cloud data monitored in real time by weather radar. Numerical Weather Prediction (NWP) simulates the changes in atmospheric state over time by solving the physical equations describing atmospheric motion. It obtains numerical weather prediction data such as cloud cover, cloud height, and solar radiation by reanalyzing meteorological data detected through ground observations, satellite remote sensing, and weather radar. A sky imager is a device used to capture sky images and analyze cloud distribution, type, and movement to predict cloud changes and their impact on solar radiation, thus obtaining sky imager data. Cloud data monitored in real time by weather radar includes: cloud location and extent, cloud thickness, cloud texture and boundaries, cloud reflectivity, cloud movement speed and direction, and cloud development and dissipation processes. Based on meteorological observation data, the irradiance received by photovoltaic modules within a first preset time range can be predicted. This first preset time range can be 24 hours, resulting in an irradiance curve that changes over time. The horizontal axis of the first irradiance curve represents time, and the vertical axis represents the irradiance value. Irradiance is a crucial parameter for calculating the power generation of photovoltaic modules. By connecting to a power trading platform, day-ahead and real-time electricity price data can be obtained. Day-ahead electricity price data refers to the electricity price data formed based on the results of day-ahead market transactions. The day-ahead market refers to the electricity trading market conducted the day before the transaction date. Market participants declare their electricity volume and price for the next day based on predicted electricity demand and their own power generation plans, and the day-ahead electricity price data is formed through a market clearing mechanism. Real-time electricity price data refers to the electricity price data formed based on the results of real-time market transactions. The real-time market refers to the electricity trading market conducted on the transaction day. Market participants declare their electricity volume and price in the real-time market based on real-time electricity supply and demand and grid operation status, and the real-time electricity price data is formed through a market clearing mechanism. Based on current electricity price data, electricity prices within a first preset time range are predicted, for example using a neural network model, to determine a first electricity price curve that changes over time. The x-axis of the first electricity price curve represents time, and the y-axis represents electricity price. A set of first angle curves that change the rotation angle of photovoltaic tracking brackets over time are randomly generated. The x-axis of the first angle curve represents time, and the y-axis represents the rotation angle of the photovoltaic tracking bracket. The time span of the first angle curve is the first preset time. Each first angle curve is treated as an individual, and multiple first angle curves are used as an initial population to perform multiple iterations using the NSGA-II algorithm.NSGA-II (Non-dominated Sorting Genetic Algorithm II) is a multi-objective optimization algorithm that maintains solution diversity while ensuring solution quality. It is applied to solving optimization problems with objective functions. The first constraint is used to filter individuals in the population. In each iteration, the population is filtered to obtain a new population that satisfies the first constraint. The first objective function is used to calculate the revenue index of photovoltaic power generation. The revenue index can consist of power generation revenue and maintenance costs. The new population, the first irradiance curve, and the first electricity price curve can be input parameters of the first objective function. The new population, the first irradiance curve, and the first electricity price curve are input into the first objective function. These parameters determine the power generation revenue and maintenance costs. The NSGA-II algorithm stops iterating after a preset number of iterations (e.g., 100 iterations) to obtain the optimal solution. The individual with the highest revenue index in the final generation is identified as the second turning curve, and the time span of the second turning curve is a first preset time.

[0073] Furthermore, based on real-time electricity price data, the electricity price within a second preset time range is predicted, for example, using a neural network model, to determine a second electricity price curve that changes over time. The horizontal axis of the second electricity price curve represents time, and the vertical axis represents electricity price. The second preset time is shorter than the first preset time, providing more accurate prediction results; for example, the second preset time can be 4 hours. Rolling predictions can be performed based on the release cycle of real-time electricity price data, thereby correcting the second electricity price curve in real time. The release cycle of electricity price data is typically 15 minutes or 1 hour. The second turning angle curve is fine-tuned within a second constraint condition based on the second electricity price curve. The second constraint condition limits the adjustment range of the photovoltaic tracking bracket's turning angle. For example, the second electricity price curve and the first electricity price curve are compared. When the price difference exceeds a preset threshold, the time range corresponding to the price difference exceeding the preset threshold is obtained. Within this time range, the turning angle of the photovoltaic tracking bracket in the second turning angle curve is fine-tuned, thereby adjusting the power generation of the photovoltaic modules to adapt to the impact of electricity price changes on photovoltaic power generation revenue. The rotation of the photovoltaic tracking bracket is controlled according to the fine-tuned second turning angle curve. For example,... Figure 2As shown, by acquiring meteorological observation data, day-ahead electricity price data, and real-time electricity price data, irradiance and electricity price are predicted, and the rotation angle curve of the photovoltaic tracking bracket changes over time is output to control the rotation of the photovoltaic tracking bracket. During the morning peak period in winter, the rotation angle is proactively increased by 15°, sacrificing some irradiance for higher revenue during peak electricity price periods. During the evening peak period, the rotation angle reset is delayed to extend the effective power generation during high electricity price periods. A 5° rotation angle deviation is allowed during the noon period to reduce mechanical movements. During the summer noon period, the rotation angle is proactively deviated by 20° to reduce excessive power generation during low price periods. During the evening peak period, the rotation angle is adjusted in the opposite direction by 10° to utilize the solar altitude angle margin to increase output during periods of rising electricity prices.

[0074] Compared with the prior art, this application embodiment integrates electricity price prediction and irradiance prediction, uses the NSGA-II algorithm to screen out the photovoltaic tracking bracket rotation angle that maximizes power generation revenue, reduces maintenance costs, and combines real-time electricity price data to fine-tune the rotation angle, thereby controlling the photovoltaic tracking bracket, maximizing power plant revenue, and reducing curtailment rate.

[0075] In another embodiment of this disclosure, in step S104 above, a first rotation angle curve of the photovoltaic tracking bracket changing with time is randomly generated as an individual. Multiple individuals are combined into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the current population is screened to obtain a new population that meets the first constraint condition. The new population, the first irradiance curve, and the first electricity price curve are input into the first objective function to obtain the income index. The individual with the highest income index in the final generation is determined as the second rotation angle curve, including:

[0076] The first rotation angle curve, which is randomly generated and varies with time, is taken as an individual. Multiple individuals are formed into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the following operation is performed until the preset number of iterations is reached and the iteration stops. The individual with the highest income index in the final generation is determined as the second rotation angle curve:

[0077] Step 1: Input each individual in the population into the first constraint condition for screening, and obtain a new population that satisfies the first constraint condition;

[0078] Step 2: Input each individual in the new population, the first irradiance curve, and the first electricity price curve into the first objective function to obtain the income indicator corresponding to each individual;

[0079] Step 3: Sort each individual in the new population according to the size of the income index to obtain a preset number of optimal individuals, and use the population composed of the preset number of optimal individuals as the parent population.

[0080] Step 4: Perform crossover on individuals in the parent population to generate intermediate individuals;

[0081] Step 5: Perform mutation operations on the intermediate individuals to generate offspring individuals;

[0082] Step 6: Determine the combination of individuals in the parent population and offspring individuals as the new population, and proceed to the next iteration.

[0083] In this embodiment, the NSGA-II algorithm is used to select the optimal solution that meets the constraints. A first angle curve representing the change of the photovoltaic tracking bracket's angle over time is taken as an individual. This first angle curve can be represented as θ(t), where t represents time, falling within a first preset time range, and θ represents the photovoltaic tracking bracket's angle. A preset population size is set, for example, 100 individuals. These individuals are formed into an initial population. The NSGA-II algorithm is used for multiple iterations, performing the following operations in each iteration until a preset number of iterations is reached. The individual with the highest income index in the final generation is determined as the second angle curve.

[0084] For step 1 above, each individual in the population is input into a first constraint for screening. The first constraint may include constraints on the rotation rate of the photovoltaic tracking bracket and the operating temperature of the photovoltaic tracking bracket. For example, the constraint on the rotation rate requires the photovoltaic tracking bracket to rotate at a rate less than or equal to 3°C per minute, and the constraint on the operating temperature requires the operating temperature to be less than or equal to 85°C. The rotation rate can be calculated for each individual, expressed by the following formula:

[0085]

[0086] The influence of the operating temperature T(t) on the first irradiance curve, ambient temperature, thermal properties of the support material, and the rotation angle of the photovoltaic tracking support can be solved based on experimental data. Individuals satisfying the first constraint condition will be grouped into a new population.

[0087] Regarding step 2 above, the formula for the first objective function is as follows:

[0088]

[0089] Where R represents the income indicator, P elec η(t) represents the electricity price at time t, η(t) represents the photoelectric conversion efficiency at time t, G(t) represents the irradiance at time t, S represents the area of ​​the solar panel, Δt represents the time step, and C main (Δθ) represents the maintenance cost caused by the change in rotation angle Δθ. elecThe photoelectric conversion efficiency η(t) can be obtained from the first electricity price curve. The photoelectric conversion efficiency η(t) is related to factors such as the solar incidence angle and temperature, and can be determined experimentally or through empirical formulas. For example, the formula for photoelectric conversion efficiency is expressed as:

[0090] η(t)=η0·(1-β·(T(t)-T0))

[0091] Where η0 represents the standard efficiency, β represents the temperature coefficient, T(t) represents the operating temperature, and T0 represents the standard temperature. Irradiance G(t) can be obtained from the first irradiance curve. The solar panel area S can be determined based on the actual photovoltaic system design. The time step Δt can be consistent with the release cycle of real-time electricity price data. Maintenance cost C main (Δθ) is related to the change in rotation angle Δθ, and the formula is expressed as:

[0092] C main =k1·∑|Δθ|+k2·∑(Δθ) 2

[0093] Here, k1 and k2 are coefficients that can be determined through accelerated life testing, such as a 100,000-cycle rotation test. Introducing maintenance costs can prevent over-optimization. By inputting each individual in the new population, the first irradiance curve, and the first electricity price curve into the first objective function, the income indicator corresponding to each individual can be obtained.

[0094] For step 3 above, each individual in the new population is sorted according to the size of the income index. A preset number of optimal individuals are obtained from the sorting from high to low. The population composed of the preset number of optimal individuals is used as the parent population.

[0095] For step 4 above, a crossover operation is performed on individuals in the parent population. For example, two individuals from the parent population are randomly selected, denoted as P1 and P2, and the two intermediate individuals generated by the crossover operation are denoted as C1 and C2. The crossover operation can be represented as:

[0096] C1=0.5·(P1+P2)+0.5·β·(P1-P2)

[0097] C2=0.5·(P1+P2)-0.5·β·(P1-P2)

[0098] Where β is the cross-distribution parameter.

[0099] Regarding step 5 above, a mutation operation is performed on the intermediate individuals to introduce randomness and generate offspring individuals, expressed by the formula:

[0100] θ′(t)=θ(t)+δ

[0101] Where δ is a random perturbation that follows a multinomial distribution.

[0102] In step 6 above, the combination of individuals from the parent population and offspring individuals is determined as a new population, and the next iteration begins. Through these steps, the optimization method based on the NSGA-II algorithm can effectively solve the optimization problem of the tracking support rotation angle and generate the optimal rotation angle curve that satisfies the constraints.

[0103] In another embodiment of this disclosure, the first constraint condition includes at least one of the following: a constraint condition on the rotational speed of the photovoltaic tracking bracket and a constraint condition on the operating temperature of the photovoltaic tracking bracket.

[0104] The first objective function includes: the difference between the first power generation revenue and the first maintenance cost;

[0105] The primary revenue from electricity generation is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price;

[0106] The primary maintenance cost is related to the rotation angle of the photovoltaic tracking bracket.

[0107] In this embodiment of the disclosure, the first constraint condition includes at least one of the following: a constraint condition on the rotation rate of the photovoltaic tracking bracket and a constraint condition on the operating temperature of the photovoltaic tracking bracket. For example, the rotation rate of the photovoltaic tracking bracket satisfies ≤ 3°C per minute and the operating temperature of the photovoltaic tracking bracket is ≤ 85°C. The rotation rate can be calculated based on the individuals in the current population, expressed by the following formula:

[0108]

[0109] The influence of the photovoltaic tracking bracket's operating temperature T(t) on the first irradiance curve, ambient temperature, and the bracket's rotation angle can be calculated based on experimental data. For example, the formula for the photovoltaic tracking bracket's operating temperature T(t) is expressed as follows:

[0110] T(t) = T amb (t)+α·G(t)·cos(θ(t))+β·G(t)·cos 2 (θ(t))

[0111] Here, G(t) represents irradiance, which is the main source of heating for the support. The higher the irradiance, the higher the support temperature, which can be obtained from the first irradiance curve. amb (t) represents the ambient temperature, which can be predicted based on historical ambient temperature data. α and β represent experimentally determined coefficients reflecting the influence of solar radiation on the support structure's temperature. The rotation angle θ(t) affects the angle at which the support structure receives solar radiation, which is determined by cos(θ(t)) and cos... 2 The term (θ(t)) reflects the effect of the incident angle of radiation on temperature.

[0112] The first objective function includes the difference between the first power generation revenue and the first maintenance cost, expressed by the formula:

[0113]

[0114] Where R represents the income indicator, P elec η(t) represents the electricity price at time t, η(t) represents the photoelectric conversion efficiency at time t, G(t) represents the irradiance at time t, S represents the area of ​​the solar panel, Δt represents the time step, and C main (Δθ) represents the initial maintenance cost resulting from the change in rotation angle Δθ. P elec The photoelectric conversion efficiency η(t) can be obtained from the first electricity price curve. The photoelectric conversion efficiency η(t) is related to factors such as the solar incidence angle and temperature, and can be determined experimentally or through empirical formulas. For example, the formula for photoelectric conversion efficiency is expressed as:

[0115] η(t)=η0·(1-β·(T(t)-T0))

[0116] Where η0 represents the standard efficiency, β represents the temperature coefficient, T(t) represents the temperature, and T0 represents the standard temperature. Irradiance G(t) can be obtained from the first irradiance curve. The solar panel area S can be determined based on the actual photovoltaic system design. The time step Δt can be consistent with the release cycle of real-time electricity price data. First maintenance cost C main (Δθ) is related to the change in rotation angle Δθ, and the formula is expressed as:

[0117] C main =k1·∑|Δθ|+k2·∑(Δθ) 2

[0118] Among them, k1 and k2 are coefficients that can be determined through accelerated life testing, such as 100,000 rotational cycles.

[0119] In another embodiment of this disclosure, the meteorological observation data includes cloud data monitored in real time by meteorological radar;

[0120] In step S106 above, the second turning angle curve is fine-tuned within the second constraint conditions based on the second electricity price curve, and the photovoltaic tracking bracket is controlled to rotate based on the fine-tuned second turning angle curve, including the following steps:

[0121] Step 1: Correct the first irradiance curve based on cloud data to determine the second irradiance curve within the second preset time range in the future;

[0122] Step 2: Based on the second turning angle curve, randomly generate the third turning angle curve of the photovoltaic tracking bracket changing with time within the second constraint condition as an individual. Combine multiple individuals into an initial population. Use the NSGA-II algorithm to perform multiple iterations. In each iteration, screen the current population to obtain a new population that meets the second constraint condition. Input the new population, the second irradiance curve and the second electricity price curve into the second objective function to obtain the income index. The individual with the highest income index in the final generation is determined as the fourth turning angle curve.

[0123] Step 3: Control the rotation of the photovoltaic tracking bracket according to the fourth turning angle curve;

[0124] The second constraint includes the constraint on the range of rotation angle variation of the photovoltaic tracking bracket;

[0125] The second objective function includes: the difference between the second power generation revenue and the second maintenance cost;

[0126] The second source of electricity revenue is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price;

[0127] The second maintenance cost is related to the rotation angle of the photovoltaic tracking bracket.

[0128] In this embodiment, the second irradiance curve is corrected using real-time electricity price data combined with the NSGA-II algorithm. For step 1 above, the meteorological observation data includes cloud data monitored in real-time by meteorological radar. Meteorological changes are random and may change due to uncertainties. To obtain a more accurate irradiance curve, real-time cloud data monitored by meteorological radar is introduced to predict irradiance within a future second preset time range, thereby correcting the first irradiance curve. The corrected first irradiance curve is then determined as the second irradiance curve within the future second preset time range. The real-time cloud data monitored by meteorological radar includes: cloud location and extent, cloud thickness, cloud texture and boundaries, cloud reflectivity, cloud movement speed and direction, and cloud development and dissipation processes. For step 2 above, the NSGA-II algorithm is used to select the optimal solution that meets the constraints. The second constraint may include a constraint on the range of rotation angle change of the photovoltaic tracking bracket. For example, the constraint requires the rotation angle change of the photovoltaic tracking bracket to be less than 15°, expressed by the formula: |Δθ|≤15°. Based on the second turning curve, a third turning curve representing the change in the turning angle of the photovoltaic tracking bracket over time is randomly generated within the second constraint conditions. Multiple individuals are then grouped into an initial population. The NSGA-II algorithm is used for multiple iterations, performing the following operations in each iteration until a preset number of iterations is reached. The individual with the highest income index in the final generation is determined as the fourth turning curve. The time span of the fourth turning curve is a first preset time.

[0129] Step 1: Input each individual in the population into the second constraint condition for screening, and obtain a new population that meets the second constraint condition. For example, compare each individual in the population with the second turning curve, and select individuals whose turning angle change is less than 15° at the same time to form a new population.

[0130] Step 2: Input each individual in the new population, the second irradiance curve, and the second electricity price curve into the second objective function to obtain the income indicator corresponding to each individual. The second objective function includes: the difference between the second power generation revenue and the second maintenance cost. The second power generation revenue is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price. The second maintenance cost is related to the rotation angle of the photovoltaic tracking bracket. For example, the formula of the second objective function is expressed as:

[0131] Revenue indicator = Secondary power generation revenue - Secondary maintenance cost;

[0132] Secondary power generation revenue = ∑ t (P elec (t)·P gen (t));

[0133] Second maintenance cost = f(Δθ) total )=c·Δθ total , Δθ total =∑ t |θ(t)-θ(t-1)|;

[0134] Where, Δθ total P represents the cumulative change in steering angle; c represents the maintenance cost coefficient per unit change in steering angle. elec (t) represents the electricity price at time t, which can be obtained from the second electricity price curve; power generation P gen The formula for (t) is expressed as:

[0135] P gen (t)=G(t)·η(t)·S·(1-Loss)

[0136] Where G(t) represents the irradiance, which can be obtained from the second irradiance curve; S represents the area of ​​the solar panel; Loss represents the photovoltaic system loss; η(t) represents the photoelectric conversion efficiency at time t, which is related to the rotation angle θ(t) and can be approximated linearly as follows:

[0137] η(t) = η0 + k·θ(t)

[0138] Where η0 represents the standard efficiency, and k represents the linear coefficient between efficiency and rotation angle.

[0139] Substitute η(t) into the power generation P gen The formula (t) is expressed as:

[0140] P gen (t)=G(t)·(η0+k·θ(t))·S·(1-Loss)

[0141] After simplification, we get: P gen (t)=a(t)+b(t)·θ(t);

[0142] a(t)=G(t)·η0·S·(1-Loss), b(t)=G(t)·k·S·(1-Loss);

[0143] P gen Substituting (t) into the second power generation revenue formula, it can be expressed as:

[0144] Secondary power generation revenue = ∑ t (P elec (t)·(a(t)+b(t)·θ(t)))

[0145] Step 3: Sort each individual in the new population according to the size of the income index to obtain a preset number of optimal individuals, and use the population composed of the preset number of optimal individuals as the parent population.

[0146] Step 4: Perform a crossover operation on individuals in the parent population to generate intermediate individuals. For example, perform a crossover operation on individuals in the parent population, such as randomly selecting two individuals from the parent population, denoted as P1 and P2, and perform the crossover operation to generate two intermediate individuals, denoted as C1 and C2. The crossover operation can be represented as:

[0147] C1=0.5·(P1+P2)+0.5·β·(P1-P2)

[0148] C2=0.5·(P1+P2)-0.5·β·(P1-P2)

[0149] Where β is the cross-distribution parameter.

[0150] Step 5: Perform a mutation operation on the intermediate individual to generate offspring individuals. For example, performing a mutation operation on the intermediate individual to introduce randomness and generate offspring individuals can be expressed by the following formula:

[0151] θ′(t)=θ(t)+δ

[0152] Where δ is a random perturbation that follows a multinomial distribution.

[0153] Step Six: Determine the combination of individuals from the parent population and offspring individuals to form a new population, and proceed to the next iteration. Regarding Step 3 above, control the rotation of the photovoltaic tracking bracket according to the fourth turning angle curve. The horizontal axis of the fourth turning angle curve represents time, and the vertical axis represents the turning angle of the photovoltaic tracking bracket. Through the above steps, the optimization method based on the NSGA-II algorithm can correct the second turning angle curve, generating the optimal turning angle curve that satisfies the constraints. Predicting future electricity prices based on day-ahead and real-time electricity price data can improve prediction accuracy, thereby enabling more accurate control of the photovoltaic tracking bracket.

[0154] In another embodiment of this disclosure, the meteorological observation data includes cloud data monitored in real time by meteorological radar;

[0155] When the price change within a future second preset time range, as represented by the second electricity price curve, exceeds a preset threshold, step S106 above involves fine-tuning the second turning curve within the second constraint based on the second electricity price curve, and controlling the rotation of the photovoltaic tracking bracket based on the fine-tuned second turning curve, including:

[0156] Step 1: Correct the first irradiance curve based on cloud data to determine the third irradiance curve within the second preset time range in the future;

[0157] Step 2: Based on the second turning curve, randomly generate the fifth turning curve of the photovoltaic tracking bracket changing with time within the second constraint condition as an individual. Combine multiple individuals into the first generation population. Use the NSGA-II algorithm to perform multiple iterations. In each iteration, screen the current population to obtain a new population that meets the third constraint condition. Input the new population, the third irradiance curve and the second electricity price curve into the third objective function to obtain the income index. The individual with the highest income index in the final generation is determined as the sixth turning curve.

[0158] Step 3: Control the rotation of the photovoltaic tracking bracket according to the sixth turning angle curve;

[0159] The second constraint includes the constraint on the range of rotation angle variation of the photovoltaic tracking bracket;

[0160] The third objective function includes: the difference between the third generation revenue and the third maintenance cost plus the sum of shadow revenue;

[0161] The third type of electricity revenue is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price;

[0162] The third maintenance cost is related to the rotation angle of the photovoltaic tracking bracket;

[0163] Shadow income is related to the preset shadow coefficient, the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price.

[0164] In this embodiment, when the predicted electricity price change is large, shadow revenue is introduced to strengthen the control of the photovoltaic tracking bracket. The second electricity price curve represents the price change within a future second preset time range that exceeds a preset threshold. For example, the horizontal axis time coordinate of the second electricity price curve is divided into multiple time periods, and the price change corresponding to each adjacent time period exceeds the preset threshold. Regarding step 1 above, the meteorological observation data includes cloud data monitored in real-time by meteorological radar. Meteorological changes are random and may change due to some uncertain factors. To obtain a more accurate irradiance curve, real-time cloud data monitored by meteorological radar is introduced to predict the irradiance within the future second preset time range, thereby correcting the first irradiance curve. The corrected first irradiance curve is then determined as the third irradiance curve within the future second preset time range. The cloud data monitored in real-time by meteorological radar includes: cloud location and extent, cloud thickness, cloud texture and boundaries, cloud reflectivity, cloud movement speed and direction, and cloud development and dissipation processes. Regarding step 2 above, the NSGA-II algorithm is used to select the optimal solution that satisfies the constraints. The second constraint can include a constraint on the range of rotation angle variation of the photovoltaic tracking bracket. For example, the constraint requires the rotation angle variation of the photovoltaic tracking bracket to be less than 15°, expressed by the formula: |Δθ|≤15°. Based on the second rotation angle curve, a fifth rotation angle curve is randomly generated within the second constraint, representing an individual. Multiple individuals are then grouped into an initial population. The NSGA-II algorithm is used for multiple iterations, performing the following operations in each iteration until a preset number of iterations is reached. The individual with the highest income index in the final generation is then determined as the sixth rotation angle curve.

[0165] Step 1: Input each individual in the population into the second constraint condition for screening, and obtain a new population that meets the second constraint condition. For example, compare each individual in the population with the second turning curve, and select individuals whose turning angle change is less than 15° at the same time to form a new population.

[0166] Step 2: Input each individual in the new population, the second irradiance curve, and the second electricity price curve into the third objective function to obtain the income indicator corresponding to each individual. The third objective function includes: the difference between the third power generation income and the third maintenance cost plus the sum of shadow income. The second power generation income is related to the photovoltaic tracking bracket rotation angle, irradiance, and electricity price. The second maintenance cost is related to the photovoltaic tracking bracket rotation angle. The shadow income is related to the preset shadow coefficient, the photovoltaic tracking bracket rotation angle, irradiance, and electricity price. For example, the formula of the third objective function is expressed as:

[0167] Revenue indicator = Third-party power generation revenue - Third-party maintenance costs + Shadow revenue;

[0168] Third-generation revenue = ∑t (P elec (t)·P gen (t));

[0169] Third maintenance cost = f(Δθ) total )=c·Δθ total , Δθ total =∑ t |θ(t)-θ(t-1)|;

[0170] Shadow income = λ·ShadowIncome;

[0171] Where λ represents the preset shadow coefficient, used to adjust the response strength to electricity price fluctuations; ShadowIncome represents the potential generation revenue resulting from adjusting the angle; Δθ total P represents the cumulative change in steering angle; c represents the maintenance cost coefficient per unit change in steering angle. elec (t) represents the electricity price at time t, which can be obtained from the second electricity price curve; power generation P gen The formula for (t) is expressed as:

[0172] P gen (t)=G(t)·η(t)·S·(1-Loss)

[0173] Where G(t) represents the irradiance, which can be obtained from the second irradiance curve; S represents the area of ​​the solar panel; Loss represents the photovoltaic system loss; η(t) represents the photoelectric conversion efficiency at time t, which is related to the rotation angle θ(t) and can be approximated linearly as follows:

[0174] η(t) = η0 + k·θ(t)

[0175] Where η0 represents the standard efficiency, and k represents the linear coefficient between efficiency and rotation angle.

[0176] Substitute η(t) into the power generation P gen The formula (t) is expressed as:

[0177] P gen (t)=G(t)·(η0+k·θ(t))·S·(1-Loss)

[0178] After simplification, we get: P gen (t)=a(t)+b(t)·θ(t);

[0179] a(t)=G(t)·η0·S·(1-Loss), b(t)=G(t)·k·S·(1-Loss);

[0180] P genSubstituting (t) into the second power generation revenue formula, it can be expressed as:

[0181] Secondary power generation revenue = ∑ t (P elec (t)·(a(t)+b(t)·θ(t)))

[0182] The potential revenue from power generation resulting from adjusting the rotation angle can be expressed as:

[0183] ShadowIncome=∑ t (P elec (t)·(a(t)+b(t)·θ(t)-P genref (t)))

[0184] Among them, P genref (t) represents the reference power generation.

[0185] The simplified formula for the third objective function is as follows:

[0186] Income indicator = (1+λ)·(∑ t (P elec (t)·a(t))+∑ t (P elec (t)·b(t)·θ(t)))-λ·∑ t (P elec (t)·P genref (t))-c·∑ t |θ(t)-θ(t-1)|

[0187] Step 3: Sort each individual in the new population according to the size of the income index to obtain a preset number of optimal individuals, and use the population composed of the preset number of optimal individuals as the parent population.

[0188] Step 4: Perform a crossover operation on individuals in the parent population to generate intermediate individuals. For example, perform a crossover operation on individuals in the parent population, such as randomly selecting two individuals from the parent population, denoted as P1 and P2, and perform the crossover operation to generate two intermediate individuals, denoted as C1 and C2. The crossover operation can be represented as:

[0189] C1=0.5·(P1+P2)+0.5·β·(P1-P2)

[0190] C2=0.5·(P1+P2)-0.5·β·(P1-P2)

[0191] Where β is the cross-distribution parameter.

[0192] Step 5: Perform a mutation operation on the intermediate individual to generate offspring individuals. For example, performing a mutation operation on the intermediate individual to introduce randomness and generate offspring individuals can be expressed by the following formula:

[0193] θ′(t)=θ(t)+δ

[0194] Where δ is a random perturbation that follows a multinomial distribution.

[0195] Step Six: Determine the combination of individuals from the parent population and offspring individuals as a new population, and proceed to the next iteration. Regarding Step 3 above, control the rotation of the photovoltaic tracking bracket according to the sixth turning angle curve. The horizontal axis of the sixth turning angle curve represents time, and the vertical axis represents the turning angle of the photovoltaic tracking bracket. The time span of the sixth turning angle curve is the first preset time. Through the above steps, when the predicted second electricity price curve fluctuates, the optimization method based on the NSGA-II algorithm can correct the second turning angle curve, generating the optimal turning angle curve that meets the constraints. By predicting future electricity prices based on day-ahead and real-time electricity price data, the accuracy of prediction can be improved, thereby enabling more accurate control of the photovoltaic tracking bracket.

[0196] In another embodiment of this disclosure, the meteorological observation data includes: numerical weather prediction data and sky imager data;

[0197] In step S102 above, predicting the irradiance within a first preset time range based on meteorological observation data and determining the first irradiance curve that varies with time includes:

[0198] Based on numerical weather forecast data and sky imager data, predict the irradiance within the first preset time range in the future, and determine the first irradiance curve that changes irradiance over time.

[0199] In this embodiment, numerical weather prediction (NWP) data and sky imager data are used to predict irradiance. Numerical Weather Prediction (NWP) is a method of predicting future weather using mathematical models and computer simulations. Based on atmospheric physics equations, it calculates using initial and boundary conditions to predict weather for the next few hours to days, providing high spatiotemporal resolution forecasts suitable for various applications. A sky imager is a device used to capture sky images and analyze cloud distribution, type, and movement. It is commonly used for solar energy forecasting and meteorological observation. It captures sky images using fisheye or wide-angle lenses, uses image processing techniques to identify clouds, the sun's position, etc., analyzes cloud movement, thickness, and type, and predicts cloud changes. This is used to predict the impact of clouds on solar radiation, optimize photovoltaic power generation, and provide high spatiotemporal resolution cloud information with strong real-time performance. Combining NWP data and sky imager data—using NWP data to provide large-scale weather trends and sky imager data to provide local cloud information—helps to correct NWP forecasts and improves the accuracy of solar energy forecasting. The prediction is made using machine learning models, such as random forests and neural networks, or physical models, such as radiative transfer models. Input data includes weather parameters from the National Weather Service (NWP), cloud information from a sky imager, and historical irradiance data. The output is the irradiance value within a first predetermined time frame. This predicts the irradiance within the first predetermined time frame and determines a first irradiance curve that reflects the change in irradiance over time. The first irradiance curve is a curve showing the change in irradiance over time, with time on the horizontal axis and irradiance on the vertical axis, reflecting the dynamic changes in irradiance within the first predetermined time frame, for example, 24 hours.

[0200] In another embodiment of this disclosure, step S103 above, predicting the electricity price within a first preset time range based on current-day electricity price data, and determining a first electricity price curve that changes over time, includes:

[0201] Step 1: Input the day-ahead electricity price data into the ARIMA model to determine the first residual sequence;

[0202] Step 2: Input the first residual sequence into the LSTM model to predict the electricity price within the first preset time range in the future, and determine the first electricity price curve that changes with time;

[0203] In step S105 above, predicting the electricity price within a second preset time range based on real-time electricity price data and determining the second electricity price curve that changes over time includes:

[0204] Step 3: Input real-time electricity price data into the ARIMA model to determine the second residual sequence;

[0205] Step 4: Input the second residual sequence into the LSTM model to predict the electricity price within the second preset time range in the future, and determine the second electricity price curve that changes with time.

[0206] In this embodiment, ARIMA and LSTM models are used to predict electricity prices. Electricity prices exhibit seasonality and randomness; for example, they fluctuate cyclically within a day, week, or year and are influenced by various factors such as weather, supply and demand, and policies, exhibiting complex nonlinear relationships. The ARIMA (Autoregressive Integrated Moving Average) model is a linear model used to handle non-stationary time series, capturing the linear components of electricity prices, such as seasonality. LSTM (Long Short-Term Memory) is suitable for handling long-term dependencies and nonlinear features in time series, capturing the nonlinear components of electricity prices, such as complex fluctuations and randomness. The ARIMA and LSTM models are combined: first, the ARIMA model is used to extract linear features, and then the LSTM model is used to model the residuals (nonlinear part). For step 1 above, the day-ahead electricity price data is input into the ARIMA model, which differs the day-ahead electricity price data to make it stationary. The ARIMA model is fitted to predict the linear part. The difference between the original day-ahead electricity price data and the ARIMA predicted value is extracted to obtain the first residual sequence. Regarding step 2 above, the first residual sequence is input into the trained LSTM model to predict the electricity price within a first preset time range in the future, thus determining the first electricity price curve that changes with time. The first electricity price prediction curve is a curve showing the electricity price changing over time, with time on the horizontal axis and electricity price on the vertical axis, reflecting the dynamic changes in electricity prices within the first preset time range in the future, for example, 24 hours. Regarding step 3 above, real-time electricity price data is input into the ARIMA model. The ARIMA model differs the real-time electricity price data to make it stable. The ARIMA model is fitted to predict the linear portion. The difference between the original real-time electricity price data and the ARIMA predicted value is extracted to obtain the second residual sequence. Regarding step 2 above, the second residual sequence is input into the trained LSTM model to predict the electricity price within a second preset time range in the future, thus determining the second electricity price curve that changes with time. The second electricity price prediction curve is a curve showing the electricity price changing over time, with time on the horizontal axis and electricity price on the vertical axis, reflecting the dynamic changes in electricity prices within the second preset time range in the future, for example, 4 hours. The second preset time is shorter than the first preset time. Combining day-ahead electricity price data and real-time electricity price data to predict future electricity prices can improve the accuracy of predictions.

[0207] In another embodiment of this disclosure, the time period of the change of the rotation angle of the photovoltaic tracking bracket in the first rotation curve is the same as the time period of the release of real-time electricity price data.

[0208] The second turning curve is used to control the rotation of the photovoltaic tracking bracket within the first preset time range in the future.

[0209] In this embodiment, real-time electricity price data refers to pricing data in the electricity market that reflects real-time changes in electricity prices based on supply and demand. The release of real-time electricity price data is periodic, typically every hour or 15 minutes, reflecting the real-time supply and demand situation of the power system. To accommodate the periodicity of real-time electricity price data release, the time period of the photovoltaic tracking bracket's rotation angle in the first turning curve is the same as the release time period of the real-time electricity price data. To accommodate the time span of the first electricity price curve and the first irradiance curve in predicting electricity prices and irradiance, the time span of the second turning curve is a first preset time, used to control the rotation of the photovoltaic tracking bracket within the future first preset time range.

[0210] Based on the same disclosed concept, this disclosure also provides a control device for a photovoltaic tracking bracket. Since the principle of solving the problem by these devices is similar to that of the aforementioned control method for photovoltaic tracking brackets, the implementation of this device can refer to the implementation of the aforementioned method, and the repeated parts will not be described again.

[0211] This disclosure provides a control device for a photovoltaic tracking bracket, such as... Figure 3 As shown, it includes:

[0212] Data acquisition module 301 is used to acquire meteorological observation data, day-ahead electricity price data and real-time electricity price data;

[0213] The day-ahead prediction module 302 is used to predict the irradiance within a first preset time range in the future based on the meteorological observation data, and determine the first irradiance curve that changes with time; predict the electricity price within the first preset time range in the future based on the day-ahead electricity price data, and determine the first electricity price curve that changes with time; randomly generate the first angle curve that changes with time for the rotation angle of the photovoltaic tracking bracket as an individual, form an initial population from multiple individuals, perform multiple iterations using the NSGA-II algorithm, screen the current population in each iteration to obtain a new population that meets the first constraint condition, input the new population, the first irradiance curve and the first electricity price curve into the first objective function to obtain the income index, and determine the individual with the highest income index in the final generation as the second angle curve;

[0214] The real-time correction module 302 is used to predict the electricity price within a second preset time range in the future based on the real-time electricity price data, determine the second electricity price curve that changes with time, fine-tune the second turning curve within the second constraint conditions based on the second electricity price curve, and control the rotation of the photovoltaic tracking bracket based on the fine-tuned second turning curve.

[0215] In another embodiment of this disclosure, the day-ahead prediction module 302 is used to take the first angle curve of the randomly generated photovoltaic tracking bracket changing with time as an individual, form an initial population of multiple individuals, and perform multiple iterations using the NSGA-II algorithm. In each iteration, the following operation is performed until the preset number of iterations is reached and the iteration stops. The individual with the highest income index in the final generation is determined as the second angle curve:

[0216] Each individual in the population is input into the first constraint for screening, resulting in a new population that satisfies the first constraint.

[0217] Each individual in the new population, the first irradiance curve, and the first electricity price curve are input into the first objective function to obtain the income indicator corresponding to each individual;

[0218] The individuals in the new population are sorted according to the size of the income index to obtain a preset number of optimal individuals. The population composed of the preset number of optimal individuals is used as the parent population.

[0219] Perform a crossover operation on the individuals in the parent population to generate intermediate individuals;

[0220] The intermediate individuals are mutated to generate offspring individuals;

[0221] The combination of individuals from the parent population and the offspring individuals is determined as a new population, and the next iteration begins.

[0222] In another embodiment of this disclosure, the first constraint condition includes at least one of the following: a constraint condition on the rotational speed of the photovoltaic tracking bracket and a constraint condition on the operating temperature of the photovoltaic tracking bracket.

[0223] The first objective function includes: the difference between the first power generation revenue and the first maintenance cost;

[0224] The first revenue from electricity generation is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price;

[0225] The first maintenance cost is related to the rotation angle of the photovoltaic tracking bracket.

[0226] In another embodiment of this disclosure, the meteorological observation data includes cloud data monitored in real time by meteorological radar;

[0227] The real-time correction module 302 is used to correct the first irradiance curve based on the cloud data and determine the second irradiance curve within a future second preset time range.

[0228] Based on the second turning curve, a third turning curve of the photovoltaic tracking bracket changing with time is randomly generated within the second constraint condition as an individual. Multiple individuals are combined into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the current population is screened to obtain a new population that meets the second constraint condition. The new population, the second irradiance curve, and the second electricity price curve are input into the second objective function to obtain the income index. The individual with the highest income index in the final generation is determined as the fourth turning curve.

[0229] The rotation of the photovoltaic tracking bracket is controlled according to the fourth turning angle curve;

[0230] The second constraint includes: the constraint on the range of rotation angle of the photovoltaic tracking bracket;

[0231] The second objective function includes: the difference between the second power generation revenue and the second maintenance cost;

[0232] The second revenue from electricity generation is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price;

[0233] The second maintenance cost is related to the rotation angle of the photovoltaic tracking bracket.

[0234] In another embodiment of this disclosure, the meteorological observation data includes cloud data monitored in real time by meteorological radar;

[0235] When the change in electricity price within a future second preset time range, as represented by the second electricity price curve, exceeds a preset threshold, the real-time correction module 302 is used to correct the first irradiance curve based on the cloud data to determine a third irradiance curve within a future second preset time range.

[0236] Based on the second turning curve, a fifth turning curve of the photovoltaic tracking bracket changing with time is randomly generated within the second constraint condition as an individual. Multiple individuals are combined into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the current population is screened to obtain a new population that meets the third constraint condition. The new population, the third irradiance curve and the second electricity price curve are input into the third objective function to obtain the income index. The individual with the highest income index in the final generation is determined as the sixth turning curve.

[0237] The rotation of the photovoltaic tracking bracket is controlled according to the sixth turning angle curve;

[0238] The second constraint includes: the constraint on the range of rotation angle of the photovoltaic tracking bracket;

[0239] The third objective function includes: the difference between the third power generation revenue and the third maintenance cost plus the sum of shadow revenue;

[0240] The third type of electricity revenue is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price;

[0241] The third maintenance cost is related to the rotation angle of the photovoltaic tracking bracket;

[0242] The shadow revenue is related to the preset shadow coefficient, the rotation angle of the photovoltaic tracking bracket, the irradiance, and the electricity price.

[0243] In another embodiment of this disclosure, the meteorological observation data includes: numerical weather prediction data and sky imager data;

[0244] The day-ahead prediction module 302 is used to predict the irradiance within a first preset time range in the future based on the numerical weather forecast data and sky imager data, and to determine the first irradiance curve of irradiance changing with time.

[0245] In another embodiment of this disclosure, the day-ahead forecasting module 302 is used to input the day-ahead electricity price data into the ARIMA model to determine the first residual sequence;

[0246] The first residual sequence is input into the LSTM model to predict the electricity price within the first preset time range in the future, and the first electricity price curve is determined as the electricity price changes over time.

[0247] The real-time correction module 302 is used to input the real-time electricity price data into the ARIMA model to determine the second residual sequence;

[0248] The second residual sequence is input into the LSTM model to predict the electricity price within a second preset time range in the future, thereby determining the second electricity price curve that changes with time.

[0249] In another embodiment of this disclosure, the time period of the change of the rotation angle of the photovoltaic tracking bracket in the first rotation curve is the same as the time period of the release of the real-time electricity price data.

[0250] The second rotation curve is used to control the rotation of the photovoltaic tracking bracket within a first preset time range in the future. Based on the same disclosed concept, embodiments of this disclosure provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the photovoltaic tracking bracket control method as described in any of the above embodiments.

[0251] Through the above description of the embodiments, those skilled in the art can clearly understand that the embodiments of this disclosure can be implemented in hardware or by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions of the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) and includes several instructions to cause a computer device (such as a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of this disclosure.

[0252] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes in the drawings are not necessarily essential for implementing this disclosure.

[0253] Those skilled in the art will understand that the modules in the apparatus of the embodiments can be distributed in the apparatus of the embodiments as described in the embodiments, or they can be located in one or more devices different from this embodiment with corresponding changes. The modules of the above embodiments can be combined into one module, or they can be further divided into multiple sub-modules.

[0254] The sequence numbers of the embodiments disclosed above are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0255] Obviously, those skilled in the art can make various modifications and variations to this disclosure without departing from its spirit and scope. Therefore, if such modifications and variations fall within the scope of the claims of this disclosure and their equivalents, this disclosure is also intended to include such modifications and variations.

Claims

1. A control method for a photovoltaic tracking bracket, characterized in that, include: Acquire meteorological observation data, day-ahead electricity price data, and real-time electricity price data; the meteorological observation data includes cloud data monitored in real time by meteorological radar; Based on the meteorological observation data, predict the irradiance within a first preset time range in the future, and determine the first irradiance curve that changes irradiance over time; Based on the current day electricity price data, predict the electricity price within a first preset time range in the future, and determine a first electricity price curve that changes with time; A first rotation angle curve, randomly generated to reflect the rotation angle of a photovoltaic tracking bracket over time, is used as an individual. Multiple individuals are grouped into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the population is filtered to obtain a new population that satisfies the first constraint condition. The new population, the first irradiance curve, and the first electricity price curve are input into a first objective function to obtain an income index. The individual with the highest income index in the final generation is determined as the second rotation angle curve. The first constraint condition includes at least one of the following: a constraint condition on the rotation rate of the photovoltaic tracking bracket and a constraint condition on the operating temperature of the photovoltaic tracking bracket. The first objective function includes: the difference between the first power generation revenue and the first maintenance cost; the first power generation revenue is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price; the first maintenance cost is related to the rotation angle of the photovoltaic tracking bracket. Based on the real-time electricity price data, predict the electricity price within a second preset time range in the future, and determine a second electricity price curve that changes with time; The second rotation angle curve is fine-tuned within the second constraint conditions based on the second electricity price curve, and the rotation of the photovoltaic tracking bracket is controlled according to the fine-tuned second rotation angle curve, including: The first irradiance curve is corrected based on the cloud data to determine the second irradiance curve within a future second preset time range; Based on the second turning curve, a third turning curve of the photovoltaic tracking bracket changing with time is randomly generated within the second constraint condition as an individual. Multiple individuals are combined into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the current population is screened to obtain a new population that meets the second constraint condition. The new population, the second irradiance curve, and the second electricity price curve are input into the second objective function to obtain the income index. The individual with the highest income index in the final generation is determined as the fourth turning curve. The rotation of the photovoltaic tracking bracket is controlled according to the fourth turning angle curve; The second constraint includes: the constraint on the range of rotation angle of the photovoltaic tracking bracket; The second objective function includes: the difference between the second power generation revenue and the second maintenance cost; The second revenue from electricity generation is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price; The second maintenance cost is related to the rotation angle of the photovoltaic tracking bracket.

2. The method as described in claim 1, characterized in that, The first rotation curve, which is randomly generated and changes in the rotation angle of the photovoltaic tracking bracket over time, is used as an individual. Multiple individuals are combined into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the current population is screened to obtain a new population that meets the first constraint condition. The new population, the first irradiance curve, and the first electricity price curve are input into the first objective function to obtain the income index. The individual with the highest income index in the final generation is determined as the second rotation curve, including: The first rotation angle curve, which is randomly generated and varies with time, is taken as an individual. Multiple individuals are formed into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the following operation is performed until the preset number of iterations is reached and the iteration stops. The individual with the highest income index in the final generation is determined as the second rotation angle curve: Each individual in the population is input into the first constraint for screening, resulting in a new population that satisfies the first constraint. Each individual in the new population, the first irradiance curve, and the first electricity price curve are input into the first objective function to obtain the income indicator corresponding to each individual; The individuals in the new population are sorted according to the size of the income index to obtain a preset number of optimal individuals. The population composed of the preset number of optimal individuals is used as the parent population. Perform a crossover operation on the individuals in the parent population to generate intermediate individuals; The intermediate individuals are mutated to generate offspring individuals; The combination of individuals from the parent population and the offspring individuals is determined as a new population, and the next iteration begins.

3. The method as described in claim 1, characterized in that, When the price change within a second preset time range, as represented by the second electricity price curve, exceeds a preset threshold, the step of fine-tuning the second turning curve within the second constraint based on the second electricity price curve, and controlling the rotation of the photovoltaic tracking bracket based on the fine-tuned second turning curve, includes: The first irradiance curve is corrected based on the cloud data to determine the third irradiance curve within a future second preset time range; Based on the second turning curve, a fifth turning curve of the photovoltaic tracking bracket changing with time is randomly generated within the second constraint condition as an individual. Multiple individuals are combined into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the current population is screened to obtain a new population that meets the third constraint condition. The new population, the third irradiance curve and the second electricity price curve are input into the third objective function to obtain the income index. The individual with the highest income index in the final generation is determined as the sixth turning curve. The rotation of the photovoltaic tracking bracket is controlled according to the sixth turning angle curve; The second constraint includes: the constraint on the range of rotation angle of the photovoltaic tracking bracket; The third objective function includes: the difference between the third power generation revenue and the third maintenance cost plus the sum of shadow revenue; The third type of electricity revenue is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price; The third maintenance cost is related to the rotation angle of the photovoltaic tracking bracket; The shadow revenue is related to the preset shadow coefficient, the rotation angle of the photovoltaic tracking bracket, the irradiance, and the electricity price.

4. The method as described in claim 1, characterized in that, The meteorological observation data also includes: numerical weather prediction data and sky imager data; The step of predicting the irradiance within a first preset time range based on the meteorological observation data and determining the first irradiance curve as a function of time includes: Based on the numerical weather forecast data and sky imager data, predict the irradiance within a first preset time range in the future, and determine the first irradiance curve that changes irradiance over time.

5. The method as described in claim 1, characterized in that, The step of predicting future electricity prices within a first preset time range based on the current daytime electricity price data and determining a first electricity price curve that changes over time includes: The day-ahead electricity price data is input into the ARIMA model to determine the first residual sequence; The first residual sequence is input into the LSTM model to predict the electricity price within the first preset time range in the future, and the first electricity price curve is determined as the electricity price changes over time. The step of predicting the electricity price within a second preset time range based on the real-time electricity price data and determining the second electricity price curve that changes over time includes: The real-time electricity price data is input into the ARIMA model to determine the second residual sequence; The second residual sequence is input into the LSTM model to predict the electricity price within a second preset time range in the future, thereby determining the second electricity price curve that changes with time.

6. The method as described in claim 1, characterized in that, The time period of the photovoltaic tracking bracket rotation angle change over time in the first rotation angle curve is the same as the time period of the real-time electricity price data release. The second rotation curve is used to control the rotation of the photovoltaic tracking bracket within a first preset time range in the future.

7. A control device for a photovoltaic tracking bracket, characterized in that, include: The data acquisition module is used to acquire meteorological observation data, day-ahead electricity price data, and real-time electricity price data; the meteorological observation data includes cloud data monitored in real time by meteorological radar. The current-day forecasting module is used to predict the irradiance within a first preset time range based on the meteorological observation data, and determine a first irradiance curve that changes with time; predict the electricity price within the first preset time range based on the current-day electricity price data, and determine a first electricity price curve that changes with time; randomly generate a first angle curve that changes with time for the rotation angle of the photovoltaic tracking bracket as an individual, and form an initial population from multiple individuals. The NSGA-II algorithm is used for multiple iterations, and in each iteration, the current population is screened to obtain a new population that meets the first constraint condition. The new population, the first irradiance curve, and the first electricity price curve are input into a first objective function to obtain an income index. The individual with the highest income index in the final generation is determined as the second angle curve. The first constraint condition includes at least one of the following: a constraint condition on the rotation rate of the photovoltaic tracking bracket and a constraint condition on the operating temperature of the photovoltaic tracking bracket; the first objective function includes: the difference between the first power generation income and the first maintenance cost; the first power generation income is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price; the first maintenance cost is related to the rotation angle of the photovoltaic tracking bracket. A real-time correction module is used to predict the electricity price within a second preset time range based on the real-time electricity price data, determine a second electricity price curve that changes over time, fine-tune the second turning angle curve within a second constraint condition based on the second electricity price curve, and control the rotation of the photovoltaic tracking bracket based on the fine-tuned second turning angle curve, including: The first irradiance curve is corrected based on the cloud data to determine the second irradiance curve within a future second preset time range; Based on the second turning curve, a third turning curve of the photovoltaic tracking bracket changing with time is randomly generated within the second constraint condition as an individual. Multiple individuals are combined into an initial population. The NSGA-II algorithm is used for multiple iterations. In each iteration, the current population is screened to obtain a new population that meets the second constraint condition. The new population, the second irradiance curve, and the second electricity price curve are input into the second objective function to obtain the income index. The individual with the highest income index in the final generation is determined as the fourth turning curve. The rotation of the photovoltaic tracking bracket is controlled according to the fourth turning angle curve; The second constraint includes: the constraint on the range of rotation angle of the photovoltaic tracking bracket; The second objective function includes: the difference between the second power generation revenue and the second maintenance cost; The second revenue from electricity generation is related to the rotation angle of the photovoltaic tracking bracket, irradiance, and electricity price; The second maintenance cost is related to the rotation angle of the photovoltaic tracking bracket.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the control method for the photovoltaic tracking bracket as described in any one of claims 1 to 6.

Images