Multi-dimensional simulation evaluation method for low-carbon life policy effect

Abstract

The invention relates to the field of low-carbon life, and discloses a multi-dimensional simulation evaluation method for a low-carbon life policy effect, and the method comprises the following steps: S1, building a macroscopic intervention conduction model and a microcosmic heterogeneous agent model; s2, based on the microscopic heterogeneous agent model, driving the agent to make a low-carbon behavior decision; s3, polymerizing low-carbon behaviors of the intelligent agent at the current simulation time step, and taking the low-carbon behaviors as input of the macroscopic intervention conduction model; s4, dynamically adjusting macroscopic intervention variables based on the influence factors, and judging a low-carbon behavior decision according to the microscopic heterogeneous agent model; and S5, iteratively executing a simulation period, and performing multi-dimensional evaluation and analysis on a simulation result generated in the simulation process. Through a double-layer simulation model technology, in combination with a macroscopic intervention conduction model and a microcosmic heterogeneous agent model, multi-dimensional evaluation of a low-carbon policy effect is realized, and the defect that the policy effect cannot be comprehensively grasped under a single view angle is overcome.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of low-carbon life, specifically to a multi-dimensional simulation evaluation method for the effect of low-carbon life policy. BACKGROUND

[0002] In recent years, the evaluation of low-carbon life policy has gradually gained attention, especially in the context of climate change and sustainable development, relevant research continues to deepen. The existing technology mainly adopts the method based on statistical model, such as linear regression and static data analysis, to evaluate the impact of macro-level policy. These technologies emphasize the analysis of the effect of policy implementation through historical data and conditional probability, aiming to provide certain basis for decision-makers. In addition, the introduction of agent model provides a new perspective for evaluation, by simulating the behavior and decision of individual, enhancing the understanding of individual difference and behavior pattern.

[0003] However, the existing technology often faces some challenges in dealing with complex low-carbon policy evaluation. Traditional statistical methods are difficult to deal with dynamic changes and nonlinear relationships, resulting in limited accuracy of evaluation results. At the same time, macro analysis usually ignores the heterogeneity of individual behavior, failing to fully reflect the real impact of policy implementation. In addition, the agent decision relying on classical algorithm will fall into local optimum in complex environment, reducing the depth and practicality of policy evaluation. On the other hand, it also lacks flexible feedback mechanism, making the policy response not timely enough, limiting the adaptability of decision-making. SUMMARY

[0004] In view of the shortcomings of the prior art, the present application provides a multi-dimensional simulation evaluation method for the effect of low-carbon life policy, which solves the problems of dynamic adaptability, individual difference and feedback mechanism in traditional evaluation technology.

[0005] To achieve the above purpose, the present application is realized by the following technical scheme: a multi-dimensional simulation evaluation method for the effect of low-carbon life policy, comprising the following steps: S1, constructing a double-layer simulation model: based on dynamic Bayesian network, establishing a macro intervention transmission model, and constructing a micro heterogeneous agent model with situation variables and social network structure for each agent to represent its heterogeneity; S2, low-carbon behavior decision of agent: based on the micro heterogeneous agent model, driving the low-carbon behavior decision of agent in the micro heterogeneous agent model through a quantum behavior heuristic optimization algorithm; S3, micro behavior aggregation and macro input: aggregating the low-carbon behavior of the agent at the current simulation time step, and taking it as the input of the macro intervention transmission model S4, Macro Reasoning and Feedback: The macro intervention transmission model infers macro behavior impact factors and dynamically adjusts macro intervention variables based on the input, and feeds back the macro behavior impact factors and dynamically adjusted macro intervention variables to the micro heterogeneous agent model to influence the low-carbon behavior decision-making of the agent at the next time step; S5, Iterative Simulation and Evaluation: At least one simulation cycle is iteratively executed from S2 to S4, and the simulation results generated during the simulation process are evaluated and analyzed in multiple dimensions.

[0006] Preferably, in step S1, the macro intervention transmission model based on dynamic Bayesian network comprises: S111, Define model structure: determine the nodes and edges of the model, the nodes represent related macro variables, and the edges represent the causal relationship between the nodes; S112, Determine node state: define the possible state of each node, which can be discrete and continuous; S113, Collect and organize data: collect historical data and expert knowledge to obtain the conditional probability distribution between nodes; S114, Build conditional probability table: build a conditional probability table for each node to describe its state dependence on the state of the parent node; S115, Parameter learning: use machine learning to learn the model parameters, fit and adjust the conditional probability table using the collected data; S116, Model verification: verify the accuracy and reliability of the model through cross-validation; S117, Establishment of dynamic reasoning function: realize the reasoning function of dynamic Bayesian network, which can update the model state in real time based on new input information, and infer the subsequent development trend and influence of macro variables; S118, Output result analysis: analyze the impact of macro intervention on low-carbon life according to the inference results output by the model, and form decision support data; The macro variables include policy intervention, economic indicators and environmental impact; The historical data includes statistical bureau, research report, literature and expert interview.

[0007] Preferably, in step S1, the micro heterogeneous agent model comprising a situation variable and a social network structure for characterizing the heterogeneity of each agent comprises: S121, Determine agent characteristics: identify the characteristics of each agent, including demographic variables, individual behavior preferences and habits, as well as cognitive level and emotional attitude related to low-carbon behavior; S122, Set situation variable: assign appropriate situation variable to each agent, so that the model can consider the background differences of different individuals; S123, Establishing social network structure: designing the social network of each agent, determining the interaction relationships between agents, including the topology of the network and its connection strength; S124, Defining agent behavior rules: formulating the behavior decision rules of each agent under different situations, describing the low-carbon behavior choices made by the agent based on its situational variables and social network influence; S125, Simulation initialization: initializing the state of each agent, including the value of its situational variables and the connection relationship in the social network, to establish the initial micro-environment; S126, Rich heterogeneity: introducing the heterogeneity between agents through clustering analysis method, ensuring the diversity of agents in behavior and decision-making in the model; S127, Model verification and testing: verifying the constructed micro-model, ensuring that the behavior of the agent is consistent with the observed social behavior, and testing the performance through simulation experiment; S128, Dynamic updating mechanism: setting each agent to dynamically update its situational variables and behavior decisions according to environmental changes, to support the continuous iteration of the model; The demographic variables include age, gender and education level; The situational variables include local economic conditions, policy environment, social culture and neighborhood relationships.

[0008] Preferably, the S2 step includes: S21, Obtaining the state information of the agent: collecting the current state information of each agent, including the value of the situational variables, the historical behavior record and the interaction in the social network; S22, Calculating individual historical optimal behavior: for each agent, calculating the individual's best low-carbon behavior choice in the past simulation based on its historical data, forming the individual historical optimal behavior record; S23, Calculating the global optimal behavior of the group: in all agents, the overall effect of low-carbon behavior is calculated, and the global optimal behavior of the group under the current simulation environment is calculated, that is, the best low-carbon behavior in all agents; S24, Assessing the influence of social neighbors: identifying and assessing the low-carbon behavior of each neighbor agent in the current agent's social network and its influence on the agent's decision, including the influence size and behavior similarity; S25, Generating quantum state probability distribution: based on the individual historical optimal behavior, the global optimal behavior of the group and the influence of the social neighbors, generating a quantum state probability distribution for each agent, describing the probability of its low-carbon behavior choice; S26, calling quantum behavior heuristic optimization algorithm: applying quantum behavior heuristic optimization algorithm, on the basis of the above probability distribution, simulating the decision of low-carbon behavior of agent, selecting the adaptive behavior path; S27, updating the state of agent: updating the state information of agent according to the selected low-carbon behavior, and recording the decision and execution result, providing feedback information for subsequent iteration.

[0009] Preferably, the S3 step comprises: S31, collecting low-carbon behavior data of current simulation time step: collecting the low-carbon behavior of all agents in the current simulation time step, forming a low-carbon behavior database; S32, calculating the sum of low-carbon behavior: summarizing the collected low-carbon behavior data, calculating the sum of low-carbon behavior of all agents in the current simulation time step, for subsequent analysis; S33, extracting key behavior features: extracting key features from low-carbon behavior data, including behavior type, execution frequency and impact benefit, helping to analyze and understand behavior patterns; S34, updating macro intervention variables: updating macro intervention variables according to the sum and feature information of low-carbon behavior in the current time step, in order to provide the latest input data for macro model; S35, inputting macro intervention conduction model: inputting the aggregated low-carbon behavior and its related features into the macro intervention conduction model for subsequent reasoning and evaluation.

[0010] Preferably, in step S4, the dynamic adjustment of macro intervention variables comprises: S411, analyzing macro behavior influence factors: analyzing the current macro behavior influence factors according to the input low-carbon behavior data and features.

[0011] S412, evaluating the effectiveness of current intervention variables: evaluating the influence of current macro intervention variables on low-carbon behavior, including its feedback effect in different situations, to judge whether it is effective.

[0012] S413, dynamically adjusting intervention variables: dynamically adjusting macro intervention variables according to the analysis results of macro behavior influence factors and the effectiveness evaluation of current intervention variables, to optimize the guiding effect of low-carbon behavior.

[0013] S414, recording the adjusted intervention variables: recording the adjusted macro intervention variables, providing the latest input information for subsequent modeling and simulation; The influence factors include policy effect, environmental impact index and social and economic conditions.

[0014] Preferably, in step S4, the influencing the low-carbon behavior decision of the agent at the next time step comprises: S421, aggregate macro feedback information: organize and summarize the updated macro behavior influence factors and macro intervention variables to provide feedback information for each agent.

[0015] S422, update the decision rule of the agent: based on the aggregated macro feedback information, update the behavior decision rule of each agent to adapt to the new macro environment.

[0016] S423, influence the low-carbon behavior decision of the agent: make the agent make low-carbon behavior choices in the new decision environment, re-evaluate and select the optimal low-carbon behavior according to the guidance of the adjusted macro intervention variables and influence factors.

[0017] Preferably, in step S5, the simulation results generated by the simulation process are multi-dimensionally evaluated and analyzed, including: S51, result data organization: organize the low-carbon behavior data, macro behavior influence factors and macro intervention variable results generated in the simulation process to form an analyzable data set; S52, quantitative analysis: quantitative analysis of the organized result data, including calculation of the overall benefit of low-carbon behavior, implementation effectiveness index and trend analysis, evaluation of the effect of different intervention strategies; S53, qualitative analysis: collect the subjective feelings and feedback of participating agents on low-carbon behavior through interviews and questionnaires, and conduct qualitative analysis on the simulation results to understand the potential factors affecting agent decision-making; S54, comparative analysis: compare the current simulation results with the previous baseline data to analyze the performance differences under different situations and obtain the influence of different factors on low-carbon behavior; S55, model verification and adjustment: according to the evaluation results, test the accuracy and adaptability of the model, and adjust the model parameters and structure if necessary; S56, result visualization: use data visualization tools to generate charts and reports to display the analysis results in a graphical way; S57, form evaluation report: summarize all analysis results to form a comprehensive evaluation report, and put forward policy suggestions and improvement measures to provide reference for future decision-making.

[0018] Preferably, the quantum behavior heuristic optimization algorithm in step S26 includes the expression, measurement and updating process of quantum state, and the specific formula includes: Quantum state initialization: ; In the formula, is the quantum state, is the number of possible behavior choices, is the ground state of each low-carbon behavior, is the probability amplitude associated with the behavior, and satisfies the normalization condition: ; Quantum behavior selection probability: ; where, is the probability of choosing behavior ; represents the square of the modulus of the probability amplitude , which determines the relative likelihood of choosing behavior ; Quantum gate operation: In quantum algorithms, the probability amplitude is adjusted by applying quantum gate operations to the qubits, thereby affecting the final measurement results. The general form is: ; where, is a quantum gate that can change the quantum state, represents the new quantum state after the quantum gate operation ; Measurement operation: In order to obtain the final behavior selection, the quantum state needs to be measured: ; where, represents the actual behavior selection result after measurement, which will be obtained from the quantum state, represents the measurement operation used to obtain the actual selection result from the quantum state; Iterative update: According to the measurement results and feedback information, update the quantum state for the next round of optimization: ; where, represents the updated quantum state, ready for the next round of decision-making, represents a function that generates a new quantum state based on the current measurement results and the old quantum state , represents the old quantum state, i.e. the quantum state obtained in the last iteration.

[0019] Preferably, the macroscopic intervention conduction model includes: ; where, represents the macroscopic intervention variable at the current time step tt, represents the macroscopic behavior influence factor at the last time step , reflecting the influence of history on current decision-making, Indicates the current time step The low-carbon behavior data includes the actual behavior results based on the agent, Indicates that external environment and policy intervention factors can affect current macroeconomic decisions. is a function that represents how to dynamically update macro intervention variables based on historical impact factors, current behavioral data, and external factors; The micro-heterogeneous agent model includes: ; Where, Representing an agent At time step low-carbon behavior decisions, Representing an agent At the previous time step The chosen behavior, Indicates the current time step The macro intervention variables come from the macro intervention transmission model, Representing an agent In time Feedback information may include the results of interactions with other agents, Represents a function that shows how the agent makes decisions based on historical behavior, macro interventions, and feedback information.

[0020] The present invention provides a multi-dimensional simulation evaluation method for the effects of low-carbon living policies. It has the following beneficial effects: 1. This invention utilizes a two-layer simulation model, combining a macro-intervention transmission model with a micro-heterogeneous agent model, to achieve a multi-dimensional assessment of the effects of low-carbon policies. This enables a comprehensive analysis of policy impacts and improves the accuracy of assessments. Compared to existing single-perspective assessment methods, this approach overcomes the limitations of traditional approaches and addresses the inability to fully grasp policy effects from a single perspective.

[0021] 2. This invention innovatively promotes low-carbon behavior decision-making by intelligent agents by introducing a quantum-inspired optimization algorithm. This technology makes the decision-making process more efficient and adaptable. Compared with traditional decision-making methods that rely on classical algorithms, this invention significantly improves decision-making speed and accuracy, overcoming the problem that classical algorithms are prone to falling into local optimality in complex decision-making environments.

[0022] 3. This invention ensures the real-time and effectiveness of the feedback mechanism by dynamically adjusting macro-intervention variables. This flexibility enables the model to quickly adapt to changes and optimize the effectiveness of guiding low-carbon behavior. Many existing models lack dynamic adjustment capabilities, resulting in delayed policy implementation. This solution addresses this shortcoming, improving policy responsiveness and practicality. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is a method step framework of the present application; Figure 2 is a macro intervention conduction model structure diagram of the present application; Figure 3 is a micro heterogeneous intelligent agent model structure diagram of the present application; Figure 4 is a quantum behavior heuristic optimization algorithm flowchart of the present application. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the specification of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0025] Embodiment: Please refer to the drawings in the specification of the present application Figure 1 - the drawings in the specification of the present application Figure 4 The embodiment of the present application provides a multi-dimensional simulation evaluation method for the effect of low-carbon life policy, which comprises the following steps: S1, constructing a double-layer simulation model: based on dynamic Bayesian network, a macro intervention conduction model is established, and a micro heterogeneous intelligent agent model is constructed, each intelligent agent has a context variable and a social network structure for representing its heterogeneity; In step S1, the macro intervention conduction model is established based on dynamic Bayesian network, which comprises: S111, defining the model structure: determining the nodes and edges of the model, the nodes represent the related macro variables, and the edges represent the causal relationship between the nodes; S112, determining the node state: defining the possible states for each node, the states can be discrete and continuous; S113, collecting and organizing data: collecting historical data and expert knowledge to obtain the conditional probability distribution between nodes; S114, constructing the conditional probability table: constructing the conditional probability table for each node to describe the state dependence on the state of the parent node; S115, parameter learning: using machine learning to learn the model parameters, fitting the collected data, and adjusting the conditional probability table; S116, model verification: verifying the accuracy and reliability of the model through cross-validation; S117, Establishing dynamic inference mechanism: Implementing the inference mechanism of dynamic Bayesian network, which can update the model state in real time based on new input information, and infer the subsequent development trend of macro variables and its impact; S118, Output result analysis: Analyzing the impact of macro intervention on low-carbon life based on the inference results output by the model, and forming decision support data; Macro variables include policy intervention, economic indicators and environmental impact; Historical data includes statistical bureau, research report, literature and expert interview; In step S1, the micro-heterogeneous agent model is constructed, which has situation variables and social network structure for each agent to represent its heterogeneity, including: S121, Determine the characteristics of the agent: Determine the characteristics of each agent, including demographic variables, individual behavior preferences and habits, as well as cognitive level and emotional attitude related to low-carbon behavior; S122, Set the situation variable: Assign appropriate situation variables to each agent, so that the model can consider the background differences of different individuals; S123, Establish social network structure: Design the social network of each agent, determine the interaction relationship between agents, including the topological structure of the network and its connection strength; S124, Define agent behavior rules: Develop behavior decision rules for each agent in different situations, describe the low-carbon behavior choices made by the agent based on its situation variables and social network influence; S125, Simulation initialization: Initialize the state of each agent, including the value of its situation variable and the connection relationship in the social network, to establish the initial micro-environment; S126, Rich heterogeneity: Introduce the heterogeneity between agents through cluster analysis method, ensure the diversity of agents in behavior and decision-making in the model; S127, Model verification and testing: Verify the micro-model constructed, ensure that the behavior of the agent is consistent with the observed social behavior, and test the performance through simulation experiment; S128, Dynamic update mechanism: Set each agent to dynamically update its situation variables and behavior decisions according to environmental changes, to support continuous iteration of the model; Demographic variables include age, gender and education level; Situation variables include local economic conditions, policy environment, social culture and neighborhood relationships.

[0026] Specifically, in step S1, a macro-intervention transmission model is first established. This model is based on a dynamic Bayesian network (DBN). By defining the relationships between nodes and edges, nodes represent related macro variables, while edges represent causal relationships between nodes. Macro variables include policy interventions, economic indicators, and environmental impacts. A conditional probability table is constructed to define the state of each node, which can be discrete or continuous.

[0027] Furthermore, defining the model structure includes determining the relevant nodes and the edge A collection of nodes Represents macro variables, edge Representing variables Influencing variables . It is described by the following conditional probability distribution: ; Where, Representation node The parent node of is the number of parent nodes, Indicates the state of a given parent node The conditional probability of describing the node The state of a node depends on the state of its parent node. Represents macro variables , is one of the nodes defined in the two-layer model. This formula shows how the state of a node depends on the state of its parent node.

[0028] Then, by collecting and organizing data, including historical data from statistical bureaus, research reports, and expert interviews, we extracted the conditional probability distribution between nodes to provide a basis for constructing a conditional probability table. The constructed conditional probability table is converted into the following form: ; Where, To describe the node Conditional probability, determined by the state of its parent node, Given the parent node state, the node The value is The conditional probability of Macro variables , is a node in the two-layer model.

[0029] Next, a machine learning algorithm is used to learn parameters to fit the conditional probability table using the collected data. The parameter learning step is achieved by minimizing the following loss function: ; Where, is the loss function used to evaluate the estimation accuracy of the model parameters, P(observed) is the probability of the observed data, P(model) is the probability of the model, N is the number of samples. The result of this step is an optimized conditional probability table, ensuring the accuracy and reliability of the model.

[0030] After completing the construction of the model structure, the model is verified by cross-validation or other performance evaluation methods. In this process, the model indicators are calculated to evaluate the prediction accuracy of the model, and the following verification formula is established: ; In the formula, , , and represent the number of true positives, true negatives, false positives and false negatives, respectively, ensuring the stability and practicality of the model, P(model) is the accuracy of the model, indicating the correctness of the model prediction.

[0031] Finally, the inference function of the dynamic Bayesian network is realized, which can process real-time data input and infer the subsequent development trend of macro variables. The inference function is represented by the following formula: ; In the formula, P(node | data) is the posterior probability of node given the observed data, P(data | node) is the likelihood probability of the observed data given that node takes a specific value, P(node) is the prior probability of node , representing the initial belief of the node state, P(data) is the probability of observing the data on the entire data set, serving as a normalization factor, and is suitable for evaluating the effect of macro behavior influencing factors on low-carbon life.

[0032] Through the above steps, the double-layer simulation model can be completely constructed to evaluate the effect of low-carbon life policy and realize the effectiveness and sustainability of agent decision-making, providing a scientific basis for subsequent optimization of low-carbon policy.

[0033] S2, agent low-carbon behavior decision: based on the micro-heterogeneous agent model, a quantum behavior heuristic optimization algorithm is used to drive the agent low-carbon behavior decision within the micro-heterogeneous agent model; The S2 step includes: S21, obtaining the state information of the agent: collecting the current state information of each agent, including the value of the situational variable, the historical behavior record and the interaction in the social network; S22. Calculate individual historical optimal behavior: For each agent, calculate the individual's optimal low-carbon behavior choice in past simulations based on its historical data, and form a record of the individual's historical optimal behavior; S23. Calculate the global optimal behavior of the group: Count the overall effects of low-carbon behaviors among all intelligent agents and calculate the global optimal behavior of the group in the current simulation environment, that is, the best low-carbon behavior among all intelligent agents; S24. Evaluate social neighbor influence: Identify and evaluate the low-carbon behavior of each neighboring agent in the current agent's social network and its impact on the agent's decision-making, including the size of influence and behavioral similarity; S25. Generate quantum state probability distribution: Based on the individual's historical optimal behavior, the group's global optimal behavior, and the influence of social neighbors, generate a quantum state probability distribution for each intelligent agent to describe the probability of it choosing low-carbon behavior; S26. Calling the quantum behavior heuristic optimization algorithm: Applying the quantum behavior heuristic optimization algorithm, based on the above probability distribution, simulates the decision-making of the intelligent agent's low-carbon behavior and selects an adaptive behavior path; S27. Update agent status: Update the agent's status information based on the selected low-carbon behavior, and record its decision and execution results to provide feedback information for subsequent iterations; The quantum behavior heuristic optimization algorithm in step S26 includes the expression, measurement and update process of quantum state. The specific formulas include: Quantum state initialization: ; Where, is a quantum state, is the number of possible behavioral choices, is the ground state of each low-carbon behavior, is the probability magnitude associated with the behavior and satisfies the normalization condition: ; Quantum behavior selection probability: ; Where, It is a choice behavior The probability of Indicates probability amplitude The square of the modulus determines the selection behavior the relative likelihood of Quantum gate operation: In quantum algorithms, quantum gate operations are applied to qubits to adjust the probability amplitude, thereby affecting the final measurement result. The general form is: ; Where, is a quantum gate that can change the quantum state, represents the new quantum state after the quantum gate operation; Measurement operation: In order to obtain the final action choice, the quantum state needs to be measured: where, represents the actual action selection result after measurement, which will be obtained from the quantum state, represents the measurement operation for obtaining the actual selection result from the quantum state; Iterative update: According to the measurement result and feedback information, update the quantum state for the next round of optimization: where, represents the updated quantum state, ready for the next round of decision-making, represents a function that generates a new quantum state based on the current measurement result and the old quantum state , represents the old quantum state, i.e. the quantum state obtained in the last iteration; Specifically, in the S2 step, first, obtain the state information of each agent, collect the current agent's context variable value, historical behavior record, and interaction in the social network. Define the state information of each agent as follows: where, represents the context variable value of the agent at time step , represents the historical behavior record of the agent , represents the interaction in the social network.

[0034] Subsequently, calculate the individual historical optimal behavior, for each agent based on historical data analysis, obtain its best low-carbon behavior selection , which can be represented by the following formula: where, represents the historical optimal low-carbon behavior selection of the agent at time step , the set of all low-carbon behaviors that the agent can choose, ​​​​​​The expected utility function is used to evaluate the behavioral effects in different situations. The best behavior in the history record is preferred to form the individual historical optimal behavior record.

[0035] Next, the social neighbor influence is evaluated, and the low-carbon behaviors of each parameter in the social network and their influence on the agent are identified and evaluated. Specifically, the influence of the neighbor agent is calculated The decision-making influence of the agent is quantified as follows: ; In the formula, represents the low-carbon behavior of the agent at time step , and is the macro intervention variable. Through the aggregation of influence, the decision basis of an agent is formed.

[0036] Generate a quantum state probability distribution, based on the individual historical optimal behavior, the global optimal behavior of the group, and the influence of the social neighbor, to generate a quantum state probability distribution for each agent in the following form: ; In the formula, represents the probability of the agent choosing a low-carbon behavior , and represents the influence of the agent on the agent .

[0037] This formula describes the probability of the agent choosing a low-carbon behavior and outputs the corresponding probability value.

[0038] Call the quantum behavior heuristic optimization algorithm to initialize and optimize the behavior selection of the quantum state. Specifically, it includes: Quantum state initialization: ; In the formula, is the quantum state, is the number of possible behavior choices, is the ground state of each low-carbon behavior, is the probability amplitude related to the behavior, and satisfies the normalization condition: ; Quantum behavior selection probability: ; In the formula, is the probability of choosing behavior , represents the square of the modulus of the probability amplitude , which determines the relative likelihood of choosing behavior . Quantum gate operation: In quantum algorithms, the probability amplitude is adjusted by applying quantum gate operations to qubits, thereby affecting the final measurement results. The general form is: ; where, is a quantum gate that can change the quantum state, represents the new quantum state after the quantum gate operation ; Measurement operation: In order to obtain the final behavior selection, the quantum state needs to be measured: ; where, represents the actual behavior selection result after measurement, which will be obtained from the quantum state, represents the measurement operation used to obtain the actual selection result from the quantum state; Iterative update: According to the measurement results and feedback information, update the quantum state for the next round of optimization: ; where, represents the updated quantum state, ready for the next round of decision-making, represents a function that generates a new quantum state based on the current measurement results and the old quantum state , represents the old quantum state, i.e. the quantum state obtained in the last iteration. Through the implementation of the above steps, the integrity of the technical features and the feasibility of the implementation are guaranteed, and the effectiveness of the low-carbon behavior decision-making is promoted.

[0039] S3, Micro-behavior aggregation and macro input: Aggregate the low-carbon behaviors of agents at the current simulation time step and use them as inputs for the macro intervention transmission model; S3 includes: S31, Collect low-carbon behavior data at the current simulation time step: Collect the low-carbon behaviors of all agents at the current simulation time step to form a low-carbon behavior database; S32, Calculate the sum of low-carbon behaviors: Summarize the collected low-carbon behavior data and calculate the sum of low-carbon behaviors of all agents at the current simulation time step for subsequent analysis; S33, Extract key behavior features: Extract key features from low-carbon behavior data, including behavior type, execution frequency and impact benefit, to help analyze and understand behavior patterns; S34, Update macro intervention variables: Update macro intervention variables based on the sum and feature information of low-carbon behaviors at the current time step to provide the latest input data for the macro model; S35. Input to the macro-intervention transmission model: The aggregated low-carbon behaviors and their related characteristics are used as inputs and passed to the macro-intervention transmission model for subsequent reasoning and evaluation; Specifically, in step S3, we first establish a low-carbon behavior data collection module. This module is responsible for collecting the low-carbon behavior data of each agent at the current simulation time step. This data will be aggregated into a low-carbon behavior database, which contains the low-carbon behaviors taken by all agents at the current time step.

[0040] Then, the sum of low-carbon behaviors at the current time step is calculated. The formula is as follows: ; Where, Indicates the current time step The sum of the low-carbon behaviors of all intelligent agents in is the total number of agents, Representing an agent At time step low-carbon behavior decisions.

[0041] Next, we extract key behavioral features from the low-carbon behavior data to help analyze and understand the behavior patterns. Key features can include behavior type, execution frequency, and environmental and economic impacts. This step will generate a feature set. , to provide necessary information for the subsequent update of macro intervention variables.

[0042] According to the low-carbon behavior characteristics, summarize the current behavior feature vector , and update the corresponding macro intervention variables. The macro intervention variable update formula is as follows: ; Where, Indicates the current time step The macro-intervention variables, The function that defines the update relationship can dynamically update the macro intervention variables by combining the low-carbon behavior data in the current time step and the extracted feature information.

[0043] Finally, the updated low-carbon behavior sum and characteristic information are input into the macro-intervention transmission model. The input is expressed as: ; Where, Indicates the current time step the input data set, including the low-carbon behavior total and the key feature set. The above steps can dynamically update the macro intervention variable by effectively aggregating the low-carbon behavior data of the agents, calculating the low-carbon behavior total, and extracting the key features, thereby affecting subsequent policy making and intervention measures. The data flow and model interaction in this process clearly constitute the basis of the multi-dimensional simulation evaluation model of the low-carbon life policy effect.

[0044] In summary, the specific way of implementing step S3 realizes the effective conversion of micro-agent behavior data to macro intervention strategy, optimizing the overall decision-making process.

[0045] S4, Macro reasoning and feedback: The macro intervention transmission model infers the macro behavior impact factor based on the input and dynamically adjusts the macro intervention variable, and feeds back the macro behavior impact factor and the dynamically adjusted macro intervention variable to the micro heterogeneous agent model to affect the low-carbon behavior decision of the agent at the next time step; In step S4, dynamically adjusting the macro intervention variable includes: S411, Analyzing the macro behavior impact factor: Analyzing the current macro behavior impact factor according to the input low-carbon behavior data and features.

[0046] S412, Evaluating the effectiveness of the current intervention variable: Evaluating the impact of the current macro intervention variable on low-carbon behavior, including its feedback effect in different situations, to determine whether it is effective.

[0047] S413, Dynamically adjusting the intervention variable: Dynamically adjusting the macro intervention variable according to the analysis results of the macro behavior impact factor and the effectiveness evaluation of the current intervention variable to optimize the guiding effect of low-carbon behavior.

[0048] S414, Recording the adjusted intervention variable: Recording the adjusted macro intervention variable value to provide the latest input information for subsequent modeling and simulation; The impact factors include policy effect, environmental impact indicators, and socio-economic conditions; In step S4, affecting the low-carbon behavior decision of the agent at the next time step includes: S421, Summarizing the macro feedback information: Summarizing the updated macro behavior impact factor and macro intervention variable to provide feedback information for each agent.

[0049] S422, Updating the decision rules of the agent: Based on the summarized macro feedback information, updating the behavior decision rules of each agent to adapt to the new macro environment.

[0050] S423. Influencing the low-carbon behavior decision of the intelligent agent: enabling the intelligent agent to make low-carbon behavior choices in the new decision-making environment, and re-evaluating and selecting the optimal low-carbon behavior based on the guidance of the adjusted macro-intervention variables and influencing factors; The macro-intervention transmission model includes: ; Where, represents the macro intervention variable at the current time step tt, Represents the previous time step The macro-behavioral influence factor reflects the impact of history on current decision-making. Indicates the current time step The low-carbon behavior data includes the actual behavior results based on the agent, Indicates that external environment and policy intervention factors can affect current macroeconomic decisions. is a function that represents how to dynamically update macro intervention variables based on historical impact factors, current behavioral data, and external factors; Micro-heterogeneous agent models include: ; Where, Representing an agent At time step low-carbon behavior decisions, Representing an agent At the previous time step The chosen behavior, Indicates the current time step The macro intervention variables come from the macro intervention transmission model, Representing an agent In time Feedback information may include the results of interactions with other agents, represents a function that shows how the agent makes decisions based on historical behavior, macro interventions, and feedback information; Specifically, in step S4, a multi-dimensional simulation evaluation method for the effectiveness of low-carbon lifestyle policies was constructed, focusing particularly on the macro-reasoning and feedback process in step S4. First, a macro-intervention transmission model was established to infer the current macro-behavioral influencing factors. Macro-intervention variables were adjusted through a dynamic update mechanism to optimize the effectiveness of guiding low-carbon behavior.

[0051] The macro-intervention transmission model enables the model to infer the subsequent development trend of macro variables based on the input low-carbon behavior data. Specifically, the model is expressed as: ; Where, represents the macro intervention variable at the current time step tt, Represents the previous time step The macro-behavioral influence factor reflects the impact of history on current decision-making. Indicates the current time step The low-carbon behavior data includes the actual behavior results based on the agent, Indicates that external environment and policy intervention factors can affect current macroeconomic decisions. is a function that represents how to dynamically update macro intervention variables based on historical impact factors, current behavioral data, and external factors.

[0052] Subsequently, the current low-carbon behavior data and its characteristics are analyzed to evaluate the effectiveness of macro-behavior influencing factors. The following formula is used to analyze and calculate the macro-behavior influencing factors: ; Where, Indicates the current time step Macro-behavioral influencing factors; is the current time step low-carbon behavior data; It is a characteristic indicator of low-carbon behavior, including guidance effect, environmental response, etc. Represents the inference function of macro-behavior influencing factors.

[0053] Furthermore, the updated macro-intervention variables are recorded for application in subsequent simulation cycles. The dynamic adjustment process is the key to ensuring the adaptability and real-time performance of the policy.

[0054] Finally, the optimized macro-intervention variables are used to influence the low-carbon behavior decisions of the intelligent agents, thereby improving the overall policy effect. The specific process includes: first, summarizing the updated macro-feedback information, then updating the decision-making rules of the intelligent agents, and finally influencing the low-carbon behavior choices of the intelligent agents in the new environment. This process can be expressed as: ; Where, Representing an agent At time step low-carbon behavior decisions, Representing an agent At the previous time step The chosen behavior, Indicates the current time step The macro intervention variables come from the macro intervention transmission model, Representing an agent In time Feedback information may include the results of interactions with other agents, represents a certain function, which represents how the agent makes decisions based on historical behavior, macro intervention, and feedback information.

[0055] Through the above steps, the constructed macro intervention transmission model and the micro heterogeneous agent model form a complete technical chain. Each part of the model provides support for the decision-making of low-carbon behavior, realizes a multi-dimensional dynamic interaction feedback mechanism, and thus helps the effective implementation of low-carbon policy.

[0056] S5, Iterative simulation and evaluation: iteratively execute at least one of S2 to S4; In step S5, the simulation results generated by the simulation process are evaluated and analyzed in multiple dimensions, including: S51, Result data organization: organize the low-carbon behavior data, macro behavior influence factors and macro intervention variable results generated during the simulation process to form an analyzable data set; S52, Quantitative analysis: quantitative analysis of the result data after organization, including calculation of the overall benefit of low-carbon behavior, effectiveness index of implementation and change trend analysis, evaluation of the effect of different intervention strategies; S53, Qualitative analysis: collect the subjective feelings and feedback of participating agents on low-carbon behavior through interviews and questionnaire surveys, and conduct qualitative analysis on the simulation results to understand the potential factors affecting agent decision-making; S54, Comparative analysis: compare the current simulation results with the previous baseline data, analyze the performance differences under different situations, and obtain the influence of different factors on low-carbon behavior; S55, Model verification and adjustment: according to the evaluation results, test the accuracy and adaptability of the model, and adjust the model parameters and structure if necessary; S56, Result visualization: use data visualization tools to generate charts and reports to display the analysis results in a graphical way; S57, Forming evaluation report: summarize all analysis results, form comprehensive evaluation report, put forward policy suggestions and improvement measures, provide reference for future decision-making; Specifically, in step S5, first, a micro heterogeneous agent model is established, which includes multiple heterogeneous agents, each agent having specific situational variables and social network structure. The low-carbon behavior decision of these agents is based on the quantum behavior heuristic optimization algorithm, and the macro intervention transmission model constructed by the dynamic Bayesian network is input and feedback to realize the simulation and evaluation of low-carbon behavior.

[0057] Then, the low-carbon behavior decision of each agent is driven by the quantum behavior heuristic optimization algorithm to generate the initialization of quantum state.

[0058] In the process of obtaining the agent state information, the context variables, historical behavior records of each agent and the interaction of each neighbor in the social network are collected. For each agent, the optimal low-carbon behavior selection of the individual history is calculated to form the individual history optimal behavior record.

[0059] Further, the overall effect of low-carbon behavior for the whole group is calculated, and the global optimal behavior of the group under the current environment is calculated. The low-carbon behavior of social neighbors and its influence on agent decision-making are evaluated, including the influence size and behavior similarity.

[0060] Then, the quantum state is processed through quantum gate operation to increase the flexibility of behavior selection: Then, the final behavior selection result is obtained by using the measurement operation, and the quantum state is updated according to the measurement result and the feedback information: In this process, the state of the agent is updated, and its decision and execution result are recorded. At the same time, the aggregated low-carbon behavior data and related features are transmitted to the macro intervention conduction model as input information, and then dynamic reasoning and feedback are realized, and the state of the intervention variable at each time step is recorded.

[0061] Next, the macro intervention conduction model infers the macro behavior influence factor based on the input and dynamically adjusts the macro intervention variable.

[0062] Finally, the simulation results are analyzed through quantity and quality evaluation to form a report and provide decision support. The implementation effect makes the dynamic feedback of low-carbon behavior possible, forming a continuous optimization process.

[0063] In the process of adjusting the macro intervention variable, the behavior decision rule of the agent is corrected through the feedback information to adapt to the new macro environment.

[0064] Although embodiments of the present application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multi-dimensional simulation evaluation method for the effect of low-carbon living policies, characterized by: The following steps are involved: S1. Construct a two-layer simulation model: Build a macro-intervention transmission model based on a dynamic Bayesian network, and simultaneously construct a micro-heterogeneous agent model in which each agent has contextual variables and social network structures to characterize its heterogeneity; S2. Low-carbon behavior decision-making of intelligent agents: Based on the micro-heterogeneous intelligent agent model, a quantum behavior heuristic optimization algorithm is used to drive the low-carbon behavior decision-making of intelligent agents within the micro-heterogeneous intelligent agent model; S3. Micro-behavior aggregation and macro-input: Aggregate the low-carbon behavior of the agent in the current simulation time step and use it as the input of the macro-intervention transmission model; S4. Macro-level reasoning and feedback: The macro-level intervention transmission model infers macro-level behavioral influencing factors based on the input and dynamically adjusts macro-level intervention variables. The macro-level behavioral influencing factors and the dynamically adjusted macro-level intervention variables are fed back to the micro-level heterogeneous agent model to influence the agent's low-carbon behavior decision at the next time step. S5. Iterative simulation and evaluation: Iteratively execute S2 to S4 for at least one simulation cycle, and perform multi-dimensional evaluation and analysis on the simulation results generated by the simulation process.

2. A multi-dimensional simulation evaluation method for the effect of a low-carbon living policy according to claim 1, characterized in that: In step S1, establishing a macro-intervention transmission model based on a dynamic Bayesian network includes: S111. Define the model structure: determine the nodes and edges of the model. Nodes represent related macro variables, and edges represent causal relationships between nodes. S112. Determine node state: define possible states for each node. The states can be discrete or continuous. S113. Collect and organize data: Collect historical data and expert knowledge to obtain the conditional probability distribution between nodes; S114, constructing a conditional probability table: constructing a conditional probability table for each node, describing how its state depends on the state of its parent node; S115, parameter learning: Use machine learning to learn model parameters, use collected data for fitting, and adjust the conditional probability table; S116. Model validation: Verify the accuracy and reliability of the model through cross-validation; S117. Establishment of dynamic inference functionality: Implementing the inference functionality of a dynamic Bayesian network, capable of updating the model state in real time based on new input information, and inferring the subsequent development trends and impacts of macro variables; S118. Output result analysis: Based on the inference results output by the model, analyze the impact of macro-intervention on low-carbon living and form decision-making support data; The macro variables include policy interventions, economic indicators and environmental impacts; The historical data include statistical bureaus, research reports, literature and expert interviews.

3. The multi-dimensional simulation evaluation method for the effect of a low-carbon life policy according to claim 1 is characterized in that: In step S1, constructing a micro-heterogeneous agent model in which each agent has contextual variables and social network structures for characterizing its heterogeneity includes: S121. Determine Agent Characteristics: Identify the characteristics of each agent, including demographic variables, individual behavioral preferences and habits, and cognitive levels and emotional attitudes related to low-carbon behavior; S122. Set context variables: Assign appropriate context variables to each agent so that the model can comprehensively consider the background differences of different individuals; S123. Establish social network structure: Design the social network of each agent and determine the interactive relationships between agents, including the network topology and connection strength; S124. Define agent behavior rules: Develop behavioral decision rules for each agent in different situations, describing the low-carbon behavior choices made by the agent based on its situational variables and social network influences; S125, Simulation Initialization: Initialize the state of each agent, including the values ​​of its contextual variables and the connections in the social network, to establish the initial micro-environment; S126. Enriching heterogeneity: By using cluster analysis methods, we introduce heterogeneity between agents to ensure the diversity of behaviors and decisions of agents in the model. S127, Model Verification and Testing: Verify the constructed micro-model to ensure that the behavior of the intelligent agent is consistent with the actual observed social behavior, and conduct performance testing through simulation experiments; S128, Dynamic Update Mechanism: Each agent is required to dynamically update its context variables and behavioral decisions based on environmental changes to support continuous iteration of the model; The demographic variables included age, gender, and education level; The contextual variables include local economic conditions, policy environment, social culture and neighborhood relations.

4. The multi-dimensional simulation evaluation method for the effect of a low-carbon living policy according to claim 1 is characterized in that: The S2 step includes: S21. Obtaining agent status information: Collecting the current status information of each agent, including the value of context variables, historical behavior records, and interactions in the social network; S22. Calculate individual historical optimal behavior: For each agent, calculate the individual's optimal low-carbon behavior choice in past simulations based on its historical data, and form a record of the individual's historical optimal behavior; S23. Calculate the global optimal behavior of the group: Count the overall effects of low-carbon behaviors among all intelligent agents and calculate the global optimal behavior of the group in the current simulation environment, that is, the best low-carbon behavior among all intelligent agents; S24. Evaluate social neighbor influence: Identify and evaluate the low-carbon behavior of each neighboring agent in the current agent's social network and its impact on the agent's decision-making, including the size of influence and behavioral similarity; S25. Generate quantum state probability distribution: Based on the individual's historical optimal behavior, the group's global optimal behavior, and the influence of social neighbors, generate a quantum state probability distribution for each intelligent agent to describe the probability of it choosing low-carbon behavior; S26. Calling the quantum behavior heuristic optimization algorithm: Applying the quantum behavior heuristic optimization algorithm, based on the above probability distribution, simulates the decision-making of the intelligent agent's low-carbon behavior and selects an adaptive behavior path; S27. Update the agent status: Update the agent status information based on the selected low-carbon behavior, and record its decision and execution results to provide feedback information for subsequent iterations.

5. The multi-dimensional simulation evaluation method for the effect of a low-carbon life policy according to claim 1, characterized in that: The S3 step includes: S31. Collect low-carbon behavior data of the current simulation time step: collect the low-carbon behaviors taken by all intelligent agents in the current simulation time step to form a low-carbon behavior database; S32. Calculate the sum of low-carbon behaviors: Summarize the collected low-carbon behavior data and calculate the sum of low-carbon behaviors of all agents in the current simulation time step for subsequent analysis; S33. Extract key behavioral features: Extract key features from low-carbon behavior data, including behavior type, execution frequency, and impact benefits, to help analyze and understand behavioral patterns; S34. Update macro intervention variables: Update macro intervention variables based on the sum and characteristic information of low-carbon behaviors in the current time step to provide the latest input data for the macro model; S35. Input the macro-intervention transmission model: The aggregated low-carbon behavior and its related characteristics are passed as input to the macro-intervention transmission model for subsequent reasoning and evaluation.

6. The multi-dimensional simulation evaluation method for the effect of a low-carbon life policy according to claim 1, characterized in that: In step S4, the dynamic adjustment of macro intervention variables includes: S411. Analyze macro-behavior influencing factors: Analyze current macro-behavior influencing factors based on the input low-carbon behavior data and characteristics. S412. Evaluate the effectiveness of current intervention variables: Evaluate the impact of current macro-intervention variables on low-carbon behavior, including their feedback effects in different scenarios, and determine whether they are effective. S413. Dynamically adjust intervention variables: Based on the analysis results of macro-behavioral influencing factors and the effectiveness evaluation of current intervention variables, dynamically adjust macro-intervention variables to optimize the guiding effect of low-carbon behavior. S414. Record adjusted intervention variables: Record the adjusted macro intervention variable values ​​to provide the latest input information for subsequent modeling and simulation; The influencing factors include policy effects, environmental impact indicators and socio-economic conditions.

7. The multi-dimensional simulation evaluation method for the effect of a low-carbon life policy according to claim 1, characterized in that: In step S4, the low-carbon behavior decision affecting the agent in the next time step includes: S421. Summarize macro feedback information: organize and summarize the updated macro behavior influencing factors and macro intervention variables to provide feedback information for each intelligent agent. S422. Update the decision rules of the intelligent agent: Based on the aggregated macro-feedback information, update the behavioral decision rules of each intelligent agent to adapt it to the new macro-environment. S423. Influencing the low-carbon behavior decision of the intelligent agent: enabling the intelligent agent to make low-carbon behavior choices in a new decision-making environment, and re-evaluate and select the optimal low-carbon behavior based on the guidance of the adjusted macro-intervention variables and influencing factors.

8. The multi-dimensional simulation evaluation method for the effect of a low-carbon life policy according to claim 1 is characterized in that: In step S5, the simulation results generated by the simulation process are evaluated and analyzed in multiple dimensions, including: S51. Result data collation: collate the low-carbon behavior data, macro-behavior influencing factors, and macro-intervention variable results generated during the simulation process to form an analyzable data set; S52. Quantitative analysis: Conduct quantitative analysis of the collated data, including calculating the overall benefits of low-carbon behavior, effectiveness indicators of implementation, and trend analysis, and evaluating the effects of different intervention strategies. S53. Qualitative analysis: Collect the subjective feelings and feedback of participating agents on low-carbon behavior through interviews and questionnaires, conduct qualitative analysis of simulation results, and understand the potential factors that affect the decision-making of agents; S54. Comparative analysis: Compare the current simulation results with the previous baseline data to analyze the performance differences under different scenarios and obtain the impact of different factors on low-carbon behavior; S55. Model verification and adjustment: Based on the evaluation results, test the accuracy and adaptability of the model and adjust the model parameters and structure if necessary; S56. Result visualization: Use data visualization tools to generate charts and reports to present the analysis results in a graphical manner; S57. Form an evaluation report: Summarize all analysis results to form a comprehensive evaluation report, put forward policy recommendations and improvement measures, and provide reference for future decision-making.

9. The multi-dimensional simulation evaluation method for the effect of a low-carbon life policy according to claim 1, characterized in that: The quantum behavior heuristic optimization algorithm described in step S26 includes the expression, measurement and update process of the quantum state, and the specific formula includes: Quantum state initialization: ； Where, is a quantum state, is the number of possible behavioral choices, is the ground state of each low-carbon behavior, is the probability magnitude associated with the behavior and satisfies the normalization condition: ; Quantum behavior selection probability: ； Where, It is a choice behavior The probability of Indicates probability amplitude The square of the modulus determines the selection behavior the relative likelihood of Quantum gate operation: In quantum algorithms, quantum gate operations are applied to qubits to adjust the probability amplitude, thereby affecting the final measurement result. The general form is: ； Where, is a quantum gate that can change the quantum state, Indicates that after quantum gate operation The new quantum state after Measurement operation: In order to obtain the final behavior, the quantum state needs to be measured: ； Where, Indicates that the actual behavior choice result after measurement will be derived from the quantum state, Represents the measurement operation, which is used to obtain the actual selection result from the quantum state; Iterative update: Based on the measurement results and feedback information, the quantum state is updated for the next round of optimization: ； Where, Indicates the updated quantum state, ready for the next round of decision-making, Represents a function that is based on the current measurement results and the old quantum state To generate new quantum states, Represents the old quantum state, that is, the quantum state obtained in the previous iteration.

10. The multi-dimensional simulation evaluation method for the effect of a low-carbon life policy according to claim 1, characterized in that: The macro-intervention transmission model includes: ； Where, represents the macro intervention variable at the current time step tt, Represents the previous time step The macro-behavioral influence factor reflects the impact of history on current decision-making. Indicates the current time step The low-carbon behavior data includes the actual behavior results based on the agent, Indicates that external environment and policy intervention factors can affect current macroeconomic decisions. is a function that represents how to dynamically update macro intervention variables based on historical impact factors, current behavioral data, and external factors; The micro-heterogeneous agent model includes: ； Where, Representing an agent At time step low-carbon behavior decisions, Representing an agent At the previous time step The chosen behavior, Indicates the current time step The macro intervention variables come from the macro intervention transmission model, Representing an agent In time Feedback information may include the results of interactions with other agents, Represents a function that shows how the agent makes decisions based on historical behavior, macro interventions, and feedback information.

Images