Power grid carbon emission reduction method based on new energy output uncertainty coordination

Abstract

The invention provides a power grid carbon emission reduction method based on new energy output uncertainty coordination, and belongs to the technical field of new energy carbon emission reduction, and the method comprises the steps: 1, obtaining real-time data of different new energy generators, and carrying out the preprocessing; 2, establishing a new energy generator model according to the preprocessed data; step 3, constructing a reinforcement learning environment, performing reinforcement learning agent training according to a method of dynamically adjusting a learning rate, and determining a reward function; 4, determining the difference between the target of carbon emission reduction and the actual output according to the reward function, and smoothing the output fluctuation of the new energy based on reactive compensation and energy storage technologies; and step 5, evaluating the carbon emission of the power grid, modifying a reward function or adjusting parameters of an agency, and carrying out new energy output uncertainty coordination on the carbon emission reduction of the power grid. The problems that a power grid carbon emission reduction method is not accurate enough in prediction of new energy output, output of a new energy generator has uncertainty, and manual intervention and supervision are needed are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of carbon emission reduction from new energy sources, and in particular to a method for carbon emission reduction in a power grid based on coordination of uncertainty in output of new energy sources. Background Art

[0002] As global climate change intensifies, reducing carbon emissions has become a top priority. However, since the output of renewable energy power generation is highly uncertain, achieving carbon emission reduction in the power grid remains a challenge.

[0003] However, traditional grid carbon emission reduction methods are not accurate enough in predicting renewable energy output, resulting in the inability to accurately coordinate the output of each generator. This may cause some generators to emit excessive carbon dioxide at low loads and fail to fully utilize renewable energy at high loads. At the same time, the output of renewable energy generators is highly uncertain and requires human intervention and supervision, which challenges the stability of the grid.

[0004] Therefore, the present invention proposes a method for reducing carbon emissions in a power grid based on the coordination of uncertainty in the output of new energy sources. Summary of the invention

[0005] The present invention provides a method for reducing carbon emissions in a power grid based on the coordination of uncertainty in the output of renewable energy, so as to solve the problem that the prediction of the output of renewable energy in the prior art is not accurate enough, resulting in the inability to accurately coordinate the output of each generator, which may cause some generators to emit excessive carbon dioxide at low loads and fail to make full use of renewable energy at high loads. At the same time, since the output of renewable energy generators is highly uncertain and requires manual intervention and supervision, the stability of the power grid is challenged.

[0006] On the one hand, the present invention provides a method for reducing carbon emissions in a power grid based on coordination of uncertainty in new energy output, comprising:

[0007] Step 1: Acquire real-time data of different new energy generators and pre-process the real-time data;

[0008] Step 2: Establish a new energy generator model based on the influence of natural factors according to the preprocessed data;

[0009] Step 3: Construct a reinforcement learning environment based on the reinforcement learning algorithm according to the new energy generator model. In the constructed environment, train the reinforcement learning agent according to the method of dynamically adjusting the learning rate, and determine the reward function according to the reinforcement learning environment after training;

[0010] Step 4: Determine the difference between the carbon emission reduction target and the actual output according to the reward function, and smooth the output fluctuation of the new energy based on the reactive power compensation and energy storage technology according to the difference;

[0011] Step 5: Evaluate the carbon emissions of the power grid based on the output fluctuations of renewable energy, modify the reward function or adjust the parameters of the agent based on the evaluation results, and coordinate the uncertainty of renewable energy output for carbon emission reduction in the power grid based on the modified and adjusted reinforcement learning environment.

[0012] According to a method for reducing carbon emissions in a power grid based on coordination of uncertainty in output of renewable energy provided by the present invention, real-time data of different renewable energy generators are obtained, and the real-time data is preprocessed, including:

[0013] Acquire the generator types of different renewable energy generators, and determine the grid topology and load characteristics of the renewable energy generators according to the generator types;

[0014] Obtain sensors and data acquisition equipment based on the grid topology, load characteristics, and data accuracy requirements of renewable energy generators;

[0015] Obtain real-time data of different renewable energy generators based on sensors and data acquisition equipment;

[0016] The real-time data is analyzed and preprocessed according to edge computing technology, and the preprocessed data is transmitted to a remote data center based on communication technology.

[0017] According to a method for reducing carbon emissions in a power grid based on coordination of uncertainty in new energy output provided by the present invention, a new energy generator model is established based on preprocessed data and the influence of natural factors, including:

[0018] Obtain environmental factors and historical power generation records related to the operation of new energy generators;

[0019] Extract relevant features of environmental factors and historical power generation records based on feature extraction technology;

[0020] Obtain the geographical location information of new energy generators and establish a new geographical topology based on the geographic information system;

[0021] Establishing a new energy generator model based on a new geographic topology according to the relevant features;

[0022] Among them, for weather factors that cannot be accurately simulated, interpolation or quantile regression methods are used for approximation.

[0023] According to a method for reducing carbon emissions in a power grid based on coordination of uncertainty in new energy output provided by the present invention, a reinforcement learning environment is constructed based on a reinforcement learning algorithm according to a new energy generator model. In the constructed environment, reinforcement learning agent training is performed according to a method for dynamically adjusting the learning rate, and a reward function is determined according to the reinforcement learning environment after training, including:

[0024] Determine the state space of the new energy generator according to the input and output of the new energy generator model, and determine the states of the state space and the transition rules between the states according to the state space;

[0025] Get the set of actions taken by the new energy generator in each state;

[0026] Constructing a reinforcement learning environment based on the reinforcement learning algorithm according to the transfer rules and the action set;

[0027] In the constructed environment, the initial learning rate is selected according to the greedy algorithm, and the learning rate is dynamically adjusted according to the value of the loss function and the size of the gradient, and the reinforcement learning agent training is performed according to the adjusted learning rate;

[0028] The reward function is determined based on the principles of reinforcement learning, the state and behavior of the new energy generator according to the trained reinforcement learning environment.

[0029] According to a method for reducing carbon emissions in a power grid based on coordination of uncertainty in output of renewable energy provided by the present invention, after obtaining a set of actions taken by the renewable energy generator in each state, the method further includes:

[0030] Acquire different working states of the new energy generator, and classify the different working states;

[0031] Monitor various status information of new energy generators in real time through telemetry and telesignaling systems, identify status boundaries based on the status information, and determine whether the generator enters or leaves a certain state;

[0032] According to the divided working status, the corresponding action set is parsed based on the status information;

[0033] The parsed action set is organized into an action sequence according to a certain order and rules, the correctness of the action sequence is verified, and the verified action sequence is stored.

[0034] According to a method for reducing carbon emissions in a power grid based on coordination of uncertainty in output of new energy, the present invention provides a method for reducing carbon emissions in a power grid, wherein the difference between the target of carbon emission reduction and the actual output is determined according to the reward function, and the output fluctuation of new energy is smoothed based on reactive power compensation and energy storage technology according to the difference, including:

[0035] Obtaining historical carbon emission data of new energy generators, and setting carbon emission reduction targets based on the historical carbon emission data and power generation performance of the new energy generators;

[0036] Determine the difference between the carbon emission reduction target and the actual output according to the reward function;

[0037] Obtaining the capacity and connection mode of the capacitor and the inductor of the new energy generator, and determining the phase difference between the current and the voltage according to the capacity and the connection mode;

[0038] Compensating the reactive power of the new energy generator according to the phase difference between the current and the voltage;

[0039] Based on the differences, the output fluctuations of new energy are smoothed out by using compensated new energy generators and energy storage technology.

[0040] According to a method for reducing carbon emissions in a power grid based on coordination of uncertainty in output of new energy, provided by the present invention, carbon emissions of the power grid are evaluated according to the output fluctuation of new energy, a reward function is modified or the parameters of the agent are adjusted according to the evaluation result, and the uncertainty in output of new energy is coordinated for carbon emissions reduction in the power grid according to the modified and adjusted reinforcement learning environment, including:

[0041] Obtain relevant data on the output fluctuation of new energy within a unit time period;

[0042] Determine the carbon emissions of the power grid based on the current power grid load and carbon emission standards according to the relevant data;

[0043] Compare the carbon emissions of the power grid with the set carbon emission targets, evaluate the carbon emission reduction of the power grid based on the comparison results, and obtain the reasons for not meeting the carbon emission reduction requirements based on the evaluation results;

[0044] The reward function is modified or the parameters of the agent are adjusted according to the reasons, and the uncertainty of the output of new energy sources is coordinated for carbon emission reduction in the power grid according to the modified and adjusted reinforcement learning environment.

[0045] According to a method for reducing carbon emissions in a power grid based on coordination of uncertainty in output of renewable energy provided by the present invention, after acquiring real-time data of different renewable energy generators, the method further includes:

[0046] Determine the mechanical energy parameters of the new energy generator according to the real-time data, and determine the power generation performance characteristics of the new energy generator based on the mechanical energy parameters;

[0047] Determine the corresponding state function of each working component of the new energy generator according to the power generation performance characteristics;

[0048] Determine the current frequency response parameters of each working component through the state response function, and determine the frequency recovery requirement according to the current frequency response parameters and the preset frequency response parameters of each working component;

[0049] Obtaining the frequency modulation condition of each working component, and determining the frequency modulation incremental loss according to the frequency modulation condition and the frequency recovery requirement of the working component;

[0050] Determine the organic control target value of the new energy generator based on the frequency modulation incremental loss and the initial loss of each working component;

[0051] Determine the working scale load of the new energy generator based on the organic control target value of the new energy generator;

[0052] Determine the coordinated control strategy of multiple working components of the new energy generator through a preset balanced optimization algorithm according to the working scale load, and determine the new energy consumption parameters according to the coordinated control strategy;

[0053] The carbon emission index is determined based on the new energy consumption parameters, and whether the working mode of the new energy generator exceeds the standard is determined according to the carbon emission index. If so, the real-time data of the new energy generator that exceeds the standard in the working mode that does not exceed the standard is re-collected.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] By constructing a new energy generator model and a reinforcement learning environment, and conducting training to determine the reward function, smoothing the output fluctuations of new energy, and modifying the reward function or adjusting the parameters of the agent, it is possible to ensure that the grid carbon emission reduction method is sufficiently accurate in predicting the output of new energy, and to accurately coordinate the output of each generator to prevent some generators from excessively emitting carbon dioxide at low loads and failing to fully utilize renewable energy at high loads. At the same time, the output of new energy generators can be made highly deterministic and do not require human intervention and supervision, thereby improving the stability of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0057] Figure 1 It is a flow chart of a method for reducing carbon emissions in a power grid based on coordination of uncertainty in new energy output provided by an embodiment of the present invention;

[0058] Figure 2 It is a schematic diagram of a flow chart for preprocessing real-time data of different new energy generators provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] Embodiment 1:

[0061] The embodiment of the present invention provides a method for reducing carbon emissions in a power grid based on coordination of uncertainty in new energy output, such as Figure 1 As shown, the method mainly includes the following steps:

[0062] Step 1: Acquire real-time data of different new energy generators and pre-process the real-time data;

[0063] Step 2: Establish a new energy generator model based on the influence of natural factors according to the preprocessed data;

[0064] Step 3: Construct a reinforcement learning environment based on the reinforcement learning algorithm according to the new energy generator model. In the constructed environment, train the reinforcement learning agent according to the method of dynamically adjusting the learning rate, and determine the reward function according to the reinforcement learning environment after training;

[0065] Step 4: Determine the difference between the carbon emission reduction target and the actual output according to the reward function, and smooth the output fluctuation of the new energy based on the reactive power compensation and energy storage technology according to the difference;

[0066] Step 5: Evaluate the carbon emissions of the power grid based on the output fluctuations of renewable energy, modify the reward function or adjust the parameters of the agent based on the evaluation results, and coordinate the uncertainty of renewable energy output for carbon emission reduction in the power grid based on the modified and adjusted reinforcement learning environment.

[0067] In this embodiment, the uncertainty of new energy output refers to the phenomenon that there is a certain gap between the actual power generation of new energy power generation equipment and the predicted power generation. The impact of the uncertainty of new energy output is mainly reflected in: the decline in power quality, the challenge of power system stability, and the increase in economic costs.

[0068] In this embodiment, grid carbon emission reduction refers to reducing the emission of greenhouse gases such as carbon dioxide in the power system by adopting low-carbon technologies, policies and management measures, thereby reducing the impact on climate change.

[0069] In this embodiment, the real-time data includes: output power, power generation frequency, and load level of the new energy generator.

[0070] In this embodiment, preprocessing includes: normalization, cleaning, denoising, etc.

[0071] In this embodiment, the influence of natural factors includes:

[0072] Weather changes: Weather changes directly affect the power generation of new energy power generation equipment. For example, changes in wind speed and temperature will affect the efficiency of wind and photovoltaic power generation.

[0073] Geographical regional differences: The climate conditions and geographical environment of different regions vary greatly, which will also lead to fluctuations in the output of new energy in different regions. For example, the northwest region of my country is mostly plateaus and mountains, with rich solar energy resources but relatively few wind energy resources.

[0074] Natural disasters: Natural disasters such as typhoons, heavy rains, snowstorms, etc. can damage new energy power generation equipment, affecting its power generation.

[0075] Human factors: For example, the decline in air quality and visibility caused by air pollution control will also have an impact on the output of new energy.

[0076] In this embodiment, the new energy generator model is used to describe and predict the output characteristics of the new energy generator and the actual power generation of the new energy generator.

[0077] In this embodiment, the reinforcement learning algorithm may be: Q-learning, Policy Gradient.

[0078] In this embodiment, the reinforcement learning environment is a computing environment that allows the reinforcement learning algorithm to learn and optimize behavior. In this environment, the reinforcement learning algorithm learns how to make optimal decisions by interacting with the environment to maximize long-term returns. The environment usually includes components such as state space, action space, and state-action value function, which together constitute the input and output of the reinforcement learning algorithm and interact with the new energy generator in a constantly changing environment.

[0079] In this embodiment, the learning rate is an important hyperparameter in reinforcement learning, which controls the learning speed of each parameter.

[0080] In this embodiment, reinforcement learning agent training refers to training an intelligent agent in a reinforcement learning scenario so that it can reach the optimal decision-making in the process of interaction and learning with the environment. The goal of agent training is to distribute tasks among multiple intelligent agents and enable them to collaborate in a game environment with common goals with the environment and peers, ultimately achieving global maximum utility.

[0081] In this embodiment, the reward function defines the reward obtained by the new energy generator when taking a certain action in a certain state.

[0082] In this embodiment, reactive power compensation refers to a technical measure in the power system to maintain the reactive power generated by the power grid within a certain range through compensation equipment, thereby ensuring stable operation of the system.

[0083] In this embodiment, energy storage technology refers to a technology that stores and manages energy over a certain period of time by utilizing a device or system that can store energy.

[0084] In this embodiment, the output fluctuation of new energy refers to the changes in power output caused by the influence of weather, seasons and other factors on renewable energy (such as solar energy, wind energy, etc.). These changes may cause grid voltage fluctuations, affecting grid stability and power supply quality. The main purpose of smoothing the output fluctuation of new energy is to reduce the risk of imbalance in power supply and demand caused by the large-scale access of new energy to the grid, and to ensure the safe and stable operation of the power system.

[0085] In this embodiment, adjusting the parameters of the agent may include increasing or decreasing certain strategies to balance the trade-off between carbon emission reduction and social benefits, such as adding some strategies with higher social benefits to improve overall satisfaction, while also trying to reduce the frequency of use of strategies with higher carbon emissions.

[0086] The beneficial effects of the above technical solution are: by constructing a new energy generator model and a reinforcement learning environment, and conducting training to determine the reward function, smoothing the output fluctuations of new energy, and modifying the reward function or adjusting the parameters of the agent, it can ensure that the grid carbon emission reduction method is sufficiently accurate in predicting the output of new energy, and can accurately coordinate the output of each generator to avoid some generators from excessively emitting carbon dioxide at low loads and failing to make full use of renewable energy at high loads. At the same time, the output of the new energy generator can be made highly certain and does not require human intervention and supervision, thereby improving the stability of the power grid.

[0087] Embodiment 2:

[0088] Based on Example 1, the embodiment of the present invention obtains real-time data of different new energy generators and pre-processes the real-time data, such as Figure 2 As shown, including:

[0089] S01: Acquire the generator types of different renewable energy generators, and determine the grid topology and load characteristics of the renewable energy generators according to the generator types;

[0090] S02: Obtain sensors and data acquisition equipment according to the grid topology, load characteristics and data accuracy requirements of the new energy generator;

[0091] S03: Obtain real-time data of different new energy generators based on sensors and data acquisition equipment;

[0092] S04: Analyze and preprocess the real-time data according to edge computing technology, and transmit the preprocessed data to a remote data center based on communication technology.

[0093] In this embodiment, the power grid topology refers to a model that describes the connection relationship between various elements in the power system, and describes the physical connection relationship between various entities in the system, such as generators, substations, transmission lines, distribution lines, load nodes, etc.

[0094] In this embodiment, the generator types of different new energy generators include: solar power generation, hydroelectric power generation, and wind power generation.

[0095] In this embodiment, the load characteristic of the new energy generator refers to the relationship between its power generation and power load within a certain period of time.

[0096] In this embodiment, the real-time data includes: output power, power generation frequency, and load level of the new energy generator.

[0097] In this embodiment, edge computing is a distributed computing model that moves data processing, machine learning, and artificial intelligence functions from the cloud to the edge of the device, which means that data is analyzed close to where it is generated, reducing latency and response time.

[0098] In this embodiment, communication technology refers to the technical means and methods used in the process of information transmission, including: wired communication and wireless communication.

[0099] The beneficial effects of the above technical solution are: according to the grid topology, load characteristics and data accuracy requirements of the new energy generator, sensors and data acquisition equipment are obtained and data is collected, and the data is pre-processed, which can improve the accuracy of the data, ensure the quality of the data, and improve the certainty and coordination of the new energy output.

[0100] Embodiment 3:

[0101] Based on Example 2, the embodiment of the present invention establishes a new energy generator model based on the preprocessed data and the influence of natural factors, including:

[0102] Obtain environmental factors and historical power generation records related to the operation of new energy generators;

[0103] Extract relevant features of environmental factors and historical power generation records based on feature extraction technology;

[0104] Obtain the geographical location information of new energy generators and establish a new geographical topology based on the geographic information system;

[0105] Establishing a new energy generator model based on a new geographic topology according to the relevant features;

[0106] Among them, for weather factors that cannot be accurately simulated, interpolation or quantile regression methods are used for approximation.

[0107] In this embodiment, the environmental factors for the operation of the new energy generator include: temperature, humidity, and wind speed.

[0108] In this embodiment, the geographic topology structure refers to a topology structure used to describe the spatial relationship between entities in a geographic information system.

[0109] In this embodiment, the new energy generator model is used to describe and predict the output characteristics of the new energy generator and the actual power generation of the new energy generator.

[0110] The beneficial effect of the above technical solution is: by establishing a new energy generator model through the environmental factors of the operation of new energy generators, the relevant characteristics of historical power generation records and the geographical topological structure, it is possible to better predict the power generation of new energy generators, thereby determining the output of new energy, better coordinate the uncertainty of new energy output, and reduce carbon emissions from the power grid.

[0111] Embodiment 4:

[0112] Based on Example 3, the embodiment of the present invention constructs a reinforcement learning environment based on a reinforcement learning algorithm according to a new energy generator model. In the constructed environment, reinforcement learning agent training is performed according to a method of dynamically adjusting the learning rate, and a reward function is determined according to the reinforcement learning environment after training, including:

[0113] Determine the state space of the new energy generator according to the input and output of the new energy generator model, and determine the states of the state space and the transition rules between the states according to the state space;

[0114] Get the set of actions taken by the new energy generator in each state;

[0115] Constructing a reinforcement learning environment based on the reinforcement learning algorithm according to the transfer rules and the action set;

[0116] In the constructed environment, the initial learning rate is selected according to the greedy algorithm, and the learning rate is dynamically adjusted according to the value of the loss function and the size of the gradient, and the reinforcement learning agent training is performed according to the adjusted learning rate value;

[0117] The reward function is determined based on the principles of reinforcement learning, the state and behavior of the new energy generator according to the trained reinforcement learning environment.

[0118] In this embodiment, the state space of the new energy generator includes:

[0119] Mechanical status: including generator start and stop, speed, output power, etc.

[0120] Electrical status: including parameters such as voltage, current, and frequency of the power system.

[0121] Control status: includes automatic control instructions of the generator, such as generator start, stop, speed regulation and other commands.

[0122] Dispatching status: refers to the dispatching and management of generators by power grid dispatchers, including the allocation of power generation plans, fault handling, load adjustment, etc.

[0123] In this embodiment, the set of actions taken by the new energy generator in each state refers to the specific actions that the generator will perform under specific circumstances to adjust its output voltage, frequency, power factor and other parameters to ensure its stable operation in the power grid and meet the needs of the power system.

[0124] In this embodiment, the transition rules between various states include:

[0125] Mechanical state transfer rules: The mechanical states of the generator, such as start and stop, speed, and output power, are usually subject to physical laws, such as electromagnetic induction and mechanical design. When these states change, they may trigger transitions to other states. For example, the start of the generator may cause the speed to increase, which in turn causes changes in voltage and current.

[0126] Electrical state transfer rules: The electrical state is affected by factors such as power supply voltage, current, and frequency. When these factors change, the mechanical state, control state, or environmental state of the generator may change. For example, a change in power supply voltage may cause the generator to start, stop, or change in speed.

[0127] In this embodiment, when performing reinforcement learning agent training according to the method of dynamically adjusting the learning rate, it is necessary to first use a certain method to select an initial learning rate, and then dynamically adjust the value of the learning rate according to some indicators (such as the value of the loss function or the size of the gradient) during the training process to keep it within an appropriate range.

[0128] In this embodiment, when constructing the reward function, it is necessary to consider the state of the new energy generator and all possible actions that the new energy generator can take, and determine the reward value corresponding to each action, so that the new energy generator can adjust its behavior according to the reward value to maximize the long-term cumulative reward.

[0129] In this embodiment, reinforcement learning is a learning process based on trial and error, allowing the new energy generator to learn how to make optimal decisions in an environment by interacting with the environment.

[0130] The beneficial effects of the above technical solution are: by constructing a reinforcement learning environment based on the new energy generator model, the best strategy can be automatically formulated through an intelligent algorithm to maximize a certain reward signal and reduce human errors. At the same time, reinforcement learning agent training is carried out according to the method of dynamically adjusting the learning rate, and the reward function is determined according to the reinforcement learning environment after training. Since the dynamic adjustment of the learning rate can dynamically change the learning rate, the adaptability of the learning environment can be improved, the output uncertainty of the new energy generator can be better coordinated, and the efficiency of carbon emission reduction can be improved.

[0131] Embodiment 5:

[0132] Based on Example 4, after obtaining the action set taken by the new energy generator in each state, the embodiment of the present invention further includes:

[0133] Acquire different working states of the new energy generator, and classify the different working states;

[0134] Monitor various status information of new energy generators in real time through telemetry and telesignaling systems, identify status boundaries based on the status information, and determine whether the generator enters or leaves a certain state;

[0135] According to the divided working status, the corresponding action set is parsed based on the status information;

[0136] The parsed action set is organized into an action sequence according to a certain order and rules, the correctness of the action sequence is verified, and the verified action sequence is stored.

[0137] In this embodiment, different working states of the new energy generator include: starting, normal operation, overload protection, and shutdown for maintenance.

[0138] In this embodiment, the state classification includes: operating state and shutdown state.

[0139] In this embodiment, the telemetry system is a technical system for real-time monitoring and transmission of the measured object. It measures physical quantities at a long distance and transmits these measurement results to the monitoring center in real time through wired or wireless means.

[0140] In this embodiment, the remote signaling system is a system for remote transmission and control of information. It transmits the status information of the equipment to the monitoring center in real time by wired or wireless means, and remotely controls the equipment according to the status information.

[0141] In this embodiment, various status information includes: voltage, current, rotation speed, and temperature.

[0142] In this embodiment, the state boundary refers to a critical point or a discontinuity point in a state change process, and describes the location of the transition between states in a system.

[0143] In this embodiment, the corresponding action set is, for example: when the generator is in the startup state, a series of actions may need to be performed, such as preheating, acceleration, speed regulation, etc., while when the generator is in the shutdown maintenance state, only specific troubleshooting actions may need to be performed.

[0144] The beneficial effect of the above technical solution is: by acquiring different working states of the new energy generator and dividing them, parsing the action set corresponding to the working state, generating an action sequence in a certain order, and storing it, the optimal action can be automatically selected according to the working state of the new energy generator equipment without human intervention, thereby greatly improving the intelligence level of the system.

[0145] Embodiment 6:

[0146] Based on Example 5, the embodiment of the present invention determines the difference between the carbon emission reduction target and the actual output according to the reward function, and smoothes the output fluctuation of new energy based on reactive power compensation and energy storage technology according to the difference, including:

[0147] Obtaining historical carbon emission data of new energy generators, and setting carbon emission reduction targets based on the historical carbon emission data and power generation performance of the new energy generators;

[0148] Determine the difference between the carbon emission reduction target and the actual output according to the reward function;

[0149] Obtaining the capacity and connection mode of the capacitor and the inductor of the new energy generator, and determining the phase difference between the current and the voltage according to the capacity and the connection mode;

[0150] Compensating the reactive power of the new energy generator according to the phase difference between the current and the voltage;

[0151] Based on the differences, the output fluctuations of new energy are smoothed out by using compensated new energy generators and energy storage technology.

[0152] In this embodiment, the historical carbon emission data of the new energy generator refers to the carbon emission data generated by different types of renewable energy power generation equipment over a period of time in the past.

[0153] In this embodiment, the power generation performance of the new energy generator mainly refers to its power generation capacity, stability and efficiency under different conditions.

[0154] In this embodiment, the capacitance of the capacitor determines the amount of charge it can hold.

[0155] In this embodiment, the capacity of the inductor determines the strength of its storage capability.

[0156] In this embodiment, the capacitor and the inductor are usually connected in parallel.

[0157] In this embodiment, the phase difference between current and voltage refers to the time difference between the current signal and the voltage signal. In an AC circuit, current and voltage always appear alternately, and their waveforms, amplitudes and frequencies are the same, but their starting points (i.e., starting times) are different.

[0158] In this embodiment, the reactive power generated by the new energy generator refers to the power caused by the phase difference between the electromotive force and the current due to the effects of inductance and capacitance in the power system.

[0159] The beneficial effects of the above technical solution are: the difference between the carbon emission reduction target and the actual output is determined according to the reward function. The reward function can provide a clear quantitative indicator for the carbon emission reduction task, which helps to avoid blind actions and waste of resources and improve the efficiency of the entire carbon emission reduction system. At the same time, since the reward function can directly affect the individual's benefits, it can stimulate the interaction and cooperation between the various parts within the system. The synergy of the entire system will be maximized, thereby improving the overall carbon emission reduction efficiency. At the same time, according to the difference, the output fluctuations of new energy are smoothed based on reactive compensation and energy storage technology, which can effectively reduce the grid voltage fluctuations and instability risks caused by large-scale access to new energy, and improve the stability and reliability of the grid.

[0160] Embodiment 7:

[0161] Based on Example 6, the embodiment of the present invention evaluates the carbon emissions of the power grid according to the output fluctuation of the new energy, modifies the reward function or adjusts the parameters of the agent according to the evaluation results, and coordinates the uncertainty of the new energy output for the carbon emission reduction of the power grid according to the modified and adjusted reinforcement learning environment, including:

[0162] Obtain relevant data on the output fluctuation of new energy within a unit time period;

[0163] Determine the carbon emissions of the power grid based on the current power grid load and carbon emission standards according to the relevant data;

[0164] Compare the carbon emissions of the power grid with the set carbon emission targets, evaluate the carbon emission reduction of the power grid based on the comparison results, and obtain the reasons for not meeting the carbon emission reduction requirements based on the evaluation results;

[0165] The reward function is modified or the parameters of the agent are adjusted according to the reasons, and the uncertainty of the output of new energy sources is coordinated for carbon emission reduction in the power grid according to the modified and adjusted reinforcement learning environment.

[0166] In this embodiment, the carbon emission standard refers to the technical requirements and inspection methods for measuring and controlling the carbon content of emissions from new energy generators.

[0167] In this embodiment, the relevant data includes: time, frequency, duration, and power.

[0168] In this embodiment, adjusting the parameters of the agent may include increasing or decreasing certain strategies to balance the trade-off between carbon emission reduction and social benefits, such as adding some strategies with higher social benefits to improve overall satisfaction, while also trying to reduce the frequency of use of strategies with higher carbon emissions.

[0169] The beneficial effect of the above technical solution is: by comparing the carbon emissions of the power grid with the set carbon emission targets and evaluating the comparison results, modifying the reward function or adjusting the parameters of the agent according to the evaluation results, it helps to ensure that the reinforcement learning environment can effectively coordinate the output of new energy and carbon emission reduction targets under various circumstances, thereby achieving beneficial carbon emission reduction effects.

[0170] Embodiment 8:

[0171] Based on Example 7, after acquiring real-time data of different new energy generators, the embodiment of the present invention further includes:

[0172] Determine the mechanical energy parameters of the new energy generator according to the real-time data, and determine the power generation performance characteristics of the new energy generator based on the mechanical energy parameters;

[0173] Determine the corresponding state function of each working component of the new energy generator according to the power generation performance characteristics;

[0174] Determine the current frequency response parameters of each working component through the state response function, and determine the frequency recovery requirement according to the current frequency response parameters and the preset frequency response parameters of each working component;

[0175] Obtaining the frequency modulation condition of each working component, and determining the frequency modulation incremental loss according to the frequency modulation condition and the frequency recovery requirement of the working component;

[0176] Determine the organic control target value of the new energy generator based on the frequency modulation incremental loss and the initial loss of each working component;

[0177] Determine the working scale load of the new energy generator based on the organic control target value of the new energy generator;

[0178] Determine the coordinated control strategy of multiple working components of the new energy generator through a preset balanced optimization algorithm according to the working scale load, and determine the new energy consumption parameters according to the coordinated control strategy;

[0179] The carbon emission index is determined based on the new energy consumption parameters, and whether the working mode of the new energy generator exceeds the standard is determined according to the carbon emission index. If so, the real-time data of the new energy generator that exceeds the standard in the working mode that does not exceed the standard is re-collected.

[0180] In this embodiment, the mechanical energy parameters of the new energy generator include: rotation speed, vibration amplitude, and temperature.

[0181] In this embodiment, the power generation performance characteristics of the new energy generator include: high power generation efficiency, environmental protection, and renewability.

[0182] In this embodiment, the corresponding state function of each working component of the new energy generator refers to the electrical characteristics or physical behaviors exhibited by each working component under a specific state. The function describes the response characteristics of each component under different conditions, such as:

[0183] Battery state function: The battery state function describes the chemical reactions and changes in physical properties that occur during the battery's charging or discharging process as energy is converted. The battery state function usually includes the battery's charge, open circuit voltage, internal resistance, and temperature parameters.

[0184] Generator output power and speed state function: The generator output power and speed state function describes the changes in the power and speed output of the generator under different loads and speeds, and is related to the generator's electromotive force, current, voltage, mechanical load and other parameters.

[0185] State function of transformer and cable: The state function of transformer and cable describes the working status under different load, temperature, voltage and other conditions, including transformer loss, efficiency, short circuit, overload and other parameters.

[0186] In this embodiment, the current frequency response parameter of each working component refers to the degree of response of each working component to excitation signals of different frequencies during the working process, reflecting the relative phase relationship between the output signal and the input excitation signal of the working component after being subjected to external excitation of different frequencies.

[0187] In this embodiment, the frequency recovery requirement means that when a fault or disturbance occurs in the system, the generator needs to quickly adjust its output frequency to maintain the stable operation of the power grid.

[0188] In this embodiment, the frequency modulation conditions of the working component include: frequency deviation, power regulation, and load variation.

[0189] In this embodiment, the frequency modulation incremental loss refers to the energy loss caused by the flow of reactive power in the circuit during the process of the generator changing the output frequency.

[0190] In this embodiment, the organic control target value of the new energy generator refers to various performance indicators and restriction conditions set under the organic control strategy to achieve precise control of the generator, including: output power, output frequency, and reactive power.

[0191] In this embodiment, the working scale load of the new energy generator refers to the maximum load that the generator can withstand under specific operating conditions.

[0192] In this embodiment, the preset equilibrium optimization algorithm is a method for solving multivariable optimization problems and finding a set of optimal solutions under the premise of presetting certain constraints and target values.

[0193] In this embodiment, the new energy consumption parameters refer to various parameters used to guide the consumption of new energy within a certain period of time, including:

[0194] Power supply side parameters: including the output curve, frequency response, voltage stability and power transmission loss of the new energy generator.

[0195] Load side parameters: including peak and valley periods of load, maximum load, peak and valley difference, and spare capacity.

[0196] Market parameters: including the price mechanism, trading rules and power demand response plan of the electricity market.

[0197] The beneficial effects of the above technical solution are: determining the carbon emission index based on the new energy consumption parameters, and determining whether the working mode of the new energy generator exceeds the standard according to the carbon emission index, which can reduce the carbon emissions of the new energy generator while ensuring the efficient operation of the new energy generator, which is in line with the development direction of the low-carbon economy. Furthermore, if the standard is exceeded, the real-time data of the new energy generator that exceeds the standard in the working mode that does not exceed the standard is re-collected, so as to have a more comprehensive understanding of the actual operation of the new energy generator, thereby more accurately evaluating the performance and emissions of the new energy generator. At the same time, it helps to optimize the control strategy of the new energy generator and improve the operating efficiency and economy of the new energy generator.

[0198] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0199] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for reducing carbon emissions in power grids based on coordination of uncertainty in new energy output, characterized in that: include: Step 1: Acquire real-time data of different new energy generators and pre-process the real-time data; Step 2: Establish a new energy generator model based on the influence of natural factors according to the preprocessed data; Step 3: Construct a reinforcement learning environment based on the reinforcement learning algorithm according to the new energy generator model. In the constructed environment, train the reinforcement learning agent according to the method of dynamically adjusting the learning rate, and determine the reward function according to the reinforcement learning environment after training; Step 4: Determine the difference between the carbon emission reduction target and the actual output according to the reward function, and smooth the output fluctuation of the new energy based on the reactive power compensation and energy storage technology according to the difference; Step 5: Evaluate the carbon emissions of the power grid based on the output fluctuations of renewable energy, modify the reward function or adjust the parameters of the agent based on the evaluation results, and coordinate the uncertainty of renewable energy output for carbon emission reduction in the power grid based on the modified and adjusted reinforcement learning environment.

2. The method for reducing carbon emissions in power grids based on coordination of uncertainty in new energy output according to claim 1 is characterized in that: Acquire real-time data of different renewable energy generators and pre-process the real-time data, including: Acquire the generator types of different renewable energy generators, and determine the grid topology and load characteristics of the renewable energy generators according to the generator types; Obtain sensors and data acquisition equipment based on the grid topology, load characteristics, and data accuracy requirements of renewable energy generators; Obtain real-time data of different renewable energy generators based on sensors and data acquisition equipment; The real-time data is analyzed and preprocessed according to edge computing technology, and the preprocessed data is transmitted to a remote data center based on communication technology.

3. The method for reducing carbon emissions in power grids based on coordination of uncertainty in new energy output according to claim 1 is characterized in that: According to the pre-processed data, a new energy generator model is established based on the influence of natural factors, including: Obtain environmental factors and historical power generation records related to the operation of new energy generators; Extract relevant features of environmental factors and historical power generation records based on feature extraction technology; Obtain the geographical location information of new energy generators and establish a new geographical topology based on the geographic information system; Establishing a new energy generator model based on a new geographic topology according to the relevant features; Among them, for weather factors that cannot be accurately simulated, interpolation or quantile regression methods are used for approximation.

4. The method for reducing carbon emissions in power grids based on coordination of uncertainty in new energy output according to claim 1 is characterized in that: According to the new energy generator model, a reinforcement learning environment is constructed based on the reinforcement learning algorithm. In the constructed environment, the reinforcement learning agent is trained according to the method of dynamically adjusting the learning rate, and the reward function is determined according to the reinforcement learning environment after training, including: Determine the state space of the new energy generator according to the input and output of the new energy generator model, and determine the states of the state space and the transition rules between the states according to the state space; Get the set of actions taken by the new energy generator in each state; Constructing a reinforcement learning environment based on the reinforcement learning algorithm according to the transfer rules and the action set; In the constructed environment, the initial learning rate is selected according to the greedy algorithm, and the learning rate is dynamically adjusted according to the value of the loss function and the size of the gradient, and the reinforcement learning agent training is performed according to the adjusted learning rate; The reward function is determined based on the principles of reinforcement learning, the state and behavior of the new energy generator according to the trained reinforcement learning environment.

5. The method for reducing carbon emissions in power grids based on coordination of uncertainty in new energy output according to claim 4 is characterized in that: After obtaining the set of actions taken by the new energy generator in each state, it also includes: Acquire different working states of the new energy generator, and classify the different working states; Monitor various status information of new energy generators in real time through telemetry and telesignaling systems, identify status boundaries based on the status information, and determine whether the generator enters or leaves a certain state; According to the divided working status, the corresponding action set is parsed based on the status information; The parsed action set is organized into an action sequence according to a certain order and rules, the correctness of the action sequence is verified, and the verified action sequence is stored.

6. The method for reducing carbon emissions in power grids based on coordination of uncertainty in new energy output according to claim 1 is characterized in that: The difference between the carbon emission reduction target and the actual output is determined according to the reward function, and the output fluctuation of new energy is smoothed based on the difference using reactive power compensation and energy storage technology, including: Obtaining historical carbon emission data of new energy generators, and setting carbon emission reduction targets based on the historical carbon emission data and power generation performance of the new energy generators; Determine the difference between the carbon emission reduction target and the actual output according to the reward function; Obtaining the capacity and connection mode of the capacitor and the inductor of the new energy generator, and determining the phase difference between the current and the voltage according to the capacity and the connection mode; Compensating the reactive power of the new energy generator according to the phase difference between the current and the voltage; Based on the differences, the output fluctuations of new energy are smoothed out by using compensated new energy generators and energy storage technology.

7. The method for reducing carbon emissions in power grids based on coordination of uncertainty in new energy output according to claim 1 is characterized in that: Evaluate the carbon emissions of the power grid according to the output fluctuations of renewable energy, modify the reward function or adjust the agent parameters according to the evaluation results, and coordinate the uncertainty of renewable energy output for carbon emission reduction in the power grid according to the modified and adjusted reinforcement learning environment, including: Obtain relevant data on the output fluctuation of new energy within a unit time period; Determine the carbon emissions of the power grid based on the current power grid load and carbon emission standards according to the relevant data; Compare the carbon emissions of the power grid with the set carbon emission targets, evaluate the carbon emission reduction of the power grid based on the comparison results, and obtain the reasons for not meeting the carbon emission reduction requirements based on the evaluation results; The reward function is modified or the parameters of the agent are adjusted according to the reasons, and the uncertainty of the output of new energy sources is coordinated for carbon emission reduction in the power grid according to the modified and adjusted reinforcement learning environment.

8. The method for reducing carbon emissions in power grids based on coordination of uncertainty in new energy output according to claim 1 is characterized in that: After obtaining real-time data of different new energy generators, it also includes: Determine the mechanical energy parameters of the new energy generator according to the real-time data, and determine the power generation performance characteristics of the new energy generator based on the mechanical energy parameters; Determine the corresponding state function of each working component of the new energy generator according to the power generation performance characteristics; Determine the current frequency response parameters of each working component through the state response function, and determine the frequency recovery requirement according to the current frequency response parameters and the preset frequency response parameters of each working component; Obtaining the frequency modulation condition of each working component, and determining the frequency modulation incremental loss according to the frequency modulation condition and the frequency recovery requirement of the working component; Determine the organic control target value of the new energy generator based on the frequency modulation incremental loss and the initial loss of each working component; Determine the working scale load of the new energy generator based on the organic control target value of the new energy generator; Determine the coordinated control strategy of multiple working components of the new energy generator through a preset balanced optimization algorithm according to the working scale load, and determine the new energy consumption parameters according to the coordinated control strategy; The carbon emission index is determined based on the new energy consumption parameters, and whether the working mode of the new energy generator exceeds the standard is determined according to the carbon emission index. If so, the real-time data of the new energy generator that exceeds the standard in the working mode that does not exceed the standard is re-collected.

Images