Energy Supply and Demand Balance Analysis and Optimization Methods for Green Electricity-Driven Zero-Carbon Industrial Parks

Abstract

This application belongs to the field of power system technology, and more specifically, relates to a method for energy supply and demand balance analysis and optimization in green electricity-driven zero-carbon industrial parks. This invention analyzes historical power consumption data of various facilities and equipment within the industrial park and uses a demand forecasting model to predict future power demand, accurately matching energy supply and demand. This not only reduces energy waste but also improves the energy efficiency of the industrial park by optimizing resource allocation. Furthermore, by utilizing photovoltaic power generation capacity, it significantly reduces the industrial park's dependence on fossil fuels. Based on photovoltaic output, park demand, and carbon emission data, it provides strong data support and decision-making basis for the planning and construction of green energy technologies such as photovoltaics and energy storage.

Description

Technical Field

[0001] This application belongs to the field of power system technology, and more specifically, relates to the energy supply and demand balance analysis and optimization method of green electricity driven zero-carbon industrial parks. Background Technology

[0002] With increasing global focus on sustainable development, all sectors are actively exploring strategies to reduce environmental impact. Particularly in energy-intensive industrial parks, achieving a green and low-carbon energy transition has become an urgent issue. Industrial parks, as a vital driver of economic development, consume a significant proportion of energy, especially electricity. Therefore, efficiently managing and optimizing energy supply and consumption in industrial parks to achieve sustainable energy use is not only crucial for economic benefits but also directly impacts environmental burden.

[0003] Traditional industrial parks generally rely on fossil fuels, which not only leads to massive resource consumption but also results in serious carbon emissions. Against the backdrop of global climate change, carbon emissions have had a significant negative impact on the environment. Under this pressure, industrial parks urgently need to transform their energy use strategies and seek greener and more sustainable energy solutions.

[0004] To address these challenges, an increasing number of industrial parks are exploring solutions that integrate renewable energy, particularly solar and wind power technologies. These technologies can generate electricity using natural resources, reducing reliance on traditional fossil fuels and thus significantly lowering carbon emissions. However, the output of renewable energy sources such as solar and wind power is not stable and is greatly affected by climate and environmental conditions, posing new challenges to the energy management systems of industrial parks. Summary of the Invention

[0005] This invention provides a method for energy supply and demand balance analysis and optimization in green electricity-driven zero-carbon industrial parks, aiming to achieve supply and demand balance through the prediction of photovoltaic output.

[0006] A method for energy supply and demand balance analysis and optimization of green-electric-driven zero-carbon industrial parks, including the following steps:

[0007] Energy demand forecasting: Obtain electricity consumption data of various facilities and equipment in the industrial park during historical periods, and use demand forecasting models to predict future electricity demand;

[0008] Photovoltaic capacity assessment: Based on historical meteorological data, assess the power output of photovoltaic power in the park and determine the photovoltaic power output;

[0009] Supply and demand balance: Dynamically adjust the operation status of photovoltaic power generation and energy storage with the goal of minimizing carbon emissions.

[0010] This invention analyzes historical electricity consumption data of various facilities and equipment within industrial parks and uses a demand forecasting model to predict future electricity demand. This allows for accurate matching of energy supply and demand, reducing energy waste and improving energy efficiency in industrial parks through optimized resource allocation. Furthermore, by utilizing photovoltaic (PV) capacity, it significantly reduces the industrial park's reliance on fossil fuels. Based on PV output, park demand, and carbon emission data, it provides strong data support and decision-making basis for the planning and construction of green energy technologies such as PV and energy storage.

[0011] Preferably, the energy demand forecasting includes the following steps:

[0012] Data acquisition: Acquire the power consumption data and related influencing factor data, wherein the influencing factor data includes at least the power consumption of each device;

[0013] Data preprocessing: Standardize all numerical data; and then format the standardized data into a time series format.

[0014] Model prediction: The preprocessed data is used as input to the LSTM model, and the future electricity demand is output based on the LSTM model.

[0015] Preferably, the specific steps for assessing the photovoltaic and wind power capacity are as follows:

[0016] Historical sunshine duration and intensity data for the industrial park area are obtained to obtain the historical meteorological data; detailed specifications of the photovoltaic panels and historical operating data are obtained, wherein the detailed specifications include at least one of the following: the power generation rate of the photovoltaic panels and the light-receiving area of ​​the photovoltaic panels;

[0017] The initial photovoltaic capacity is predicted based on the detailed specifications of the photovoltaic panels and the historical meteorological data using a machine learning model; and the predicted capacity is adjusted based on the performance degradation of the photovoltaic panels and environmental factors to obtain the photovoltaic power output.

[0018] Preferably, the machine learning model is a random forest model, and the training of the random forest model includes the following steps:

[0019] Data acquisition: Acquire the specifications, environmental parameters, and power output data of the photovoltaic panels within the historical time period;

[0020] Data preprocessing: The acquired data is processed to handle missing and outlier values, and finally standardized to obtain the historical dataset;

[0021] Model training: The dataset is divided into training and test sets. The random forest model is trained on the training set data, and cross-validation is used to adjust the model parameters.

[0022] Model testing: The performance of the random forest model is evaluated on the test set based on the mean squared error. The parameters of the model are adjusted based on the performance evaluation results to obtain the random forest model.

[0023] Preferably, the specific steps for achieving supply and demand balance are as follows:

[0024] Calculate the supply difference ΔE:

[0025] ΔE=E demand -E supply ;

[0026] In the formula: E demand This represents a projected value for energy demand; E supply This represents the predicted value for photovoltaic power generation.

[0027] Energy storage system adjustment: If ΔE is greater than 0, the remaining power of the energy storage system is obtained. If the remaining power of the energy storage system is greater than ΔE, it is discharged from the energy storage system to meet additional demand. If the remaining power of the energy storage system is less than ΔE, the difference in demand after the energy storage system provides power is calculated. The reduceable load of the zero-carbon industrial park is obtained, and the demand is calculated based on the reduceable load and the difference. If the demand is met, the operating status of the corresponding equipment is adjusted based on the reduceable load. If the requirements are not met, fossil fuel power is connected.

[0028] If ΔE is less than 0, the excess remaining electricity will be stored in the energy storage system.

[0029] Preferably, when the fossil fuel power is connected, the demand for fossil fuels is calculated based on the required power, and carbon emissions are calculated based on the demand:

[0030] E fossil =ΔE remaining -E reducible ;

[0031] In the formula: E fossil ΔE represents the electrical power obtained from fossil fuels. remaining E represents the difference; reducible This indicates that the load can be reduced;

[0032] Carbon emission calculation:

[0033] C emissions =E fossil ×EF;

[0034] In the formula: C emissions It represents the amount of carbon emissions generated by using fossil fuels; EF represents the carbon emission factor of fossil fuels.

[0035] Preferably, the specific steps for achieving supply and demand balance are as follows:

[0036] Initialize the parameters of the particle swarm optimization algorithm and calculate the fitness:

[0037] Fitness function:

[0038] F(X i )=(α×F1(X i ),β×F2(X i ),γ×F3(X i ));

[0039] In the formula: F1(X i F2(X) represents the cost function; i F3(X) represents the carbon emission function; i F(X) represents the system reliability function; α, β, γ represent weighting factors, and F(X) represents the system reliability function. i ) indicates fitness;

[0040] Update particles:

[0041]

[0042] In the formula: X i,j V represents the position of the i-th particle in the j-th dimension, and represents the decision variable within a specific time period; i,j P represents the velocity of the i-th particle in the j-th dimension; best,i,j Gbest,i,j represents the individual optimal historical position of the i-th particle in the j-th dimension; Gbest,i,j represents the global optimal position in the j-th dimension; w(t) represents the inertia weight; c1 and c2 represent the learning factors; rand represents returning a random number in the range of 0 to 1. This represents the updated velocity of the i-th particle in the j-th dimension; w represents the updated position of the i-th particle in the j-th dimension; max w represents the initial maximum inertia weight. min c2 represents the minimum inertia weight during the iteration process; T represents the total number of iterations; t represents the number of iterations in this iteration; when the global optimum is not improved after n consecutive iterations, increase the value of c2 and decrease the value of c1.

[0043] After updating the particle positions, the performance of each particle is evaluated using non-dominated sorting and crowding distance; and when updating the optimal and global optimal positions of particles, the non-dominated relationships of multiple targets and the distribution of ions in the target space are considered.

[0044] If a particle fails to update its optimal position in the k-th iteration, a local search is performed on that particle to optimize its position.

[0045] The iteration stops after reaching the maximum number of iterations, thus obtaining the operating status of photovoltaic power generation and energy storage at the lowest carbon emission level.

[0046] Preferably, calculating the non-dominated ranking includes:

[0047] Iterate through all particles and calculate two properties for each particle: the number of solutions that govern the particle and the set of solutions that are governed by the particle.

[0048] The first frontier, i.e. the optimal frontier, is formed by particles that are not dominated by any other particles.

[0049] Use a recursive method to identify other fronts after the first front. For each particle in the k-th front, decrement the number of solutions for each particle in the set by one. When a particle is reduced to zero, add that particle to the (k+1)-th front.

[0050] Calculating congestion distance includes:

[0051] For each non-dominated frontier, the particles are sorted according to each objective function value;

[0052] Calculate the crowding distance for each particle: For each target, the distance of a particle is the sum of the distances of its two neighboring particles to the target function value;

[0053] For each target dimension, the crowding distance of particles located at the boundary is set to infinity to ensure priority during the selection process.

[0054] Preferably, the consideration of the non-dominated relationships of multiple targets and the distribution of ions in the target space is as follows:

[0055] At each iteration, the particle’s current fitness is compared with its best fitness. If the current fitness is better than the best fitness on at least one objective and no worse than the best fitness on other objectives, then the best fitness is updated to the current fitness.

[0056] Use non-dominated sorting to find the non-dominated set among all particles; select the particle with the largest crowding distance from the non-dominated set as the new global optimal position.

[0057] The beneficial effects of this invention include:

[0058] This invention analyzes historical electricity consumption data of various facilities and equipment within industrial parks and uses a demand forecasting model to predict future electricity demand. This allows for accurate matching of energy supply and demand, reducing energy waste and improving energy efficiency in industrial parks through optimized resource allocation. Furthermore, by utilizing photovoltaic (PV) capacity, it significantly reduces the industrial park's reliance on fossil fuels. Based on PV output, park demand, and carbon emission data, it provides strong data support and decision-making basis for the planning and construction of green energy technologies such as PV and energy storage. Attached Figure Description

[0059] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0060] Figure 1 This is a system architecture block diagram provided for an embodiment of the present invention. Detailed Implementation

[0061] To make the technical problems, technical solutions, and beneficial effects to be solved by this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this application.

[0062] It should be noted that when a component is referred to as being "fixed to" or "set on" another component, it can be directly on or indirectly on that other component. When a component is referred to as being "connected to" another component, it can be directly connected to or indirectly connected to that other component.

[0063] It should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0064] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0065] Example 1

[0066] See Figure 1 As shown, the energy supply and demand balance analysis and optimization method for green electricity-driven zero-carbon industrial parks includes the following steps:

[0067] Energy demand forecasting: Obtain electricity consumption data of various facilities and equipment in the industrial park during historical periods, and use demand forecasting models to predict future electricity demand;

[0068] Photovoltaic capacity assessment: Based on historical meteorological data, assess the power output of photovoltaic power in the park and determine the photovoltaic power output;

[0069] Supply and demand balance: Dynamically adjust the operation status of photovoltaic power generation and energy storage with the goal of minimizing carbon emissions.

[0070] This invention analyzes historical electricity consumption data of various facilities and equipment within industrial parks and uses a demand forecasting model to predict future electricity demand. This allows for accurate matching of energy supply and demand, reducing energy waste and improving energy efficiency in industrial parks through optimized resource allocation. Furthermore, by utilizing photovoltaic (PV) capacity, it significantly reduces the industrial park's reliance on fossil fuels. Based on PV output, park demand, and carbon emission data, it provides strong data support and decision-making basis for the planning and construction of green energy technologies such as PV and energy storage.

[0071] Preferably, the above energy demand forecasting includes the following steps:

[0072] Data Acquisition: Acquire the above-mentioned power consumption data and related influencing factor data. The above-mentioned influencing factor data includes at least the power consumption of each piece of equipment, and also the production plan of the industrial park.

[0073] Data preprocessing: Standardize all numerical data; and then format the standardized data into a time series format.

[0074] Model prediction: The preprocessed data is used as input to the LSTM model, and the future electricity demand is output based on the LSTM model.

[0075] Preferably, the specific steps for the above-mentioned photovoltaic and wind power capacity assessment are as follows:

[0076] Obtain historical sunshine duration and intensity data for the area where the industrial park is located, and obtain the aforementioned historical meteorological data; obtain detailed specifications data for the photovoltaic panels, as well as historical operating data, wherein the aforementioned detailed specifications data shall include at least one of the following: the power generation rate of the aforementioned photovoltaic panels, and the light-receiving area of ​​the aforementioned photovoltaic panels;

[0077] Based on the detailed specifications of the photovoltaic panels and the historical meteorological data mentioned above, the initial photovoltaic capacity is predicted using a machine learning model. The predicted capacity is then adjusted based on the performance degradation of the photovoltaic panels and environmental factors to obtain the photovoltaic power output.

[0078] Preferably, the above machine learning model uses a random forest model, and the training of the random forest model includes the following steps:

[0079] Data Acquisition: Acquire the specifications, environmental parameters, and power output data of the aforementioned photovoltaic panels within a historical time period;

[0080] Data preprocessing: The acquired data is processed to handle missing and outlier values, and finally standardized to obtain the historical dataset;

[0081] Model training: The dataset is divided into training and test sets. The random forest model is trained on the training set data, and cross-validation is used to adjust the model parameters.

[0082] Model testing: The performance of the random forest model was evaluated on the test set based on the mean squared error. The model parameters were adjusted based on the performance evaluation results to obtain the random forest model.

[0083] Preferably, the specific steps for achieving the above-mentioned supply and demand balance are as follows:

[0084] Calculate the supply difference ΔE:

[0085] ΔE=E demand -E supply ;

[0086] In the formula: E demand This represents a projected value for energy demand; E supply This represents the predicted value for photovoltaic power generation.

[0087] Energy storage system adjustment: If ΔE is greater than 0, the remaining power of the energy storage system is obtained. If the remaining power of the energy storage system is greater than ΔE, it is discharged from the energy storage system to meet additional demand. If the remaining power of the energy storage system is less than ΔE, the difference in demand after the energy storage system provides power is calculated. The reduceable load of the zero-carbon industrial park is obtained, and the demand is calculated based on the reduceable load and the difference. If the demand is met, the operating status of the corresponding equipment is adjusted based on the reduceable load. If the requirements are not met, fossil fuel power is connected.

[0088] If ΔE is less than 0, the excess remaining electricity will be stored in the energy storage system.

[0089] Preferably, when connecting to the aforementioned fossil fuel power, the demand for fossil fuels is calculated based on the required electricity, and carbon emissions are calculated based on the aforementioned demand:

[0090] Efossil =ΔE remaining -E reducible ;

[0091] In the formula: E fossil ΔE represents the electrical power obtained from fossil fuels. remaining E represents the difference; reducible This indicates that the load can be reduced;

[0092] Carbon emission calculation:

[0093] C emissions =E fossil ×EF;

[0094] In the formula: C emissions It represents the amount of carbon emissions generated by using fossil fuels; EF represents the carbon emission factor of fossil fuels.

[0095] Example 2 differs from the previous example in that it employs a different supply and demand balancing technology; the details are as follows:

[0096] The specific steps for achieving the above-mentioned supply and demand balance are as follows:

[0097] Initialize the parameters of the particle swarm optimization algorithm and calculate the fitness:

[0098] Fitness function:

[0099] F(X i )=(α×F1(X i ),β×F2(X i ),γ×F3(X i ));

[0100] In the formula: F1(X i F2(X) represents the cost function; i F3(X) represents the carbon emission function; i F(X) represents the system reliability function; α, β, γ represent weighting factors, and F(X) represents the system reliability function. i ) indicates fitness;

[0101] Update particles:

[0102]

[0103] In the formula: X i,j V represents the position of the i-th particle in the j-th dimension, and represents the decision variable within a specific time period; i,j P represents the velocity of the i-th particle in the j-th dimension; best,i,jGbest,i,j represents the individual optimal historical position of the i-th particle in the j-th dimension; Gbest,i,j represents the global optimal position in the j-th dimension; w(t) represents the inertia weight; c1 and c2 represent the learning factors; rand represents returning a random number in the range of 0 to 1. This represents the updated velocity of the i-th particle in the j-th dimension; w represents the updated position of the i-th particle in the j-th dimension; max w represents the initial maximum inertia weight. min c2 represents the minimum inertia weight during the iteration process; T represents the total number of iterations; t represents the number of iterations in this iteration; when the global optimum is not improved after n consecutive iterations, increase the value of c2 and decrease the value of c1.

[0104] After updating the particle positions, the performance of each particle is evaluated using non-dominated sorting and crowding distance; and when updating the optimal and global optimal positions of particles, the non-dominated relationships of multiple targets and the distribution of ions in the target space are considered.

[0105] If a particle fails to update its optimal position in the k-th iteration, a local search is performed on that particle to optimize its position.

[0106] The iteration stops after reaching the maximum number of iterations, thus obtaining the operating status of photovoltaic power generation and energy storage at the lowest carbon emission level.

[0107] Preferably, calculating the above non-dominated sort includes:

[0108] Iterate through all particles and calculate two properties for each particle: the number of solutions that govern the particle and the set of solutions that are governed by the particle.

[0109] The first frontier, i.e. the optimal frontier, is formed by particles that are not dominated by any other particles.

[0110] Use a recursive method to identify other fronts after the first front. For each particle in the k-th front, decrement the number of solutions for each particle in the set by one. When a particle is reduced to zero, add that particle to the (k+1)-th front.

[0111] Calculating congestion distance includes:

[0112] For each non-dominated frontier, the particles are sorted according to each objective function value;

[0113] Calculate the crowding distance for each particle: For each target, the distance of a particle is the sum of the distances of its two neighboring particles to the target function value;

[0114] For each target dimension, the crowding distance of particles located at the boundary is set to infinity to ensure priority during the selection process.

[0115] Preferably, the above considerations regarding the non-dominated relationships of multiple targets and the distribution of ions in the target space are as follows:

[0116] At each iteration, the particle’s current fitness is compared with its best fitness. If the current fitness is better than the best fitness on at least one objective and no worse than the best fitness on other objectives, then the best fitness is updated to the current fitness.

[0117] Use non-dominated sorting to find the non-dominated set among all particles; select the particle with the largest crowding distance from the non-dominated set as the new global optimal position.

[0118] In this embodiment:

[0119] Dynamic adjustment: By dynamically adjusting the inertia weight w(t), the algorithm can explore the solution space with a larger inertia weight in the early stage of the search, and refine the local search by reducing the inertia weight in the later stage of the search. This helps to achieve a balance between global exploration and local development.

[0120] Enhanced exploration and utilization: By adjusting the learning factors c1 and c2, the tendency of particles to learn towards individual or global optimal solutions can be enhanced when necessary. This helps the algorithm to strengthen exploration or utilization as needed at different stages.

[0121] Diversity Preservation: By calculating the crowding distance, the algorithm can maintain the diversity of solutions in the population, avoid the algorithm from converging to a local optimum too early, and ensure the potential for exploring the global optimum.

[0122] Balanced multi-objective optimization: Non-dominated sorting helps the algorithm achieve balance among different objectives, ensuring that no single objective is over-prioritized, which is especially important for multi-objective optimization problems.

[0123] Increasing search depth: When a particle fails to update its optimal position in multiple iterations, local search can help the particle escape possible local optima and explore new potential solution spaces, thereby improving the quality of solutions and the search efficiency of the algorithm.

[0124] Real-time optimization feedback: By continuously updating the individual optimal and global optimal positions, the particle swarm optimization algorithm can provide real-time feedback on the latest optimal solution, accelerating the convergence speed while ensuring responsiveness to real-time changes in needs and environment.

[0125] Balancing pressure and diversity: For each objective dimension, setting the crowding distance of boundary particles to infinity ensures the priority of these particles in the selection process, which helps maintain the breadth and depth of the solution.

[0126] Example 1 uses traditional data analysis and forecasting models to predict energy demand and assess the capacity of photovoltaic and wind power. This example mainly relies on acquiring and analyzing historical data, using an LSTM (Long Short-Term Memory) network model to predict future electricity demand, and employing machine learning models (such as random forests) based on historical meteorological data and photovoltaic panel performance to predict photovoltaic capacity.

[0127] Beneficial effects:

[0128] Accurate demand forecasting: LSTM models can effectively process time series data and improve the accuracy of electricity demand forecasting.

[0129] Optimize resource allocation: Through accurate forecasting models, the power generation plans for photovoltaic and wind power can be arranged more rationally, reducing energy waste.

[0130] Reduced carbon emissions: Significantly increased the proportion of renewable energy used in industrial parks, reducing dependence on fossil fuels.

[0131] Application scenarios:

[0132] Suitable for industrial parks with relatively stable electricity demand and minimal changes in environmental conditions (such as sunshine and wind speed). It can effectively manage and optimize the existing energy structure, making it suitable for initial transitions to green energy.

[0133] Example 2

[0134] Technical solution:

[0135] Example 2 employs Particle Swarm Optimization (PSO) for dynamic supply and demand balance adjustment. This is a more complex and dynamic optimization method that considers a multi-objective optimization problem involving cost, carbon emissions, and system reliability. It seeks the optimal solution among multiple objectives through non-dominated sorting and congestion distance calculation.

[0136] Beneficial effects:

[0137] Multi-objective optimization: It provides a comprehensive optimization solution that simultaneously considers cost, carbon emissions, and system reliability.

[0138] Dynamic responsiveness: By adjusting strategies in real time, it can respond more flexibly to changes in demand and supply, thereby improving energy efficiency.

[0139] Maintaining solution diversity: Crowded distance and non-dominated sorting are used to ensure solution diversity and prevent solutions from converging to local optima prematurely.

[0140] Application scenarios:

[0141] Suitable for industrial parks where electricity demand and renewable energy supply fluctuate significantly. Particularly suitable for situations with drastic changes in environmental conditions or highly uncertain electricity demand, enabling effective real-time adjustments and optimizations.

[0142] Differences and choices:

[0143] Example 1 focuses more on using historical data and stable predictive models for energy management, while Example 2 employs a more dynamic and adaptive multi-objective optimization method to address complex and changing environments. Based on the specific needs and conditions of the industrial park (such as the stability of electricity demand, the availability of renewable energy, and environmental variability), a more suitable implementation plan can be selected to optimize the energy supply and demand balance. Example 2 is particularly suitable for applications requiring high flexibility and the ability to respond to rapidly changing environments, while Example 1 is suitable for relatively stable application environments.

[0144] The above are merely preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for energy supply and demand balance analysis and optimization in green electricity-driven zero-carbon industrial parks, characterized in that, Includes the following steps: Energy demand forecasting: Obtain electricity consumption data of various facilities and equipment in the industrial park during historical periods, and use demand forecasting models to predict future electricity demand; Photovoltaic capacity assessment: Based on historical meteorological data, assess the power output of photovoltaic power in the park and determine the photovoltaic power output; Supply and demand balance: Dynamically adjust the operation status of photovoltaic power generation and energy storage with the goal of minimizing carbon emissions. The specific steps for the photovoltaic capacity assessment are as follows: obtain historical sunshine duration and intensity data for the area where the industrial park is located, and obtain the historical meteorological data; obtain detailed specification data of the photovoltaic panels, as well as historical operating data, wherein the detailed specification data includes at least one of the following: the power generation rate of the photovoltaic panels and the light-receiving area of ​​the photovoltaic panels; The initial photovoltaic (PV) capacity is predicted based on the detailed specifications of the PV panels and historical meteorological data using a machine learning model. The predicted capacity is then adjusted based on the performance degradation of the PV panels and environmental factors to obtain the PV power output. The specific steps for achieving supply and demand balance are as follows: Calculate supply differences : ； In the formula: This represents a forecast of energy demand. This represents the predicted value for photovoltaic power generation. Energy storage system adjustment: If If the remaining energy level is greater than 0, then the remaining energy level of the energy storage system is obtained. If the remaining energy level of the energy storage system is greater than 0, then the remaining energy level of the energy storage system is obtained. Then, the energy is discharged from the energy storage system to meet additional demand; If the remaining power of the energy storage system is less than Then calculate the difference in demand after the energy storage system provides electricity; Obtain the load that can be reduced in the zero-carbon industrial park, and calculate whether the demand is met based on the load that can be reduced and the difference: if the demand is met, adjust the operating status of the corresponding equipment based on the load that can be reduced; if the demand is not met, connect to fossil fuel power. like If the value is less than 0, the excess remaining electricity will be stored in the energy storage system.

2. The method for energy supply and demand balance analysis and optimization of green electricity-driven zero-carbon industrial parks according to claim 1, characterized in that, The energy demand forecasting includes the following steps: Data acquisition: Acquire the power consumption data and related influencing factor data, wherein the influencing factor data includes at least the power consumption of each device; Data preprocessing: Standardize all numerical data; and then format the standardized data into a time series format. Model prediction: The preprocessed data is used as input to the LSTM model, and the future electricity demand is output based on the LSTM model.

3. The method for energy supply and demand balance analysis and optimization of green electricity-driven zero-carbon industrial parks according to claim 1, characterized in that, The machine learning model uses a random forest model, and the training of the random forest model includes the following steps: Data acquisition: Acquire the specifications, environmental parameters, and power output data of the photovoltaic panels within the historical time period; Data preprocessing: The acquired data is processed to handle missing and outlier values, and finally standardized to obtain the historical dataset; Model training: The dataset is divided into training and test sets. The random forest model is trained on the training set data, and cross-validation is used to adjust the model parameters. Model testing: The performance of the random forest model is evaluated on the test set based on the mean squared error. The parameters of the model are adjusted based on the performance evaluation results to obtain the random forest model.

4. The method for energy supply and demand balance analysis and optimization of green electricity-driven zero-carbon industrial parks according to claim 1, characterized in that, When the fossil fuel power is connected, the demand for fossil fuels is calculated based on the required electricity, and carbon emissions are calculated based on the demand: ； In the formula: It represents the electrical power obtained from fossil fuels; This represents the difference; This indicates that the load can be reduced; Carbon emission calculation: ； In the formula: This indicates the amount of carbon emissions generated from the use of fossil fuels; This indicates the carbon emission factor of fossil fuels.

5. A method for energy supply and demand balance analysis and optimization in green electricity-driven zero-carbon industrial parks, characterized in that... Includes the following steps: Energy demand forecasting: Obtain electricity consumption data of various facilities and equipment in the industrial park during historical periods, and use demand forecasting models to predict future electricity demand; Photovoltaic capacity assessment: Based on historical meteorological data, assess the power output of photovoltaic power in the park and determine the photovoltaic power output; Supply and demand balance: The operation status of photovoltaic power generation and energy storage is dynamically adjusted with the goal of minimizing carbon emissions. The specific steps of the photovoltaic capacity assessment are as follows: obtain historical sunshine duration and intensity data of the industrial park area to obtain the historical meteorological data; obtain detailed specification data of photovoltaic panels and historical operation data. The detailed specification data includes at least one of the following: power generation rate of photovoltaic panels and light-receiving area of ​​photovoltaic panels. The initial photovoltaic (PV) capacity is predicted based on the detailed specifications of the PV panels and historical meteorological data using a machine learning model. The predicted capacity is then adjusted based on the performance degradation of the PV panels and environmental factors to obtain the PV power output. The specific steps for achieving supply and demand balance are as follows: Initialize the parameters of the particle swarm optimization algorithm and calculate the fitness: Fitness function: ； In the formula: Represents the cost function; A function representing carbon emissions; A function representing system reliability; Indicates the weighting factor. Indicates fitness; Update particles: ； ； ； In the formula: This represents the position of the i-th particle in the j-th dimension, and represents the decision variable within a specific time period; This represents the velocity of the i-th particle in the j-th dimension; This represents the optimal historical position of the i-th particle in the j-th dimension. This represents the globally optimal position in the j-th dimension; Indicates inertia weight; and Indicates the learning factor; rand This indicates that a random number between 0 and 1 will be returned. This represents the updated velocity of the i-th particle in the j-th dimension; This represents the updated position of the i-th particle in the j-th dimension; This represents the initial maximum inertia weight; This represents the minimum inertia weight during the iteration process; T Indicates the total number of iterations; t Indicates the number of iterations; If the global optimum remains unchanged after n consecutive iterations, increase... The value decreases. The value; After updating the particle positions, the performance of each particle is evaluated using non-dominated sorting and crowding distance; and when updating the optimal and global optimal positions of particles, the non-dominated relationships of multiple targets and the distribution of ions in the target space are considered. If a particle fails to update its optimal position in the k-th iteration, a local search is performed on that particle to optimize its position. The iteration stops after reaching the maximum number of iterations, thus obtaining the operating status of photovoltaic power generation and energy storage at the lowest carbon emission level.

6. The method for energy supply and demand balance analysis and optimization of green electric-driven zero-carbon industrial parks according to claim 5, characterized in that, Calculating the non-dominated order includes: Iterate through all particles and calculate two properties for each particle: the number of solutions that govern the particle and the set of solutions that are governed by the particle. The first frontier, i.e. the optimal frontier, is formed by particles that are not dominated by any other particles. Use a recursive method to identify other fronts after the first front. For each particle in the k-th front, decrement the number of solutions for each particle in the set by one. When a particle is reduced to zero, add that particle to the (k+1)-th front. Calculating congestion distance includes: For each non-dominated frontier, the particles are sorted according to each objective function value; Calculate the crowding distance for each particle: For each target, the distance of a particle is the sum of the distances of its two neighboring particles to the target function value; For each target dimension, the crowding distance of particles located at the boundary is set to infinity to ensure priority during the selection process.

7. The method for energy supply and demand balance analysis and optimization of green electric-driven zero-carbon industrial parks according to claim 5, characterized in that, The consideration of the non-dominated relationships of multiple targets and the distribution of ions in the target space is as follows: At each iteration, the particle’s current fitness is compared with its best fitness. If the current fitness is better than the best fitness on at least one objective and no worse than the best fitness on other objectives, then the best fitness is updated to the current fitness. Use non-dominated sorting to find the non-dominated set among all particles; select the particle with the largest crowding distance from the non-dominated set as the new global optimal position.

Images