Method for dynamically adjusting green energy and load by using virtual power plant

Abstract

The invention discloses a method for dynamically adjusting green energy and load by using a virtual power plant, which comprises the following steps of: acquiring historical and real-time weather data, and processing the data; establishing a weather prediction model by using the preprocessed weather data, and predicting a future weather change condition by using an HHO algorithm; the energy storage device stores redundant energy in a low-load period, releases and balances supply and demand in a high-load period, energy supply and demand data are imported into the current-day big data module, and the current-day big data load module predicts the current-day energy demand load according to predicted weather data and historical load data; and transmitting the weather data prediction model and the data of the current day big data module to a comprehensive data receiving and processing unit. Energy supply and demand imbalance caused by sudden weather change is reduced, and the energy utilization efficiency is improved; the absorption capability and the operation efficiency of the power grid are obviously improved; scientific basis is provided for energy transaction, system operation cost is reduced, and flexibility and economical efficiency of market transaction are improved.

Description

Technical Field

[0001] The present invention relates to the field of distributed energy integration and market trading, and particularly to a method for dynamically adjusting green energy and load using a virtual power plant. Background Art

[0002] Traditional energy systems mainly rely on centralized power stations and fixed energy distribution models, making it difficult to adapt to the volatility and uncertainty of distributed energy resources (such as solar energy, wind energy, etc.). In addition, with the popularization of electric vehicles and the development of energy storage technologies, the flexibility and complexity on the energy demand side are also increasing continuously. These problems have led to the difficulty of spatio-temporal matching of energy supply and demand, restricted the efficient utilization of renewable energy, and at the same time increased the operating cost and carbon emission intensity of the power grid.

[0003] In the prior art, although some energy management systems can achieve basic load forecasting and scheduling functions, there are still obvious deficiencies in coping with the volatility of renewable energy and the dynamic changes in load demand. For example, traditional energy management systems usually lack the ability to anticipate weather changes in advance and cannot adjust energy supply and demand strategies in advance, resulting in a lag in response during sudden weather changes and low energy utilization efficiency. In addition, the existing systems have limited capabilities in integrating distributed energy resources (such as energy storage devices, electric vehicles, etc.) and optimizing market trading strategies, and it is difficult to achieve precise control and efficient utilization of energy. Therefore, it is necessary to propose an effective method for dynamically adjusting green energy and load to solve the deficiencies of traditional energy management systems in terms of precision, flexibility, and energy conservation. Summary of the Invention

[0004] Object of the Invention: The object of the present invention is to provide a method for dynamically adjusting green energy and load using a virtual power plant.

[0005] Technical Solution: The method described in the present invention includes the following steps:

[0006] S1: Collect historical and real-time weather data, covering key meteorological parameters such as temperature, humidity, and wind speed, and transmit the collected weather data to the weather data receiving and processing unit for data processing;

[0007] S2: Use the preprocessed weather data to establish a weather prediction model and adopt the HHO algorithm to predict future weather changes;

[0008] S3: The VPP virtual power plant forms a virtual power plant by integrating distributed energy resources. The controllable load adjusts the energy consumption behavior of users through demand response. The energy storage device stores excess energy during low load periods and releases it during high load periods to balance supply and demand;

[0009] S4: Import the energy supply and demand data in S3 into the big data module of the day. The big data load module of the day predicts the energy demand load of the day based on the predicted weather data and historical load data, and divides it into high-load periods and low-load periods;

[0010] S5: Transmit the weather data prediction model combined with the data of the big data module of the day to the comprehensive data reception and processing unit. The green electricity price trend prediction model predicts the trend of electricity prices based on the energy supply and demand situation and market mechanisms, providing a reference for energy trading.

[0011] Further, the weather data processing steps in step S1 are as follows:

[0012] S11: The weather data collection formula is as follows:

[0013]

[0014] pi = p

[0015]

[0016] Among them, the historical weather data searched by the weather data reception and processing unit is divided into k layers according to certain characteristics, each layer has N i units, n i units are extracted from the i-th layer, and the sampling ratio of the i-th layer is p i , and the overall sampling ratio is p;

[0017] S12: The weather data cleaning formula is as follows:

[0018]

[0019] Among them, x i is a variable, n is the total value of weather observations, μ is the mean value, and σ is the standard deviation;

[0020] S13: The weather data preprocessing formula is as follows:

[0021]

[0022] Among them, x is the original data screened by the weather data reception and processing unit, x min is the minimum value of extreme weather data, and x max is the maximum value of extreme weather data;

[0023] S14: Extract the key weather features from the preprocessed weather data. The weather feature extraction formula is as follows:

[0024]

[0025] H max= max(H1, H2, ..., H i )

[0026] where T a , W a , H max are the average temperature, average wind speed, and maximum humidity respectively. n1 and n2 are the total numbers of temperature observation values and wind speed observation values in the historical weather data respectively. T i is the i-th temperature observation value, W i is the i-th wind speed observation value, and H i is the i-th humidity observation value.

[0027] Furthermore, the weather prediction model in step S2 is constructed as follows:

[0028] S21: Input of the weather model; The input Y includes two parts. Y1 is the historical meteorological information for a past period, and Y2 is the prediction value of the supercomputer. For T consecutive moments (1, 2, …, T) in history, the corresponding weather index observation values are denoted as Y = (y 1 , y 2 , …, y T ), and the corresponding supercomputer prediction values are denoted as Among them, the meteorological information data y t corresponding to a certain moment t1 includes k weather index values, which can be expressed as

[0029] S22: Output of the weather model; It includes the weather index values corresponding to the future P consecutive moments, which can be expressed as Among them, the data at each moment also includes k weather index values;

[0030] S23: Define the weather prediction model:

[0031]

[0032] S24: The weather prediction objective function is as follows:

[0033] f(T, H, W) = w T ·T + w H ·H + w W ·W

[0034] where w T , w H , w W are the weight factors for temperature, humidity, and wind speed respectively.

[0035] Furthermore, in step S2, the HHO algorithm introduces the PWLCM chaotic mapping and the non-linear escape energy update strategy to ensure the global convergence efficiency.

[0036] Furthermore, the introduction of the PWLCM chaotic mapping:

[0037] The original HHO algorithm randomly generates a population in the population space, resulting in a lack of population quality and low optimization accuracy. Therefore, the PWLCM chaotic mapping is introduced to adjust the initialization parameter r of the HHO population. The adjustment formula is as follows:

[0038]

[0039] where r i is the initialization random parameter.

[0040] Furthermore, the introduction of the non-linear escape energy update strategy:

[0041] In the original HHO, the escape energy E is used to control the switching of the eagle group from the exploration stage to the exploitation stage. During the iteration process, E linearly decreases from 2 to 0; the linear decrease will lead to a decrease in the global search ability in the later stage of the algorithm iteration and increase the probability of falling into the local optimum. A new non-linear escape energy update strategy is introduced, and the calculation formula is as follows:

[0042]

[0043] E = E0×(2×rand - 1)

[0044] where T0 is the maximum number of iterations. The improved HHO algorithm will preferentially select individuals with higher fitness values and use them as seeds for the next generation, thereby accelerating the convergence speed of the algorithm and finally obtaining a better global optimal solution X best .

[0045] Furthermore, the calculation formula of the VPP virtual power plant in step S3 is as follows:

[0046] E total (t) = E RE (t) + E ESS (t) + E EV (t)

[0047] where E total (t) is the total energy output of the virtual power plant, E RE (t) is the power generation of the controllable load at time t, E ESS (t) is the energy output of the energy storage device at time t, E EV (t) is the energy output of the electric vehicle at time t;

[0048] Controllable load power generation E RE (t) The calculation formula is as follows:

[0049]

[0050] Among them, N RE is the type of controllable load, P RE，i (t) is the power output of the i-th controllable load at time t, η RE，i is the conversion efficiency of the i-th controllable load,

[0051] The energy output E ESS (t) of the energy storage device has the following calculation formula:

[0052] E ESS (t) = P ESS (t)·Δt·η ESS (t)

[0053] Among them, P ESS (t) is the charge-discharge power of the energy storage device at time t, Δt is the time interval, and η ESS (t) is the charge-discharge efficiency of the energy storage device at time t.

[0054] Furthermore, the calculation formula for the energy demand load in step S4 is as follows:

[0055]

[0056] Among them, L(t) is the energy demand load, N is the number of weather factors affecting the energy demand, α i is the weight coefficient of the i-th weather factor, W i (t) is the predicted value of the i-th weather factor at time t, β i is the weight coefficient of the i-th historical load factor, H i is the i-th historical load data, and γ is a comprehensive adjustment coefficient used to balance the influence of weather and historical load.

[0057] Furthermore, the division criteria for high load periods and low load periods are as follows:

[0058] Load average value L avg :

[0059]

[0060] Among them, T1 is the total number of time intervals in a day;

[0061] Set the load threshold L threshold :

[0062] L threshold = L avg·(1 + δ)

[0063] where δ represents the load fluctuation coefficient, and its value usually ranges from 0.1 to 0.3;

[0064] Divide the high-load period and the low-load period:

[0065] T high ={t | L(t) ≥ L threshold}

[0066] T low ={t | L(t) < L threshold}

[0067] where T high is the high-load period, and T low is the low-load period.

[0068] Furthermore, the calculation formula of the green electricity price trend prediction model in step S5 is as follows:

[0069]

[0070] where P price (t) is the predicted electricity price at time t, ζ is the weight coefficient of the energy supply-demand ratio, E total (t) is the total energy output at time t, η is the weight coefficient of the market mechanism, and P market (t) is the market base electricity price at time t.

[0071] Beneficial effects: Compared with the prior art, the present invention has the following remarkable advantages: By using the weather data prediction model and the HHO algorithm to optimize weather prediction, the present invention can anticipate in advance the impact of weather changes on renewable energy power generation, reduce the energy supply-demand imbalance caused by sudden weather changes, and improve energy utilization efficiency; By integrating distributed energy resources (such as energy storage devices, electric vehicles, etc.) through a virtual power plant, dynamically adjusting energy supply and demand, and optimizing the energy distribution in high-load and low-load periods, the present invention can significantly enhance the grid's accommodation capacity and operation efficiency; By using the green electricity price trend prediction model, combining energy supply-demand conditions and market mechanisms, the present invention can provide a scientific basis for energy trading, reduce system operation costs, and enhance the flexibility and economy of market trading. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 is the flowchart of the present invention;

[0073] Figure 2 is the flowchart of the HHO algorithm of the present invention;

[0074] Figure 3 is the optimization flowchart of the HHO algorithm of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0075] The technical solution of the present invention will be further described below in conjunction with the accompanying drawings.

[0076] As Figure 1 shown, the method for dynamically adjusting green energy and load using a virtual power plant according to the present invention includes the following steps:

[0077] S1: The weather data set is specifically responsible for collecting historical and real-time weather data, covering key meteorological parameters such as temperature, humidity, and wind speed. The collected weather data is then accurately transmitted to the weather data receiving and processing unit via a high-speed and stable transmission channel. In the weather data receiving and processing unit, a series of rigorous data processing is performed on the transmitted massive weather data for subsequent use by the weather prediction model;

[0078] The calculation formulas for weather data parameters are as follows:

[0079] The temperature calculation formula is as follows:

[0080]

[0081] where T(t, l) is the temperature at time t and location l, T0 is the average temperature in a year, A and B are the amplitude and phase shift coefficients of daily temperature variation, C and D are the amplitude and phase shift coefficients of annual temperature variation, E is the influence coefficient of air pressure P on temperature, F is the influence coefficient of humidity H on temperature, and G is the influence coefficient of wind speed W on temperature.

[0082] The humidity calculation formula is as follows:

[0083]

[0084] where H(t, l) is the humidity at time t and location l, H0 is the average humidity in a year, A H and B H are the amplitude and phase shift coefficients of daily humidity variation, C H and D H are the amplitude and phase shift coefficients of annual humidity variation, E H is the influence coefficient of air pressure P on humidity, F H is the influence coefficient of temperature T on humidity, G H is the influence coefficient of wind speed W on humidity.

[0085] The wind speed calculation formula is as follows:

[0086]

[0087] where W(t, l) is the wind speed at time t and location l, W0 is the average wind speed in a year, A W and B Wis the amplitude and phase shift coefficient of the daily wind speed variation, C W and D W are the amplitude and phase shift coefficient of the annual wind speed variation, E W is the influence coefficient of air pressure P on wind speed, F W is the influence coefficient of temperature T on wind speed, G W is the influence coefficient of humidity H on wind speed. The weather data processing steps are as follows:

[0088] S11: The weather data acquisition formula is as follows:

[0089]

[0090] pi = p

[0091]

[0092] Among them, the historical weather data searched by the weather data receiving and processing unit is divided into k layers according to certain characteristics, and each layer has N i units, n i units are extracted from the i-th layer, and the sampling ratio of the i-th layer is p i , and the overall sampling ratio is p.

[0093] S12: The weather data cleaning formula is as follows:

[0094]

[0095] Among them, x i is a variable, n is the total number of weather observations, μ is the mean value, and σ is the standard deviation.

[0096] S13: The weather data preprocessing formula is as follows:

[0097]

[0098] Among them, x is the original data screened by the weather data receiving and processing unit, x min is the minimum value of extreme weather data, x max is the maximum value of extreme weather data.

[0099] S14: Extract the key weather features from the preprocessed weather data, such as the average values of temperature and wind speed, and the maximum value of humidity, etc. The weather feature extraction formula is as follows:

[0100]

[0101] H max = max(H$_1$, H$_2$,..., H i )

[0102] Among them, T a, W a , H max are the average temperature, average wind speed, and maximum humidity respectively. n1 and n2 are the total numbers of temperature observation values and wind speed observation values in the historical weather data respectively. T i is the i-th temperature observation value, W i is the i-th wind speed observation value, H i is the i-th humidity observation value.

[0103] S2: Using the preprocessed weather data, establish a weather prediction model and adopt the HHO algorithm to predict the future weather changes.

[0104] The construction process of the weather prediction model is as follows:

[0105] S21: Input of the weather model:

[0106] The input Y includes two parts. Y1 is the historical meteorological information of the past period of time, and Y2 is the predicted value of the supercomputer. For T consecutive moments (1, 2,..., T) in history, the corresponding weather index observation values are recorded as Y = (y 1 , y 2 , …, y T ), and the corresponding predicted value of the supercomputer is recorded as Among them, the meteorological information data y t corresponding to a certain moment t1 contains k weather index values, which can be expressed as

[0107] S22: Output of the weather model:

[0108] The output is the meteorological information of the future period of time, including the weather index values corresponding to P consecutive moments in the future, which can be expressed as Among them, the data of each moment also contains k weather index values.

[0109] S23: Define the weather prediction model:

[0110]

[0111] S24: The weather prediction objective function is as follows:

[0112] f(T, H, W) = w T ·T + w H ·H + w W ·W

[0113] Among them, w T , w H , w W are the weight factors of temperature, humidity, and wind speed respectively.

[0114] S25: The specific steps of the HHO algorithm are as follows:

[0115] Exploration phase:

[0116] At the initial stage of HHO, the parameters of the eagle group are randomly generated, indicating that the eagle group is distributed at random positions. In the exploration phase, two strategies are used to discover the prey, and probabilities are used to determine the chance of adopting which strategy. Harris hawks perch according to the position of the prey or other individuals, and the position update formula is as follows:

[0117]

[0118] where t is the number of the current iteration, X(t + 1) is the position of the population individual in the next iteration, X rabbit (t) is the position of the prey, and X(t) is the position of the current population individual. In each iteration process, r1, r2, r3, r4, and q are updated to random values within (0, 1). LB and UB represent the upper and lower boundaries of the population individual positions. X rand (t) is a random individual from the current population, and X m is the average position of the current population. r3 is the scaling factor. The closer r4 is to 1, the more it will increase the randomness of the individual. The formula for calculating the average position of the population is as follows:

[0119]

[0120] where X i (t) represents the position of each individual in iteration t, and N represents the number of population individuals.

[0121] Exploitation phase:

[0122] The escape energy of the rabbit will gradually decrease during the pursuit by the eagle, and the eagle will also switch its exploitation behavior. The formula for calculating the rabbit escape energy is as follows:

[0123]

[0124] where E is the escape energy of the rabbit, T is the number of iterations of the algorithm, and E0 is the initial state of the rabbit energy. During the HHO iteration, E0 is randomly taken within the range of (1, 1). E0 approaching -1 indicates that the prey energy is exhausted, and E0 approaching 1 indicates that the prey energy is strong. The prey escape energy E decreases as the number of iterations increases. When |E| ≥ 1, the eagle is in the exploration phase, and when |E| < 1, it is in the surprise attack phase.

[0125] Surprise attack phase:

[0126] The hawk will attack the target detected in the previous stage, but the prey will try to escape when encountering the attack. Therefore, the hawk will choose several different hunting strategies. Based on the escape behavior of the prey and the hunting behavior of the hawk, four strategies are proposed in this stage to simulate the attack process.

[0127] 1. Soft Siege

[0128] When r≥0.5 and |E|≥0.5, the prey still has enough energy and uses random misleading jumps to escape, but finally fails. Because in this process, the Harris hawk approaches the prey gently, making the prey feel persistently tired, and then suddenly pounces on the prey. The calculation formula for this process is as follows:

[0129] X(t + 1) = ΔX(t) - E|JX rabbit (t) - X(t)|

[0130] ΔX(t) = X rabbit (t) - X(t)

[0131] Where ΔX(t) is the position difference between the hawk and the prey in the current iteration, J is a random number within (0, 2), which changes randomly in each iteration, simulating the random jump intensity of the prey.

[0132] 2. Hard Siege

[0133] When r≥0.5 and |E|<0.5, the prey has consumed a large amount of energy and is already in a state of exhaustion, with low escape energy. At the same time, the Harris hawk has almost surrounded the prey and finally completes the attack and capture. The calculation formula for this process is as follows:

[0134] X(t + 1) = X rabbit (t) - E|ΔX(t)|

[0135] 3. Soft Siege with Progressive Rapid Dive

[0136] When |E|≥0.5 but r<0.5, that is, when the prey has high escape energy, before the attack, the Harris hawk will adopt the soft siege strategy and perform misleading jump movements on the prey before implementing the attack. In the actual process, the Harris hawks will dive around the prey several times collectively, and adjust their positions and directions according to the misleading jump movements of the prey. The calculation formula for this process is as follows:

[0137]

[0138] Where Levy(D) is the Lévy flight random step length and S is the Lévy flight vector.

[0139] 4. Hard Siege with Progressive Rapid Dive

[0140] When |E| < 0.5 and r < 0.5, the prey's energy weakens, and the eagle can execute a surprise attack through a hard siege to capture the target. During this process, the distance between the average position of the population and the prey position is gradually reduced. The calculation formula for this process is as follows:

[0141]

[0142] Among them, X m (t) is the average position of the population under this strategy.

[0143] S26: Since the HHO algorithm relies on the escape energy mechanism for local search during the exploitation stage, when the prey's escape energy is low, the algorithm tends to perform fine exploitation, which may lead to a decrease in population diversity, an inability to effectively jump out of the local optimal solution, and the HHO contains multiple position update strategies, increasing the computational complexity and resulting in a longer execution time for the algorithm. Therefore, it is necessary to introduce the PWLCM chaotic mapping and the non-linear escape energy update strategy to ensure the global convergence efficiency.

[0144] Introduction of the PWLCM chaotic mapping:

[0145] The original HHO algorithm randomly generates the population in the population space, resulting in a lack of population quality and low optimization accuracy. Therefore, the PWLCM chaotic mapping is introduced to adjust the initialization parameter r of the HHO population, and the adjustment formula is as follows:

[0146]

[0147] Among them, r i is the initialization random parameter.

[0148] Introduction of the non-linear escape energy update strategy:

[0149] In the original HHO, the switch of the eagle group from the exploration stage to the exploitation stage is controlled according to the escape energy E. During the iteration process, E linearly decreases from 2 to 0. The linear decrease will lead to a decrease in the global search ability in the later stage of the algorithm iteration and increase the probability of falling into the local optimum. Therefore, a new non-linear escape energy update strategy is introduced, and the calculation formula is as follows:

[0150]

[0151] E = E0 × (2 × rand - 1)

[0152] Among them, T0 is the maximum number of iterations.

[0153] The improved HH0 algorithm will preferentially select individuals with higher fitness values and use them as seeds for the next generation, thereby accelerating the convergence speed of the algorithm and finally obtaining a better global optimal solution X best .

[0154] S3: The VPP virtual power plant forms a virtual power plant by integrating distributed energy resources (such as renewable energy, energy storage devices, and electric vehicles) to provide flexible energy supply. The controllable load adjusts the energy consumption behavior of users through means such as demand response to adapt to changes in energy supply and demand. The energy storage device stores excess energy during low load periods and releases it during high load periods to balance supply and demand. As part of the energy storage device, electric vehicles can charge during low load periods and discharge during high load periods to participate in energy balancing.

[0155] The calculation formula of the VPP virtual power plant is as follows:

[0156] E total (t) = E RE (t) + E ESS (t) + E EV (t)

[0157] Where, E total (t) is the total energy output of the virtual power plant, E RE (t) is the power generation of the controllable load at time t, E ESS (t) is the energy output of the energy storage device at time t, E EV (t) is the energy output of the electric vehicle at time t.

[0158] The power generation of the controllable load E RE (t) The calculation formula is as follows:

[0159]

[0160] Where, N RE is the type of the controllable load, P RE，i (t) is the power output of the i-th controllable load at time t, η RE，i is the conversion efficiency of the i-th controllable load.

[0161] The energy output of the energy storage device E ESS (t) The calculation formula is as follows:

[0162] E ESS (t) = P ESS (t)·Δt·η ESS (t)

[0163] Where, P ESS (t) is the charge and discharge power of the energy storage device at time t, Δt is the time interval, η ESS (t) is the charge and discharge efficiency of the energy storage device at time t.

[0164] The energy output of the electric vehicle E EV (t) The calculation formula is as follows:

[0165] E EV E(t) = P EV (t)·Δt·η EV (t)

[0166] Wherein, P EV (t) is the charging and discharging power of the electric vehicle at time t, and η EV (t) is the charging and discharging efficiency of the electric vehicle at time t.

[0167] S4: Import the energy supply and demand data described in S3 into the big data module of the day. The big data load module of the day predicts the energy demand load of the day based on the predicted weather data and historical load data, and divides it into high load periods and low load periods.

[0168] The calculation formula for the energy demand load L(t) is as follows:

[0169]

[0170] Wherein, N is the number of weather factors affecting energy demand, and α i is the weight coefficient of the i-th weather factor, W i (t) is the predicted value of the i-th weather factor at time t, and β i is the weight coefficient of the i-th historical load factor, H i is the i-th historical load data, and γ is a comprehensive adjustment coefficient used to balance the influence of weather and historical load.

[0171] The division criteria for high load periods and low load periods are as follows:

[0172] 1. Load average value L avg :

[0173]

[0174] In the formula, T1 is the total number of time intervals in a day.

[0175] 2. Set the load threshold L threshold :

[0176] L threshold = L avg ·(1 + δ)

[0177] In the formula, δ represents the load fluctuation coefficient, and usually takes values between 0.1 and 0.3.

[0178] 3. Divide high load periods and low load periods:

[0179] T high = {t | L(t) ≥ L threshold}

[0180] T low ={t | L(t) < L threshold}

[0181] In the formula, T high is the high - load period, and T low is the low - load period.

[0182] S5: Transmit the weather data prediction model and the data of the big data module of the current day to the integrated data receiving and processing unit together. The green electricity price trend prediction model predicts the trend of electricity prices according to the energy supply - demand situation and market mechanism, providing a reference for energy trading.

[0183] The calculation formula of the green electricity price trend prediction model is as follows:

[0184]

[0185] In the formula, P price (t) is the predicted electricity price at time t, ζ is the weight coefficient of the energy supply - demand ratio, E total (t) is the total energy output at time t, η is the weight coefficient of the market mechanism, and P market (t) is the market basic electricity price at time t.

Claims

1. A method for dynamically regulating green energy and load using a virtual power plant, characterized in that It includes the following steps: S1: Collect historical and real-time weather data, covering key meteorological parameters such as temperature, humidity, and wind speed, and transmit the collected weather data to the weather data receiving and processing unit for data processing; S2: Use the preprocessed weather data to establish a weather prediction model, adopt the HHO algorithm, and predict future weather changes; S3: The VPP virtual power plant forms a virtual power plant by integrating distributed energy resources. The controllable load adjusts the energy consumption behavior of users through demand response. The energy storage device stores excess energy during low load periods and releases it during high load periods to balance supply and demand. As part of the energy storage device, electric vehicles can charge during low load periods and discharge during high load periods to participate in energy balancing; S4: Import the energy supply and demand data in S3 into the daily big data module. The daily big data load module predicts the daily energy demand load based on the predicted weather data and historical load data, and divides it into high load periods and low load periods; S5: Transmit the weather data prediction model combined with the data in the daily big data module to the comprehensive data receiving and processing unit. The green electricity price trend prediction model predicts the trend of electricity prices based on the energy supply and demand situation and market mechanisms, providing a reference for energy trading.

2. The method for dynamically adjusting green energy and load using a virtual power plant according to claim 1, characterized in that, The weather data processing steps in step S1 are as follows: S11: The weather data collection formula is as follows: pi = p Among them, the weather data receiving and processing unit searches that the historical weather data is generally divided into k layers according to certain characteristics, and each layer has N i units. n i units are extracted from the i-th layer, and the sampling ratio of the i-th layer is p i , and the overall sampling ratio is p; S12: The weather data cleaning formula is as follows: where x i is a variable, n is the total value of weather observations, μ is the mean, and σ is the standard deviation; S13: The weather data preprocessing formula is as follows: Among them, x is the original data screened by the weather data receiving and processing unit, x min is the minimum value of extreme weather data, x max is the maximum value of extreme weather data; S14: Extract key weather features from the preprocessed weather data. The weather feature extraction formula is as follows: H max = max(H1, H2,..., H i ) Among them, T a , W a , H max are the average temperature, average wind speed, and maximum humidity respectively. n1 and n2 are the total numbers of temperature observation values and wind speed observation values in the historical weather data respectively. T i is the i-th temperature observation value, W i is the i-th wind speed observation value, and H i is the i-th humidity observation value.

3. The method for dynamically adjusting green energy and load using a virtual power plant according to claim 1, wherein The weather prediction model in step S2 is constructed as follows: S21: Input of the weather model; the input Y consists of two parts. Y1 is the historical meteorological information over a period of time in the past, and Y2 is the predicted value of the supercomputer. For T consecutive moments (1, 2,..., T) in history, the corresponding weather index observation values are denoted as Y = (y 1 , y 2 , …, y T ), and the corresponding predicted value of the supercomputer is denoted as Among them, the meteorological information data y t corresponding to a certain moment t1 contains k weather index values, which can be expressed as S22: Output of the weather model; including weather index values corresponding to the next P consecutive moments, which can be expressed as where the data for each moment also includes k weather index values; S23: Define the weather prediction model: S24: The weather prediction objective function is as follows: f(T, H, W) = w T ·T + w H ·H + w W ·W where, w T , w H , w W are the weighting factors of temperature, humidity and wind speed respectively.

4. The method for dynamically adjusting green energy and load using a virtual power plant according to claim 1, characterized in that In step S2, the PWLCM chaotic mapping and nonlinear escape energy update strategy are introduced into the HHO algorithm to ensure the global convergence efficiency.

5. The method for dynamically adjusting green energy and load using a virtual power plant according to claim 4, characterized in that The introduction of the PWLCM chaotic mapping: The original HHO algorithm randomly generates a population in the population space, resulting in a lack of population quality and low optimization accuracy. Therefore, the PWLCM chaotic mapping is introduced to adjust the initial parameter r of the HHO population. The adjustment formula is as follows: where r i is the initialization random parameter.

6. The method for dynamically regulating green energy and load using a virtual power plant according to claim 4, characterized in that, The introduction of the nonlinear escape energy update strategy: In the original HHO, the switch of the eagle group from the exploration stage to the exploitation stage is controlled according to the escape energy E. During the iteration process, E linearly decreases from 2 to 0; the linear decrease will lead to a decrease in the global search ability in the later stage of the algorithm iteration and increase the probability of falling into the local optimum. A new nonlinear escape energy update strategy is introduced, and the calculation formula is as follows: E = E0×(2×rand - 1) Among them, T0 is the maximum number of iterations. The improved HHO algorithm will preferentially select individuals with higher fitness values and use them as seeds for the next generation, thereby accelerating the convergence speed of the algorithm and finally obtaining a better global optimal solution X best .

7. The method for dynamically adjusting green energy and load using a virtual power plant according to claim 1, characterized in that The calculation formula of the VPP virtual power plant in step S3 is as follows: E total E(t) = E RE + E(t)+ E ESS + E(t)+ E EV (t) Among them, E total (t) is the total energy output of the virtual power plant, E RE (t) is the power generation of the controllable load at time t, E ESS (t) is the energy output of the energy storage device at time t, E EV (t) is the energy output of the electric vehicle at time t; Controllable load generated electricity E RE (t) The calculation formula is as follows: where N RE is the type of controllable load, P RE，i (t) is the power output of the i-th controllable load at time t, η RE，i is the conversion efficiency of the i-th controllable load, The energy output E of the energy storage device ESS (t) The calculation formula is as follows: E ESS E(t) = P ESS (t)·Δt·η ESS E(t) Among them, P ESS (t) is the charging and discharging power of the energy storage device at time t, Δt is the time interval, and η ESS (t) is the charging and discharging efficiency of the energy storage device at time t.

8. The method for dynamically adjusting green energy and load using a virtual power plant according to claim 1, characterized in that, The calculation formula of the energy demand load in step S4 is as follows: Among them, L(t) is the energy demand load, N is the number of weather factors affecting energy demand, and α i is the weight coefficient of the i-th weather factor, and W i (t) is the predicted value of the i-th weather factor at time t, and β i is the weight coefficient of the i-th historical load factor, and H i is the i-th historical load data, and γ is the comprehensive adjustment coefficient used to balance the impacts of weather and historical load.

9. The method for dynamically adjusting green energy and load using a virtual power plant according to claim 1, characterized in that, The division criteria for high load periods and low load periods are as follows: Load average value L avg : Where, T1 is the total number of time intervals in a day; Set the load threshold L threshold : L threshold = L avg ·(1 + δ) Where, δ represents the load fluctuation coefficient, usually taking values between 0.1 and 0.3; Divide high load periods and low load periods: T high = {t | L(t) ≥ L threshold} T low = {t | L(t) < L threshold} Among them, T high is the high-load period, and T low is the low-load period.

10. The method for dynamically adjusting green energy and load using a virtual power plant according to claim 1, characterized in that, The calculation formula of the green electricity price trend prediction model in step S5 is as follows: Among them, P price (t) is the predicted electricity price at time t, ζ is the weight coefficient of the energy supply-demand ratio, and E total (t) is the total energy output at time t, η is the weight coefficient of the market mechanism, and P market (t) is the market base electricity price at time t.

Images