Central air conditioner green power consumption optimization control system based on carbon footprint tracing

Abstract

The invention relates to the technical field of building power system energy saving optimization, and discloses a central air conditioner green power consumption optimization control system based on carbon footprint tracing. The system comprises a carbon footprint traceability data acquisition unit, a green power consumption priority evaluation unit, an energy scheduling optimization unit, a real-time load matching unit and a carbon footprint verification unit. The carbon footprint traceability data acquisition unit acquires related data; the green power consumption priority evaluation unit generates a consumption strategy sequence based on a dynamic game theory model; the energy scheduling optimization unit adopts a specific algorithm to construct an optimization model; the real-time load matching unit adjusts a matching relationship by using deep reinforcement learning; the carbon footprint verification unit verifies the carbon footprint based on a zero-knowledge proof algorithm. According to the method, optimal control over green power consumption of the central air conditioner is achieved, the energy utilization efficiency is improved, carbon emission is reduced, and carbon footprints are accurately managed.

Description

Technical Field

[0001] The present invention relates to the technical field of energy-saving optimization of building power systems, specifically to an optimized control system for green power consumption of central air conditioners based on carbon footprint traceability. Background Art

[0002] Under the background of the global advocacy of energy conservation, emission reduction and sustainable development, the energy consumption and carbon emissions in the building field have attracted much attention. As a "big consumer" of building energy consumption, the energy utilization efficiency and carbon emissions of central air conditioners directly affect the green development level of buildings. During the operation of traditional central air conditioners, there is a lack of effective management for the consumption of green power, resulting in problems such as energy waste and high carbon emissions, and it is difficult to meet the current strict requirements for low-carbon buildings.

[0003] From the perspective of energy supply, the power grid energy structure is complex and diverse, including various types such as thermal power, hydropower, wind power, and photovoltaic power. The carbon emission intensities of different energies vary greatly. At present, when obtaining the carbon emission intensity data of the power grid energy, there is often a lack of accurate dynamic acquisition methods, and it is impossible to timely and accurately reflect the impact of the change in the real-time power supply ratio of different energy types on the carbon emission intensity. For example, in some areas, with the intermittent fluctuations in the supply of green power such as wind power and photovoltaic power, the traditional acquisition methods cannot quickly adapt to this change, resulting in data lag and making it difficult to support subsequent precise energy scheduling decisions.

[0004] The proportion of green power consumption also faces problems in monitoring and management. On the one hand, the supply of green power is unevenly distributed in time and space, affected by natural factors such as weather and seasons, as well as energy policies. On the other hand, existing technologies are difficult to effectively evaluate the priority of green power consumption at different times and cannot formulate reasonable green power consumption strategies. For example, during peak electricity consumption periods, due to the lack of scientific judgment on the priority of green power consumption, there may be situations of green power waste or excessive reliance on high-carbon energy, which not only reduces the utilization efficiency of green power but also increases carbon emissions.

[0005] In terms of air-conditioning load demand, it has significant spatio-temporal distribution characteristics. The air-conditioning load demands in different building functional areas (such as office buildings, shopping malls, hospitals, etc.) vary significantly in time. For example, the demand in office buildings is high during the day on weekdays and low at night; the demand in shopping malls is strong on weekends and holidays. The traditional load matching methods do not fully consider these characteristics and cannot achieve dynamic and precise matching between the proportion of green power consumption and the air-conditioning load. This not only causes energy waste but may also affect the operation stability and comfort of the air-conditioning system.

[0006] In addition, during the process of green electricity consumption, there is a lack of effective technical means for monitoring and verifying carbon footprints. It is impossible to accurately verify the consistency between the actual carbon footprint and the theoretical predicted value, making it difficult to ensure the authenticity and reliability of carbon footprint data. Moreover, the existing data management methods cannot form an immutable carbon footprint traceability record, which is not conducive to the supervision of carbon emissions and the tracing of responsibilities. For example, when enterprises conduct carbon trading or respond to environmental protection reviews, they may face problems with insufficient data credibility due to the lack of reliable carbon footprint traceability records. Summary of the Invention

[0007] The purpose of the present invention is to provide an optimized control system for green electricity consumption of central air conditioners based on carbon footprint traceability to solve the problems raised in the above background technology.

[0008] To achieve the above purpose, the present invention provides the following technical solution: An optimized control system for green electricity consumption of central air conditioners based on carbon footprint traceability, the system includes: A carbon footprint traceability data acquisition unit, which is used to obtain real-time grid energy carbon emission intensity data, green electricity supply ratio data, and air-conditioning load demand data associated with the central air-conditioning system through a dynamic weighted acquisition algorithm; A green electricity consumption priority evaluation unit, which is used to quantitatively sort the multi-period green electricity consumption priorities based on a dynamic game theory model and generate a green electricity consumption strategy sequence; An energy scheduling optimization unit, which is used to construct an optimized model for green electricity consumption by using an adaptive particle swarm algorithm with carbon footprint constraints according to the green electricity consumption strategy sequence and real-time grid energy data; A real-time load matching unit, which is used to dynamically adjust the matching relationship between the green electricity consumption ratio and the air-conditioning load according to the optimized model for green electricity consumption and the operating parameters of the central air conditioner through a deep reinforcement learning strategy; A carbon footprint verification unit, which is used to verify the consistency between the actual carbon footprint and the theoretical predicted value during the process of green electricity consumption based on a zero-knowledge proof algorithm and output a carbon footprint verification result; Among them, the execution steps of the green electricity consumption priority evaluation unit include: Construct a multi-period green electricity consumption game matrix and define the profit functions of green electricity suppliers, grid operators, and user entities; Iteratively optimize the green electricity consumption priority weights for each period through a Nash equilibrium solution algorithm; Generate the green electricity consumption strategy sequence including period priorities and green electricity allocation ratios according to the game results.

[0009] Preferably, the execution steps of the carbon footprint traceability data acquisition unit include: Divide the carbon emission weight factors according to the grid energy types and dynamically collect the real-time power supply ratios of different energy types; Based on the sliding time window algorithm, perform multi-scale smoothing processing on the green electricity supply ratio data to generate a normalized benchmark value for green electricity consumption and absorption; Integrate the spatio-temporal distribution characteristics of air-conditioning load demand data to construct a dynamically weighted carbon footprint traceability data set; Preferably, the execution steps of the carbon footprint traceability data acquisition unit further include: Divide the carbon emission intensity sub-regions according to the power grid regions, and use distributed data acquisition nodes to synchronously obtain the real-time energy type data of each sub-region; Perform noise elimination on the collected data through the Kalman filtering algorithm to generate high-confidence carbon footprint traceability data.

[0010] Preferably, the execution steps of the green electricity consumption and absorption priority evaluation unit further include: Introduce a time decay factor to dynamically adjust the influence weight of historical green electricity consumption and absorption data; Construct a Pareto front solution set for multi-objective games, and screen the optimal green electricity consumption and absorption strategy sequence through the entropy weight method.

[0011] Preferably, the execution steps of the energy scheduling optimization unit include: Take the minimization of the carbon footprint as the objective function, and take the power grid stability, green electricity consumption and absorption ratio, and air-conditioning load fluctuation as constraints to establish a multi-objective optimization problem; Adopt an adaptive particle swarm optimization algorithm to dynamically adjust the inertia weight and search range of the particle swarm, and solve the green electricity consumption and absorption optimization model; Preferably, the execution steps of the energy scheduling optimization unit further include: Embed a carbon footprint sensitivity analysis module in the adaptive particle swarm optimization algorithm to dynamically identify key constraints; Initialize the particle swarm position through the chaotic mapping algorithm to avoid local optimal solutions.

[0012] Preferably, the execution steps of the real-time load matching unit include: Construct a deep reinforcement learning network, with the air-conditioning operation parameters, environmental temperature, and green electricity consumption and absorption ratio as the state space, and the load regulation instruction as the action space; Update the policy network parameters through the Q-learning algorithm to generate real-time green electricity consumption and absorption and load matching instructions; Preferably, the execution steps of the real-time load matching unit further include: Use a long short-term memory network to predict the air-conditioning load demand in the future period as the input feature of the deep reinforcement learning strategy; Optimize the exploration and exploitation balance of the action space through the twin-delayed deep deterministic policy gradient algorithm.

[0013] Preferably, the execution steps of the carbon footprint verification unit include: Encoding the actual carbon footprint data and the theoretical prediction value into hash values to generate commitment parameters for zero-knowledge proof; Verifying the consistency of the commitment parameters through a non-interactive zero-knowledge proof protocol and outputting the verification result.

[0014] Preferably, the execution steps of the carbon footprint verification unit further include: binding the verification result to the blockchain ledger to generate an immutable carbon footprint traceability record; aggregating multi-source data through a Merkle tree structure.

[0015] Compared with the prior art, the beneficial effects of the present invention are: The carbon footprint traceability data acquisition unit of the present invention can accurately obtain power grid energy carbon emission intensity data, green power supply ratio data, and air-conditioning load demand data through advanced technologies such as dynamic weighted acquisition algorithm, sliding time window algorithm, and Kalman filtering algorithm. By dividing the carbon emission weight factor according to the power grid energy type and dynamically collecting the real-time power supply ratio of different energy types, it can timely reflect the impact of energy structure changes on carbon emission intensity; perform multi-scale smoothing processing on the green power supply ratio data to generate a normalized green power consumption benchmark value, providing a stable and reliable data basis for the formulation of green power consumption strategies; integrate the spatio-temporal distribution characteristics of air-conditioning load demand data to construct a dynamically weighted carbon footprint traceability data set, making the data more in line with the actual load change situation. At the same time, the application of distributed data acquisition nodes and Kalman filtering algorithm effectively eliminates data noise, improves the confidence of data, and ensures the accuracy of subsequent decisions.

[0016] Based on the dynamic game theory model, the green power consumption priority evaluation unit comprehensively considers the interests of green power suppliers, power grid operators, and user entities. By constructing a multi-period green power consumption game matrix and defining a revenue function, it uses the Nash equilibrium solution algorithm to iteratively optimize the green power consumption priority weights in each period to generate a reasonable sequence of green power consumption strategies. Introducing a time decay factor to dynamically adjust the influence weight of historical green power consumption data makes the decision more timely; constructing the Pareto front solution set of multi-objective games and screening the optimal sequence of green power consumption strategies through the entropy weight method realizes the balanced optimization of multiple objectives, improves the green power consumption efficiency, and promotes the rational allocation of energy.

[0017] The energy dispatch optimization unit takes the minimization of carbon footprint as the objective function, combines constraints such as grid stability, green power consumption ratio, and air-conditioning load fluctuation, and establishes a multi-objective optimization problem. An adaptive particle swarm optimization algorithm with carbon footprint constraints is used to dynamically adjust the inertia weight and search range of the particle swarm, effectively solving the green power consumption optimization model. A carbon footprint sensitivity analysis module is embedded in the algorithm, which can dynamically identify key constraint conditions and improve the optimization efficiency; the position of the particle swarm is initialized through the chaotic mapping algorithm, avoiding local optimal solutions, making the energy dispatch more scientific and reasonable, minimizing the carbon footprint to the greatest extent under the premise of meeting various constraints, and improving the energy utilization efficiency.

[0018] The real-time load matching unit uses a deep reinforcement learning strategy to dynamically adjust the matching relationship between the green power consumption ratio and the air-conditioning load according to the green power consumption optimization model and the operating parameters of the central air-conditioning. A deep reinforcement learning network is constructed, with the operating parameters of the air-conditioning, environmental temperature, and green power consumption ratio as the state space, and the load regulation instruction as the action space. The parameters of the policy network are updated through the Q-learning algorithm to generate real-time green power consumption and load matching instructions. A long short-term memory network is used to predict the air-conditioning load demand in the future period as the input feature of the deep reinforcement learning strategy, providing a basis for load matching in advance; the exploration and utilization balance of the action space is optimized through the double delayed deep deterministic policy gradient algorithm, further improving the accuracy and stability of load matching and ensuring the efficient operation of the central air-conditioning system.

[0019] The carbon footprint verification unit is based on the zero-knowledge proof algorithm to verify the consistency between the actual carbon footprint and the theoretical prediction value in the process of green power consumption. The actual carbon footprint data and the theoretical prediction value are encoded as hash values to generate the commitment parameters of the zero-knowledge proof, and the consistency of the commitment parameters is verified through the non-interactive zero-knowledge proof protocol, ensuring the reliability of the verification results. The verification results are bound to the blockchain ledger to generate an immutable carbon footprint traceability record, and multi-source data is aggregated through the Merkle tree structure, realizing the effective supervision and responsibility traceability of the carbon footprint, enhancing the credibility and data security of the system, and providing strong support for the carbon management and carbon trading of enterprises. Brief Description of the Drawings

[0020] Figure 1 It is the working principle diagram of the central air-conditioning green power consumption optimization control system based on carbon footprint traceability described in the present invention; Figure 2 It is the regional data processing and noise reduction flow chart of the carbon footprint traceability data acquisition unit; Figure 3 It is the optimization strategy flow chart of the green power consumption priority evaluation unit; Figure 4 It is the basic flow chart of deep reinforcement learning of the real-time load matching unit. Detailed implementation manners

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0022] Please refer to Figures 1-4 , the present invention provides an optimized control system for green electricity consumption of central air conditioners based on carbon footprint traceability, and its overall implementation scheme is as follows: Carbon footprint traceability data acquisition unit: This unit obtains the carbon emission intensity data of the power grid energy associated with the central air-conditioning system, the green electricity supply ratio data, and the air-conditioning load demand data in real time through a dynamic weighted acquisition algorithm. Its working principle is to divide the carbon emission weight factors according to the power grid energy types, and dynamically collect the real-time power supply ratios of different energy types; based on the sliding time window algorithm, perform multi-scale smoothing processing on the green electricity supply ratio data to generate a normalized green electricity consumption benchmark value; integrate the spatio-temporal distribution characteristics of the air-conditioning load demand data to construct a dynamically weighted carbon footprint traceability data set. At the same time, divide the carbon emission intensity sub-regions according to the power grid regions, use distributed data acquisition nodes to synchronously obtain the real-time energy type data of each sub-region, and eliminate the noise of the collected data through the Kalman filter algorithm to generate high-confidence carbon footprint traceability data.

[0023] Green electricity consumption priority evaluation unit: Based on the dynamic game theory model, quantitatively rank the green electricity consumption priorities in multiple time periods to generate a green electricity consumption strategy sequence. The specific implementation steps include constructing a multi-time period green electricity consumption game matrix and defining the profit functions of the green electricity suppliers, power grid operators, and user entities; through the Nash equilibrium solution algorithm, iteratively optimize the green electricity consumption priority weights in each time period; generate a green electricity consumption strategy sequence including time period priorities and green electricity allocation ratios according to the game results. In addition, a time decay factor is introduced to dynamically adjust the influence weight of historical green electricity consumption data; construct a Pareto front solution set for multi-objective games, and screen the optimal green electricity consumption strategy sequence through the entropy weight method.

[0024] Energy Dispatch Optimization Unit: An adaptive particle swarm algorithm with carbon footprint constraints is adopted. According to the green power consumption strategy sequence and real-time power grid energy data, a green power consumption optimization model is constructed. The execution steps are as follows: taking the minimization of carbon footprint as the objective function, and taking grid stability, green power consumption ratio and air-conditioning load fluctuation as constraints, a multi-objective optimization problem is established; an adaptive particle swarm algorithm is used to dynamically adjust the inertia weight and search range of the particle swarm to solve the green power consumption optimization model. At the same time, a carbon footprint sensitivity analysis module is embedded in the adaptive particle swarm algorithm to dynamically identify key constraints; the position of the particle swarm is initialized through the chaotic mapping algorithm to avoid local optimal solutions.

[0025] Real-time Load Matching Unit: Through a deep reinforcement learning strategy, according to the green power consumption optimization model and the operating parameters of the central air-conditioning, the matching relationship between the green power consumption ratio and the air-conditioning load is dynamically adjusted. The specific steps are as follows: a deep reinforcement learning network is constructed, with the air-conditioning operating parameters, ambient temperature and green power consumption ratio as the state space, and the load regulation instruction as the action space; the parameters of the policy network are updated through the Q-learning algorithm to generate real-time green power consumption and load matching instructions. In addition, a long short-term memory network is used to predict the air-conditioning load demand in the future period as the input feature of the deep reinforcement learning strategy; the exploration and exploitation balance of the action space is optimized through the double delayed deep deterministic policy gradient algorithm.

[0026] Carbon Footprint Verification Unit: Based on the zero-knowledge proof algorithm, the consistency between the actual carbon footprint and the theoretical prediction value in the process of green power consumption is verified, and the carbon footprint verification result is output. Its execution steps include encoding the actual carbon footprint data and the theoretical prediction value into hash values to generate the commitment parameters of the zero-knowledge proof; through the non-interactive zero-knowledge proof protocol, the consistency of the commitment parameters is verified, and the verification result is output. At the same time, the verification result is bound to the blockchain ledger to generate an immutable carbon footprint traceability record; the multi-source data is aggregated through the Merkle tree structure.

[0027] The following further illustrates the implementation of the present invention in conjunction with Embodiments 1 to 5.

[0028] Embodiment 1: The specific method for data collection and processing of the carbon footprint traceability data collection unit is as follows: By accurately dividing the carbon emission weight factor, smoothing the green power supply ratio data, and eliminating the noise of the collected data, a high-quality data basis is provided for subsequent system decision-making, ensuring the accuracy of the system for the green power consumption and carbon footprint management of the central air-conditioning.

[0029] The carbon footprint traceability data acquisition unit plays the role of a data cornerstone in the entire system. When collecting data on the carbon emission intensity of grid energy, the proportion of green power supply, and the air-conditioning load demand, the carbon emission weight factors are first divided according to the types of grid energy. For example, for thermal power, different carbon emission weight factors are assigned according to its different power generation methods (such as coal power generation, natural gas power generation, etc.). The carbon emission weight factor for coal power generation is relatively high, while that for natural gas power generation is relatively low. By real-time monitoring the operating parameters of power generation equipment of different energy types, such as the consumption of coal and the use of natural gas, the real-time power supply ratio of different energy types is dynamically collected.

[0030] When processing the data on the proportion of green power supply, multi-scale smoothing processing is carried out based on the sliding time window algorithm. The size of the sliding time window is set according to the actual situation. Suppose it is set to 15 minutes, that is, each 15 minutes is a window. Within each window, smoothing calculations such as weighted averaging are performed on the data of the proportion of green power supply. For example, for the data of the proportion of green power supply [0.3, 0.32, 0.28] within the current window, through weighted averaging calculation (assuming the weights are 0.4, 0.3, and 0.3 respectively), the smoothed value of the proportion of green power supply is , generating a normalized benchmark value for green power consumption. This process can effectively remove the noise fluctuations in the data, making the data on the proportion of green power supply more stable and valuable for reference.

[0031] For the air-conditioning load demand data, its spatio-temporal distribution characteristics are integrated. The air-conditioning load data in different time periods and different regions are collected. For example, in the office area, the load demand is relatively high during the day on weekdays and low at night; in the shopping mall area, the load demand is higher on weekends and holidays than on weekdays. According to these characteristics, a dynamically weighted carbon footprint traceability data set is constructed to make the data better reflect the actual load changes.

[0032] To further improve the accuracy of the data, the carbon emission intensity sub-regions are divided according to the grid regions. The entire grid is divided into multiple sub-regions according to factors such as geographical regions and energy structures. Distributed data acquisition nodes are set in each sub-region. These nodes synchronously obtain the real-time energy type data of each sub-region, such as the power generation data of thermal power, hydropower, wind power, etc. in a certain sub-region. The Kalman filter algorithm is used to eliminate the noise in the collected data. The Kalman filter algorithm is based on the state equation and observation equation of the system, and makes an optimal estimate of the observed data containing noise. Suppose the state equation of the system is , and the observation equation is , where is the state of the system at time k, , , are system parameters, is the control input, and They are process noise and observation noise respectively. Through the iterative calculation of the Kalman filtering algorithm, the noise in the data can be effectively removed, generating carbon footprint traceability data with high confidence and providing reliable data support for the work of subsequent units.

[0033] Example 2: By introducing a time decay factor and constructing a Pareto front solution set, the green power consumption priority evaluation unit more reasonably considers the influence of historical data, screens out the optimal green power consumption strategy sequence from the perspective of multi-objective game, making the green power consumption priority evaluation more scientific and forward-looking, and improving the overall green power consumption efficiency of the system.

[0034] In the green power consumption priority evaluation unit, a time decay factor is introduced to dynamically adjust the influence weight of historical green power consumption data. The setting of the time decay factor is based on the passage of time, giving lower weights to earlier historical data, so that recent data has a greater impact on current decisions. Assume the time decay factor is , whose value range is between 0 and 1. As time increases, the weight of historical data decays exponentially. For example, for the historical green power consumption data in n time periods before time t , its weight in the current evaluation is . This can make the evaluation results better reflect the current green power consumption situation and trends, and avoid decision-making lag caused by the excessive influence of historical data.

[0035] Construct a Pareto front solution set for multi-objective game, considering multiple objectives such as green power suppliers, grid operators, and user entities. Green power suppliers hope to maximize the sales revenue of green power, grid operators expect to maintain the stable operation of the grid, and user entities pursue lower electricity costs and good air-conditioning usage experiences. In the game process, multiple possible green power consumption strategies are obtained through the Nash equilibrium solution algorithm, and these strategies constitute the Pareto front solution set. In this solution set, each strategy balances the interests of each entity to varying degrees.

[0036] The optimal green power consumption strategy sequence is screened by the entropy weight method. The entropy weight method is a method to determine weights based on the dispersion degree of data. For each strategy in the Pareto front solution set, calculate its entropy value under different objectives. Assume there are m strategies under a certain objective, and the value of the i-th strategy under this objective is , then the entropy value is , where Calculate the weight of each objective according to the entropy value. The smaller the entropy value, the greater the dispersion degree of the data under this objective, the greater the impact on the decision-making, and the higher its weight. By synthesizing the weights of each objective, the optimal green power consumption strategy sequence is screened to achieve a better green power consumption effect under the constraints of multiple objectives for the system.

[0037] Embodiment 3: In this embodiment, for the energy dispatch optimization unit, the process of constructing a multi-objective optimization problem and solving the green power consumption optimization model using the adaptive particle swarm optimization algorithm is described in detail. At the same time, the method of embedding the carbon footprint sensitivity analysis module and initializing the particle swarm position using the chaotic mapping algorithm is introduced, which helps to improve the optimization effect of energy dispatch and ensure the minimization of carbon footprint under the satisfaction of multiple constraint conditions.

[0038] The energy dispatch optimization unit takes the minimization of carbon footprint as the core objective to construct a multi-objective optimization problem. The objective function is , where represents the carbon footprint. The constraint conditions include grid stability, green power consumption ratio, and air-conditioning load fluctuation. Grid stability is ensured by restricting parameters such as grid voltage deviation and frequency deviation. For example, it is stipulated that the grid voltage deviation is within , and the frequency deviation is within . A certain lower limit is set for the green power consumption ratio, assumed to be , to ensure that the consumption amount of green power reaches a certain standard. The air-conditioning load fluctuation is controlled by restricting the load change rate, such as stipulating that the load change rate is within .

[0039] Use the adaptive particle swarm optimization algorithm to solve the green power consumption optimization model. In the particle swarm optimization algorithm, each particle represents a possible solution, and its position and velocity are continuously updated. In the adaptive particle swarm optimization algorithm, the inertia weight and search range of the particle swarm are dynamically adjusted. The inertia weight is adjusted according to the fitness value of the particle. When the fitness value of the particle is good, the inertia weight is appropriately increased to make it more inclined to global search; when the fitness value is poor, the inertia weight is decreased to strengthen the local search ability. Assume that the adjustment formula of the inertia weight is , where , are the maximum and minimum values of the inertia weight respectively, , are the maximum and minimum fitness values in the current population, and is the fitness value of the current particle.

[0040] Embed a carbon footprint sensitivity analysis module in the adaptive particle swarm optimization algorithm. This module dynamically identifies key constraint conditions by performing perturbation analysis on the parameters in the model. For example, by changing the constraint value of the green electricity consumption ratio and observing the change in the carbon footprint objective function. If a small change in the green electricity consumption ratio causes a large change in the carbon footprint, then the green electricity consumption ratio is a key constraint condition. By identifying key constraint conditions, more attention can be paid to these conditions during the optimization process, improving the optimization efficiency.

[0041] Initialize the particle swarm position through the chaotic mapping algorithm. The chaotic mapping algorithm has the characteristics of randomness and ergodicity, which can prevent the particle swarm from concentrating in a local area during initialization, thus avoiding local optimal solutions. For example, the Logistic chaotic mapping is adopted. , where takes values in the range of (3.5699456, 4], takes values in the range of (0, 1). By iterating the Logistic chaotic mapping multiple times, a series of chaotic sequences are generated, and these sequences are mapped to the solution space of the problem as the initial positions of the particle swarm, providing a wider starting point for subsequent optimization searches.

[0042] Example 4: This example mainly elaborates on the working process of the real-time load matching unit. By constructing a deep reinforcement learning network, using the Q-learning algorithm and the twin-delayed deep deterministic policy gradient algorithm, combined with the long short-term memory network to predict future load demands, the precise dynamic matching of the green electricity consumption ratio and the air-conditioning load is achieved, improving the energy utilization efficiency and ensuring the stable operation of the central air-conditioning system.

[0043] The real-time load matching unit constructs a deep reinforcement learning network, with the air-conditioning operating parameters (such as the air-conditioning temperature set value, compressor speed, etc.), ambient temperature, and green electricity consumption ratio as the state space, and the load adjustment command (such as increasing or decreasing the air-conditioning load) as the action space. In this network, the policy network parameters are updated through the Q-learning algorithm. The core formula of the Q-learning algorithm is , where is the Q-value of executing action in state , is the learning rate, is the reward obtained after executing action , is the discount factor, is the new state after executing action , is in the new state The maximum Q value below. By continuously performing actions and updating the Q value according to the rewards, an optimal load regulation strategy is gradually learned.

[0044] Use a long short-term memory network to predict the air-conditioning load demand in future periods. The long short-term memory network (LSTM) can effectively handle the long-term dependence problems in time series data. Input the historical air-conditioning load data into the LSTM network, and the network learns the time features and trends in the data to predict the air-conditioning load demand in future periods. For example, take the hourly air-conditioning load data of the past 24 hours as the input, and the LSTM network outputs the predicted value of the air-conditioning load for the next 1 hour. This predicted value is used as an input feature for the deep reinforcement learning strategy, enabling the system to adjust the green electricity consumption ratio in advance according to the predicted load demand and achieve more accurate load matching.

[0045] Optimize the balance between exploration and exploitation of the action space through the twin-delayed deep deterministic policy gradient algorithm. In deep reinforcement learning, it is necessary to find a balance between exploring new actions and exploiting existing experience. The twin-delayed deep deterministic policy gradient algorithm makes the policy update more stable by introducing two networks with delayed updates (the target network and the evaluation network). During the training process of this algorithm, by adjusting the magnitude and decay rate of the exploration noise, the randomness of exploration is gradually reduced, and more learned experience is used for decision-making, thereby improving the accuracy and stability of load matching.

[0046] Example 5: This example is used to describe the workflow of the carbon footprint verification unit. By encoding the actual carbon footprint data and the theoretical prediction value as hash values and using the zero-knowledge proof protocol for verification, and binding the verification result to the blockchain ledger and aggregating data using the Merkle tree structure, reliable verification of the carbon footprint in the process of green electricity consumption and an immutable traceability record are achieved, enhancing the credibility and data security of the system.

[0047] The carbon footprint verification unit encodes the actual carbon footprint data and the theoretical prediction value as hash values to generate the commitment parameters of the zero-knowledge proof. The hash function has the characteristics of one-wayness and uniqueness, and can convert data of any length into a hash value of a fixed length. For example, use the SHA-256 hash function to calculate the hash values for the actual carbon footprint data and the theoretical prediction value respectively and , and these hash values are used as the commitment parameters of the zero-knowledge proof.

[0048] Verify the consistency of commitment parameters through a non-interactive zero-knowledge proof protocol. A non-interactive zero-knowledge proof protocol allows a prover to prove the truth of a statement without revealing any additional information to the verifier. In this system, the prover (such as the carbon footprint verification module) generates proof information through a specific algorithm, and the verifier (such as the management end of the system) verifies whether the actual carbon footprint is consistent with the theoretical prediction value based on this proof information and the commitment parameters. If the verification passes, the output verification result is true; otherwise, it is false.

[0049] Bind the verification result to the blockchain ledger to generate an immutable carbon footprint traceability record. Blockchain technology has the characteristics of decentralization and immutability. Add the verification result as a transaction record to the blockchain ledger, and each transaction contains information such as a timestamp and the verification result. For example, on the Ethereum blockchain, the verification result is written to the blockchain through a smart contract, making the carbon footprint traceability record tamper-proof and ensuring the authenticity and reliability of the data.

[0050] Aggregate multi-source data through a Merkle tree structure. A Merkle tree is a hash binary tree structure that can efficiently verify the integrity of data. Use carbon footprint-related data from different sources (such as actual carbon footprint data, theoretical prediction values, verification results, etc.) as leaf nodes, calculate the hash value of each node, and then calculate the hash value of the parent node layer by layer until the root hash value is obtained. When verifying the integrity of the data, only need to verify whether the root hash value is correct. For example, during data transmission, if the data is tampered with, the calculated root hash value will be different from the original root hash value, thus enabling the timely detection of data anomalies and ensuring data security.

[0051] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0052] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An optimized control system for green electricity consumption of central air conditioners based on carbon footprint traceability, characterized in that, Including: A carbon footprint traceability data acquisition unit, which is used to obtain real-time grid energy carbon emission intensity data, green electricity supply ratio data, and air-conditioning load demand data associated with the central air-conditioning system through a dynamic weighted acquisition algorithm; A green electricity consumption priority evaluation unit, which is used to quantitatively rank the green electricity consumption priorities in multiple time periods based on a dynamic game theory model and generate a green electricity consumption strategy sequence; An energy scheduling optimization unit, which is used to construct a green electricity consumption optimization model by using an adaptive particle swarm algorithm with carbon footprint constraints according to the green electricity consumption strategy sequence and real-time grid energy data; A real-time load matching unit, which is used to dynamically adjust the matching relationship between the green electricity consumption ratio and the air-conditioning load according to the green electricity consumption optimization model and the central air-conditioning operation parameters through a deep reinforcement learning strategy; A carbon footprint verification unit, which is used to verify the consistency between the actual carbon footprint and the theoretical prediction value during the green electricity consumption process based on a zero-knowledge proof algorithm and output a carbon footprint verification result; Among them, the execution steps of the green electricity consumption priority evaluation unit include: Construct a multi-time period green electricity consumption game matrix and define the profit functions of green electricity suppliers, grid operators, and user entities; Through the Nash equilibrium solution algorithm, iteratively optimize the green electricity consumption priority weights in each time period; Generate the green electricity consumption strategy sequence including time period priorities and green electricity allocation ratios according to the game results.

2. The system according to claim 1, wherein The execution steps of the carbon footprint traceability data acquisition unit include: Divide the carbon emission weight factors according to the grid energy types and dynamically collect the real-time power supply ratios of different energy types; Based on the sliding time window algorithm, perform multi-scale smoothing processing on the green electricity supply ratio data to generate a normalized green electricity consumption benchmark value; Integrate the spatio-temporal distribution characteristics of the air-conditioning load demand data and construct a dynamically weighted carbon footprint traceability data set.

3. The system according to claim 2, wherein The execution steps of the carbon footprint traceability data acquisition unit further include: Divide the carbon emission intensity sub-regions according to the grid regions and use distributed data acquisition nodes to synchronously obtain the real-time energy type data of each sub-region; Eliminate the noise of the collected data through the Kalman filter algorithm to generate high-confidence carbon footprint traceability data.

4. The system according to claim 1, wherein The execution steps of the green electricity consumption priority evaluation unit further include: Introduce a time decay factor to dynamically adjust the influence weight of historical green electricity consumption data; Construct a Pareto front solution set for multi-objective games and screen the optimal green electricity consumption strategy sequence through the entropy weight method.

5. The system according to claim 1, wherein The execution steps of the energy scheduling optimization unit include: Establish a multi-objective optimization problem with the minimum carbon footprint as the objective function and the grid stability, green electricity consumption ratio, and air-conditioning load fluctuation as the constraint conditions; Use the adaptive particle swarm algorithm to dynamically adjust the inertia weight and search range of the particle swarm and solve the green electricity consumption optimization model.

6. The system according to claim 5, wherein ​ ​ ​ 7. The system according to claim 1, wherein ​ Build a deep reinforcement learning network with the air conditioner operating parameters, environmental temperature, and green electricity consumption ratio as the state space and the load regulation instruction as the action space; Update the policy network parameters through the Q-learning algorithm to generate real-time green electricity consumption and load matching instructions.

8. The system according to claim 7, wherein The execution steps of the real-time load matching unit further include: Use a long short-term memory network to predict the air conditioner load demand in the future period as the input feature of the deep reinforcement learning strategy; Optimize the exploration and exploitation balance of the action space through the twin-delayed deep deterministic policy gradient algorithm.

9. The system according to claim 1, wherein The execution steps of the carbon footprint verification unit include: Encode the actual carbon footprint data and the theoretical prediction value into hash values to generate commitment parameters for zero-knowledge proof; Verify the consistency of the commitment parameters through a non-interactive zero-knowledge proof protocol and output the verification result.

10. The system according to claim 9, characterized in that The execution steps of the carbon footprint verification unit further include: binding the verification result to the blockchain ledger to generate an immutable carbon footprint traceability record; aggregating multi-source data through a Merkle tree structure.

Images