Regional carbon emission permit trading system modeling based on multi-agent simulation

Abstract

The invention provides regional carbon emission permit trading system modeling based on multi-agent simulation, and a multi-agent modeling and simulation method is adopted to establish a model for a complex regional carbon emission permit trading system. A carbon emission permit transaction system including a government, a carbon emission permit exchange and numerous emission control enterprises in reality is abstracted into a simulation-based virtual system model, so that the dynamic performance of the system under the influence of a total amount control and transaction mechanism and related risks and the emerging phenomenon caused by numerous individual decentralized decisions are analyzed; and carrying out relevant policy induction, description, depiction and evaluation and description, depiction and analysis on relevant functions of the carbon emission permit trading market according to the carbon emission permit trading market and the coping strategy of the emission control enterprise. According to the invention, some beneficial countermeasures and suggestions can be provided for the government, the carbon emission permit exchange and the enterprise, and platform support can also be provided for the government, the carbon emission permit exchange and the enterprise to carry out related policy research and evaluation and coping strategies.

Description

Technical Field

[0001] The present invention belongs to the field of carbon emission trading market and relates to modeling of a regional carbon emission trading system based on multi-agent simulation. Background Art

[0002] Globally, carbon emission trading markets have been gradually established and developed, such as the European Union Emission Trading System (EUETS) and the Clean Development Mechanism (CDM) project trading market, which provide economic incentives for emission reduction. Mandatory carbon emission reduction policies have brought new challenges to both enterprises and governments. Enterprises need to deal with the risk of insufficient carbon quotas on the basis of traditional operating risks, including purchasing additional quotas through the market or implementing internal emission reduction measures. These decisions are accompanied by financial and technical uncertainties. For the government, the pressure to reduce emissions comes from both domestic and foreign aspects. It is necessary to balance the relationship between economic development and environmental protection and build an effective carbon emission management system.

[0003] At present, the research methods for studying the problem of mandatory carbon emission reduction (whether from a macro or micro perspective) are basically limited to traditional research methods such as qualitative analysis, mathematical programming and econometric models, which have certain limitations. It is rare to use multi-agent modeling and simulation to study the new thing of mandatory carbon emission reduction (whether from a macro or micro perspective); and it is even rarer to use multi-agent modeling and simulation to study the complex management system composed of the government, carbon emission rights exchanges and a large number of emission-controlled enterprises, especially the emergent phenomena caused by the decentralized decision-making of many individuals and their impact on the overall performance of the system. Summary of the invention

[0004] The purpose of this invention is to provide a regional carbon emission trading system modeling based on multi-agent simulation, which expands the research ideas and content of the complex behavior and management of the carbon emission trading market under the background of mandatory carbon emission reduction, can provide some useful countermeasures and suggestions for the government, carbon emission trading exchanges and enterprises, and can also provide a platform support for the government, carbon emission trading exchanges and enterprises to conduct relevant policy research and evaluation and response strategies.

[0005] In order to solve the above technical problems, the present invention provides a regional carbon emission trading system modeling based on multi-agent simulation, which is characterized by including the attributes, functions and behavior logic design of enterprise agents, the attributes, functions and behavior logic design of government agents, the attributes, functions and behavior logic design of exchange agents and the interaction between agents;

[0006] The enterprise intelligent body has the functions of meeting the carbon gap, handling carbon surplus, no need for decision-making, viewing account information and updating account information. When the enterprise intelligent body has a carbon gap, it can meet its own carbon gap by purchasing carbon, self-purification and being fined. When there is a carbon surplus, it can choose to sell or hold. When it has neither a carbon gap nor a carbon surplus, it does not need to make any decision from purchasing carbon, self-purification, being fined, selling and holding. Its logic includes the top-level logic of a single enterprise intelligent body, the calculation of expected price, self-purification cost and fine cost, the logic of possible buyers, and the calculation of related risks.

[0007] The government agent has the functions of total control, quota allocation, reserve release, quota repurchase, auction, entry approval, verification, quota clearance and fines. The government agent determines the total carbon emissions of the region according to the grandfather principle and the baseline principle, that is, the budget quota, and then allocates quotas to the enterprise agent for free without exceeding the total amount. The government agent has its own reserve mechanism. Once the market situation triggers the government's reserve mechanism, the government agent will release or repurchase quotas accordingly. The government agent will hold a carbon quota auction by first issuing an announcement and then auctioning. For enterprises that meet the entry conditions, the government agent will approve their entry. At the specified time, the government agent will verify the emissions of the enterprise agent and then clear its quota. For enterprise agents that do not meet the requirements, the government agent will impose fines on them. Its logic includes total control mechanism and quota allocation mechanism, new entrant mechanism, government carbon reserve mechanism, auction mechanism, and punishment mechanism.

[0008] The exchange agent has the functions of facilitating transactions, creating enterprise accounts, updating enterprise information, deleting enterprise information, publishing market information, updating market information and deleting market information. Its logical process is as follows:

[0009] (1) Sort the sellers by selling price from low to high;

[0010] (2) Starting from the seller with the lowest price, a buyer is randomly selected from the set of buyers who can accept the price to trade with. If: ① the buyer fails to complete the transaction in full, the seller continues to wait for the transaction, leaves the trading market, and updates the seller set; ② the seller fails to complete the transaction in full, the buyer continues to wait for the transaction, leaves the trading market, and updates the buyer set; ③ the seller and the buyer both complete the transaction in full, and both leave the trading market, and update the seller set and the buyer set;

[0011] (3) This cycle continues until all tradable activities for the day are completed, leaving only: ① buyers who failed to complete a transaction, whose expected buying prices are less than all current selling prices; ② sellers who failed to complete a transaction, whose expected selling prices are greater than all current buying prices;

[0012] (4) Sellers and buyers who failed to complete a transaction adjust their respective expected prices and prepare to enter the next trading day;

[0013] The interaction between intelligent entities includes transactions between companies selling carbon quotas and companies purchasing carbon quotas in the trading market. The carbon exchange facilitates transactions between buyers and sellers. When the government reserve release is triggered, the government releases the government reserve in the carbon emission rights trading market. When the government repurchases carbon quotas, the government repurchases carbon quotas in the carbon emission rights trading market. When new emission-controlled enterprises join the carbon emission rights trading system, the government issues free quotas to the emission-controlled enterprises. The government holds carbon quota auctions for emission-controlled enterprises, and emission-controlled enterprises report their willingness to bid to the government. The government will allocate quotas to each emission-controlled enterprise, and each emission-controlled enterprise must summarize its annual output value, carbon emissions, etc. to the government every year.

[0014] Further preferably, the top-level logic of the single enterprise intelligent entity is:

[0015] At the beginning of the month, the enterprise agent will first estimate its excess emission demand Δe(t). If Δe(t) is positive, it means there is a carbon gap, and the enterprise agent must purchase carbon or perform self-purification or wait to be fined by the management department. The enterprise agent then enters the logic of possible buyers. If Δe(t) is negative, it means there is a carbon surplus, and the enterprise agent must decide whether to sell the remaining carbon quota or hold it without selling it. The enterprise agent then enters the logic of possible sellers. After exiting the logic of possible buyers or possible sellers, the relevant variables are updated, and then the next period is entered, repeating the same logic until the simulation ends.

[0016] The calculation process of the emission demand Δe(t) when the enterprise agent makes production plans at the beginning of each month is as follows:

[0017] (1) Estimate the market demand of the product d(t)~Uniform(dl,du), where dl represents the lower bound of the market demand and du represents the upper bound of the market demand;

[0018] (2) Determine the output q(t) to meet market demand and expand production;

[0019] (3) Estimate total output value TP(t) = q(t) × pd, where pd = unit product price;

[0020] (4) Convert TP(t) to standard coal equivalent: TP(t) → STC(t) = TP(t) × α1, where α1 is the coefficient for converting total output value to standard coal equivalent;

[0021] (5) Convert standard coal equivalent STC(t) to carbon dioxide equivalent e(t): STC(t)→e(t)=STC(t)×α2, where α2 is the coefficient for converting standard coal equivalent to carbon dioxide equivalent;

[0022] (6) Calculate the emission demand Δe(t) = e(t) - EQ(t), where EQ(t) = SQ(t-1) + EQA / 12, SQ(t-1) is the remaining quota in period t-1, and EQA is the annual carbon quota allocated free of charge by the government. It is assumed that the corporate intelligent agent divides the annual carbon quota allocated free of charge by the government equally into each month.

[0023] Further preferably, the calculation of the expected price, self-purification cost and penalty cost includes:

[0024] ①The first day of each month

[0025] Assume that the uncorrected expected carbon quota price is PcarbonU, the carbon quota reference price is PcarbonR, the expected price is PcarbonA, the smoothing constant is θ, the value in the model is 0.4, the carbon quota price on the previous day is PcarbonL, the initial value is 64.32, the predicted price of the carbon quota on the previous day is PcarbonFL, the initial value is 109, the influence coefficient of crude oil is Coil, the value in the model is 0.08, the influence coefficient of coal is Ccoal, the value in the model is 0.3, then

[0026] PcarbonU=θ*PcarbonL+(1-θ)*PcarbonFL

[0027] PcarbonR=PcarbonU*(1+Ccoal*Fcoal)*(1+Coil*Foil)

[0028] PcarbonA=PcarbonR*Cd

[0029] Where Cd is the coefficient of variation. In the model, Cd for possible buyers is uniformly distributed (uniform(0.9, 1.05)), and Cd for possible sellers is uniformly distributed (uniform(0.95, 1.1)). These two distributions are our assumptions and may be adjusted as the research progresses.

[0030] ②Except the first day of each month

[0031] Assume that the expected price of the previous day is PcarbonAL, the price fluctuation of carbon quota is R, and the model is uniformly distributed uniform(0, 0.5). This distribution is what we assume, that is, the fluctuation does not exceed 0.5 yuan. As the research goes deeper, it may be adjusted.

[0032] For potential buyers: PcarbonA = PcarbonAL + R

[0033] For possible sellers: PcarbonA = PcarbonAL-R

[0034] The calculation process of self-purification cost is as follows: when the cumulative self-purification emission reduction is m, the self-purification emission reduction cost of self-purification emission reduction nm is:

[0035] Among them, C0 is the initial value of the unit self-purification emission reduction cost, λ is the difficulty coefficient of self-purification emission reduction, the variable qp represents the cumulative self-purification emission reduction, and Qp is the total target self-purification emission reduction during the planning period;

[0036] The penalty cost is calculated as follows:

[0037] (1) When the unfulfilled self-purification target, i.e. the unfulfilled mandatory emission reduction, accounts for less than 10% of the mandatory emission reduction planned by the government, the unit penalty cost shall be three times the average price of the carbon emission trading market quota for the six consecutive months before the current month;

[0038] (2) When the unfulfilled self-purification target, i.e., the unfulfilled mandatory emission reduction, accounts for more than 10% of the mandatory emission reduction planned by the government, the unit penalty cost shall be four times the average price of the carbon emission trading market quota for the six consecutive months before the current month.

[0039] Further preferably, the logic of the possible buyers is:

[0040] At the beginning of each month, the expected price is compared with the unit self-cleaning cost and the unit penalty:

[0041] (1) If the expected price is the lowest, then the logic of purchasing carbon will be adopted;

[0042] (2) If the unit self-purification cost is the lowest, then the logic of self-purification will be implemented;

[0043] (3) If the unit fine is the lowest, then the enterprise agent will choose to be fined by the management department;

[0044] As a possible buyer, the business agent compares the expected price with the unit penalty at the beginning of each month:

[0045] (1) If the expected price is low, then the logic of choosing to purchase carbon will be entered;

[0046] (2) If the unit fine is low, then the enterprise agent will choose to be fined by the management department;

[0047] The logic behind choosing to purchase carbon is as follows:

[0048] First, the enterprise intelligence will determine whether the market system risk occurs;

[0049] Secondly, if the market system risk does not occur, the corporate intelligent agent will enter the carbon emission trading market to purchase carbon at the originally expected futures price; if the market system risk occurs, the corporate intelligent agent will adjust the originally expected expected price, and then enter the carbon emission trading market to purchase carbon at the adjusted expected price.

[0050] Further preferably, the calculation of the relevant risk is:

[0051] There are two marginal risks of carbon purchase and one self-purification risk involved, namely: the first risk of carbon purchase - the risk of carbon price fluctuations, the second risk of carbon purchase - the risk of failing to complete the remaining self-purification targets on schedule due to carbon purchases, and the risk of self-purification - the risk that the actual completion time is significantly longer than the planned completion time:

[0052] (1) Carbon purchase risk assessment

[0053] ① Consider the first risk of purchasing carbon

[0054] Defining risk events Carbon price fluctuations are characterized by the actual carbon price significantly exceeding the average price of the past three days;

[0055] Let p(t) be the acceptable price in period t, ons be the standard deviation of carbon prices in the past three days, p(t) = the moving average carbon price in the past three days (t-1, t-2, t-3), the probability of risk occurrence of p3(t)<p(t) is 100%, p(t)≤p3(t)<p(t)+σ M3 The probability of risk occurrence is 85%, p(t)+σ M3 ≤p3(t)<p(t)+2σ M3 The probability of risk occurrence is 60%, p(t)+2σ M3 ≤p3(t)<p(t)+3σ M3 The probability of risk occurrence is ≤15%, p(t)+3σ M3 ≤p3(t) The probability of risk occurrence is ≤15%;

[0056] ② Consider the second risk of purchasing carbon

[0057] Defining risk events Because: due to carbon purchase, the remaining self-purification target cannot be completed on schedule.

[0058] Let the remaining time be rt), t g The time required to complete the remaining self-purification target = the remaining self-purification target amount / C, where C is a coefficient, σ gThe standard deviation of the time required to complete the remaining self-purification target = CVSP*t g , (Tt<t g The probability of risk occurrence is 100%, t g ≤(Tt)<t g +σ g The probability of risk occurrence is 85%, t g +σ g ≤(Tt)<t g +2σ g The probability of risk occurrence is ≤60%, t g +2σ g ≤(Tt)<t g +3σ g The probability of risk occurrence is ≤15%, t g +3σ g ≤(Tt) The probability of risk occurrence is ≤5%;

[0059] ③ Consider two risks of carbon purchase at the same time and

[0060] When considering the combined risk of carbon purchase, the “average method” is used based on the probability of occurrence of the above two types of risks. The calculation formula is p ij =p ji =a i +b j / 2, a i for The upper limit of risk probability at different levels, b j for The probability upper limits of different risk levels are calculated to be the possibility of the occurrence of joint risks, which are organized into the following numerical form and semantic form. The numerical form can be simulated on a computer simulation platform, and the semantic form provides the logic of decision-making. When the higher The corresponding upper limit of risk probability is 0.60, and the corresponding upper limit of risk probability is 0.85. According to the above formula, the joint risk probability is 0.725, which is greater than 0.60 and less than 0.85. Therefore, the joint risk level is judged to be "high";

[0061] (2) Self-purification risk assessment

[0062] Consider the risks of self-purification

[0063] Defining risk events The actual completion time is significantly longer than the planned completion time;

[0064] Let D,(q) be the mean of the equivalent duration of the self-cleaning project = self-cleaning project equivalent / C, C is a coefficient, the standard deviation of the equivalent duration of oa = CVSP*D(q), D(q) is the actual completion time, the probability of the risk of D0(q)<D(q) is = 100%, D(q)≤D0(q)<D(q)+σ d The probability of risk occurrence is ≤85%, D(q)+σ d ≤D0(q)<D(q)+2σ d The probability of risk occurrence is ≤60%, D(q)+2σ d ≤D0(q)<D(q)+3σ d The probability of risk occurrence is ≤15%, D(q)+3σ d The probability of occurrence of ≤D0(q) risk is ≤5%.

[0065] Further preferably, the calculation process of the total amount control mechanism and the quota allocation mechanism is:

[0066] (1) Assuming that the actual regional emissions in the base year are Q0, the budget quota correction coefficient is d%, and the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, then the government's budget quota for this year can be obtained as QT = Q0*rT, where r T =r T-1 *(1-d%)*(1-R%)*(1+e%), r0=1; and the government's mandatory emission reduction target for this year, that is, the self-net emission reduction target, is: G T =Q T-1 *(1-d%)*R%*(1+e%), which does not include the self-net emission reduction targets of new entrants;

[0067] (2) Assume that the unit self-purification emission reduction costs of each enterprise in the previous year are C1, C2, ... C n The output value scale is V1, V2...V n , the standard coal quantity per 10,000 yuan of GDP is E1, E2...E n , then, we can get the self-purification emission reduction targets of each enterprise this year: G i =(V i E i / C i ) / (V1E1 / C1+V2E2 / C2+…+V n E n / C n )*G T ;

[0068] (3) Assuming that the reserved auction ratio of the region is a%, the reserved government reserve ratio is b%, and the reserved new entrant ratio is c%, then the free quota for each enterprise this year is: Qi =(V i E i / C i ) / (V1E1 / C1+V2E2 / C2+…+V n E n / C n )*Q T *(1-a%-b%-c%).

[0069] Further preferably, the calculation process of the initial quota amount of the new entrant reserve quota is:

[0070] (1) Assuming that the actual regional emissions in the base year are Q0, the budget quota correction coefficient is d%, and the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, then the government's budget quota for this year can be obtained as Q T =Q0*r T , where r T =r T-1 *(1-d%)*(1-R%)*(1+e%), r0=1;

[0071] (2) Assuming that the proportion of the region's reserved new entrants is c%, then the reserved new entrants reserve for this year can be obtained as: QR = QT * c%;

[0072] The calculation process for updating the free quota for new entrants, the self-purification emission reduction target and the reserved reserve quota for new entrants is as follows:

[0073] Assuming that the regional budget quota correction coefficient is d%, the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, the emissions of the newly entered enterprises in the same period last year are Q T-1 , the reserved auction ratio of the region is a%, the reserved government reserve ratio is b%, and the reserved new entrant ratio is c%, then the free quota for new entrants is Q T =Q T-1 *(1-d%)*(1-R%)*(1+e%)*(1-a%-b%-c%), we can also get the mandatory emission reduction for new entrants: G T =Q T-1 *(1-d%)*R%*(1+e%), at this time, the reserved quota for new entrants is Q R Update according to the following formula: Q R →Q R -Q T .

[0074] Further preferably, the calculation process of the initial quota amount of the government carbon reserve quota is:

[0075] 1. Assuming that the actual regional emissions in the base year are Q0, the budget quota correction coefficient is d%, the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, then the government's budget quota for this year can be obtained as Q T =Q0*r T , where r T =r T-1 *(1-d%)*(1-R%)*(1+e%), r0=1;

[0076] 2. Assuming that the proportion of reserved government reserves in the region is b%, then the reserved government reserves this year can be obtained as: Q b =Q T * b%;

[0077] The detailed government carbon reserve mechanism is as follows:

[0078] (1) Repurchase quota

[0079] 1) Repurchase conditions

[0080] A repurchase is taken when one of the following conditions is met:

[0081] ① When the market carbon price is lower than the given threshold P0 and is expected to continue to fall (the carbon price has fallen for three consecutive days and is expected to fall again today), and the supply exceeds the demand, a repurchase is adopted;

[0082] P0=P min -K1*σ, where P min represents the lowest carbon price in the previous year, σ represents the standard deviation of the carbon price in the previous year, K1=(CV Q +CV P ) / (1+CV Q +CV P ) represents the price-related buyback trigger correction factor, where CV Q The coefficient of variation of the previous year's trading volume, CV P represents the coefficient of variation of carbon prices in the previous year;

[0083] ② The amount of carbon quota in the carbon quota reserve is less than Q min = K*half of the highest daily trading volume in the previous year, repurchase;

[0084] 2) Repurchase quantity Q1

[0085] Considering that supply is greater than demand at this time, the repurchase quantity should be related to the changes in supply and demand. The repurchase quantity should make supply and demand tend to a balanced state. Let the supply quantity at this time be Q s1 , the demand is Q d1 , let ΔQ1=Q s1 -Q d1 , K2 is the adjustment coefficient;

[0086] Repurchase quantity Q1 = K2*ΔQ1, where K2 = (CV q +CV p ) / (1+CV q +CV p ) represents the repurchase correction coefficient, where CV q The coefficient of variation of the trading volume in the previous three days, CV p represents the coefficient of variation of carbon prices in the previous three days;

[0087] 3) Buyback price

[0088] ① The repurchase price for stabilizing the market is P1*

[0089] P1*=Mt-1+θ*σ 3(t-1) , where M t-1 represents the market carbon price of the previous day, σ 3(t-1) is the standard deviation of the market carbon price in the previous three days, θ=(CV q +CV p ) / (1+CV q +CV p ) represents the repurchase price correction factor, where CV q Indicates the coefficient of variation of the trading volume in the previous three days, CV p represents the coefficient of variation of the market carbon price in the previous three days;

[0090] ②The repurchase price for filling the carbon reserve is P2*

[0091] At this point, the repurchase department is similar to a market participant, and the repurchase price P2* = C* reference price;

[0092] (2) Release of reserves

[0093] 1) Conditions for placement

[0094] The market carbon price is higher than the given threshold P M , and it is expected to continue to rise and the market demand is greater than the supply at this time, so release is adopted;

[0095] P M =P max +K3*σ, where P max represents the highest carbon price in the previous year, σ represents the standard deviation of the carbon price in the previous year, K3=(CV Q +CV P ) / (1+CV Q +CV P ) represents the price-related trigger correction coefficient, where CV Q The coefficient of variation of the previous year's trading volume, CV Prepresents the coefficient of variation of carbon prices in the previous year;

[0096] 2) Quantity of delivery Q2

[0097] Considering that the market demand is greater than the supply at this time, the quantity of supply should be related to the changes in supply and demand, and the quantity of supply should make the supply and demand tend to a balanced state. Suppose the supply at this time is Q s2 , the demand is Q d2 , let ΔQ2=Q s2 -Q d2 , K4 is the adjustment coefficient;

[0098] The quantity of delivery Q2 = K4*ΔQ2, where K4 = (CV q +CV p ) / (1+CV q +CV p ) represents the correction factor of the delivery amount, where CV q The coefficient of variation of the trading volume in the previous three days, CV p represents the coefficient of variation of carbon prices in the previous three days;

[0099] 3) The price of the advertisement

[0100] The repurchase price for market stabilization is P3*

[0101] P3*=M t-1 -θ*σ 3(t-1) , where M t-1 represents the market carbon price of the previous day, σ 3(t-1) is the standard deviation of the market carbon price in the previous three days, θ=(CV q +CV p ) / (1+CV q +CV p ) represents the price correction factor, where CV q The coefficient of variation of the trading volume in the previous three days, CV p Represents the coefficient of variation of the market carbon price in the previous three days.

[0102] Further preferably, the calculation process of the initial quota amount of the auction mechanism is:

[0103] 1. Assuming that the actual regional emissions in the base year are Q0, the budget quota correction coefficient is d%, the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, then the government's budget quota for this year can be obtained as Q T =Q0*r T , where r T =r T-1 *(1-d%)*(1-R%)*(1+e%), r0=1;

[0104] 2. Assuming that the proportion of the reserved auction quota in the region is a%, then the reserved auction quota for this year can be obtained as: Q A =Q T *a%.

[0105] Further preferably, the penalty mechanism is:

[0106] When the unfulfilled self-purification target, i.e. the unfulfilled mandatory emission reduction, accounts for less than 10% of the mandatory emission reduction planned by the government, the unit penalty cost shall be three times the average price of the carbon emission trading market quota for the six consecutive months before the current month;

[0107] When the unfulfilled self-purification target, that is, the unfulfilled mandatory emission reduction, accounts for more than 10% of the government's planned mandatory emission reduction, the unit penalty cost shall be four times the average price of the carbon emission trading market quota for the six consecutive months before that month.

[0108] Compared with the prior art, the present invention has the following beneficial effects:

[0109] First, under the mandatory carbon emission reduction conditions, we should focus on the new and huge challenges faced by enterprises in actual operations from the perspective of emission sources (micro perspective). For example, under the mandatory carbon emission reduction environment, how should enterprises analyze and evaluate various risks based on the environment and their own characteristics, and make an effective balance between different risks and interests to minimize risks and increase profits;

[0110] Second, the government, carbon emission rights exchanges and a large number of emission-controlled enterprises, that is, the various entities of the carbon emission rights trading system, are integrated into a complex system to study the emergence phenomenon caused by the decentralized decision-making of many individuals and its impact on the overall performance of the system;

[0111] Third, use multi-agent modeling and simulation to study the complex system of regional carbon emission trading system composed of the government, carbon emission trading exchanges and a large number of emission-controlled enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0112] Figure 1 It is the top-level logical flow chart of a single enterprise intelligent entity;

[0113] Figure 2 It is the historical trend chart of international crude oil futures prices;

[0114] Figure 3 It is a historical chart of coal prices;

[0115] Figure 4 It is a logical flow chart of the multi-source risk decision-making and analysis experimental system;

[0116] Figure 5 It is a logical flow chart of possible buyers;

[0117] Figure 6 It is the calculation process of budget quota, self-purification emission reduction target and free quota;

[0118] Figure 7 It is the process of calculating the initial quota amount of the reserve quota of the new entrant;

[0119] Figure 8 It is the calculation process of updating the free quotas for new entrants, the self-purification emission reduction targets and the reserved reserve quotas for new entrants;

[0120] Fig. 9 It is the process of calculating the initial quota amount of the government's carbon reserve quota;

[0121] Fig.10 It is the process of calculating the initial quota amount of auctionable quotas;

[0122] Fig.11 It is a market transaction logic flow chart;

[0123] Fig.12 is the interaction graph between agents. DETAILED DESCRIPTION

[0124] The following is a further detailed description of the modeling of a regional carbon emission trading system based on multi-agent simulation proposed by the present invention in combination with the accompanying drawings and specific embodiments. According to the following description, the advantages and features of the present invention will become clearer. It should be noted that the accompanying drawings are all in a very simplified form and are not in precise proportions, which are only used to conveniently and clearly assist in explaining the purpose of the embodiments of the present invention. The same or similar figure numbers in the accompanying drawings represent the same or similar components.

[0125] Embodiment: The present invention provides a modeling of a regional carbon emission trading system based on multi-agent simulation. The conceptual modeling of the present invention is mainly developed from four aspects: the attributes, functions and behavioral logic design of enterprise agents, the attributes, functions and behavioral logic design of government agents, the attributes, functions and behavioral logic design of exchange agents, and the interaction between agents.

[0126] 1. Attributes, functions and behavior logic design of enterprise intelligent agents

[0127] The enterprise intelligent body has the functions of meeting the carbon gap, handling the carbon surplus, not requiring decision-making, viewing account information, and updating account information. When the enterprise intelligent body has a carbon gap, it can meet its carbon gap through the following channels: (1) purchasing carbon; (2) self-purification; (3) being fined; when there is a carbon surplus, it can choose to sell or hold; and when it has neither a carbon gap nor a carbon surplus, it does not need to make any decision from purchasing carbon, self-purification, being fined, selling, and holding.

[0128] 1.1 The top-level logic of a single enterprise intelligent entity

[0129] The present invention defines the attributes, functions and behavior logic of the enterprise intelligent body in detail. The top-level logic flow chart of a single enterprise intelligent body is as follows: Figure 1 shown.

[0130] At the beginning of the month, the corporate agent will first estimate its excess emission demand Δe(t). If Δe(t) is positive, it means there is a carbon gap, and the corporate agent must purchase carbon or perform self-purification or wait to be fined by the management department. The corporate agent then enters the logic of possible buyers. If Δe(t) is negative, it means there is a carbon surplus, and the corporate agent must decide whether to sell the remaining carbon quota or hold it without selling it. The corporate agent then enters the logic of possible sellers. After coming out of the logic of possible buyers or possible sellers, the relevant variables are updated, and then the next period is entered, repeating the same logic until the simulation terminates.

[0131] The calculation process of the emission demand Δe(t) when the enterprise agent makes production plans at the beginning of each month is as follows:

[0132] (1) Estimate the market demand of the product d(t)~Uniform(dl,du), where dl represents the lower bound of the market demand and du represents the upper bound of the market demand;

[0133] (2) Determine the output q(t) to meet market demand and expand production;

[0134] (3) Estimate total output value TP(t) = q(t) × pd, where pd = unit product price;

[0135] (4) Convert TP(t) to standard coal equivalent: TP(t) → STC(t) = TP(t) × α1, where α1 is the coefficient for converting total output value to standard coal equivalent;

[0136] (5) Convert standard coal equivalent STC(t) to carbon dioxide equivalent e(t): STC(t)→e(t)=STC(t)×α2, where α2 is the coefficient for converting standard coal equivalent to carbon dioxide equivalent;

[0137] (6) Calculate the emission demand Δe(t) = e(t) - EQ(t), where EQ(t) = SQ(t-1) + EQA / 12, SQ(t-1) is the remaining quota in period t-1, and EQA is the annual carbon quota allocated free of charge by the government. It is assumed that the corporate intelligent agent divides the annual carbon quota allocated free of charge by the government equally into each month.

[0138] 1.2 Calculation of expected price, self-purification cost and penalty cost

[0139] Factors affecting carbon prices include crude oil, coal, electricity, temperature, etc. Considering the impact of exogenous factors on carbon prices, the reference price at the beginning of each month in the model of the present invention is obtained by using the exponential smoothing forecasting method and correcting the results of the exponential smoothing forecasting method using the fluctuation range of crude oil and coal and their influence coefficients. Details are as follows:

[0140] (1) Assume that the forecast price of international crude oil futures on the current day is PoilFT, the smoothing constant is α (the value in the model is 0.4), the international crude oil futures price on the previous day is PoilL (the initial value is 109), the forecast price of international crude oil futures on the previous day is PoilFL (the initial value is 109), and the fluctuation range of the forecast price of international crude oil futures on the current day compared with the international crude oil futures price on the previous day is Foil, then: PoilFT = α*PoilL + (1-α)*PoilFLFoil = (PoilFT-PoilL) / PoilL

[0141] The actual price of international crude oil futures on that day, PoilAT, is from the normal distribution normal (109, 3.89 2 ), that is, PoilAT = normal (109, 3.89 2 ), which is obtained by fitting the distribution of international crude oil futures prices (in US dollars) through the simulation platform Arena. The detailed information of the fitting results is as follows:

[0142] Distribution Summary

[0143] Distribution: Normal

[0144] Expression:NORM(109,3.89 2 )SquareError:0.008412

[0145] ChiSquareTest

[0146] Numberofintervals=13

[0147] Degrees of freedom = 10

[0148] Test Statistic = 33.3

[0149] Corresponding p-value<0.005

[0150] Kolmogorov-Smirnov Test

[0151] Test Statistic = 0.0632

[0152] Corresponding p-value>0.15

[0153] Data Summary

[0154] Number of Data Points = 316

[0155] Min DataValue=98.9

[0156] Max DataValue=119

[0157] Sample Mean=109

[0158] Sample Std Dev=3.89

[0159] Histogram Summary

[0160] Histogram Range = 98 to 119

[0161] Number ofIntervals=17

[0162] It can be seen that the t-value of the chi-square test is 33.3, and its corresponding p-value is less than 0.05; while the t-value of the KS test is 0.0632, and its corresponding p-value is greater than 0.15.

[0163] (2) Coal (the forecast price of coal for the day is calculated using the trend exponential smoothing method)

[0164] Assume that the predicted price of coal for the day is PcoalFT, the smoothing constant is β (the value in the model is 0.4), the trend smoothing constant is γ (the value in the model is 0.4), the coal price of the previous day is PcoalL (the initial value is 590), the coal price of the day before is PcoalLB (the initial value is 590), the predicted price of coal the day before is PcoalFL (the initial value is 590), the predicted price of coal the day before is PcoalFLB (the initial value is 590), the trend of coal is estimated to be Tcoal, the last trend of coal is estimated to be TcoalL (the initial value is 0), the recent trend of coal is TcoalR, and the fluctuation range of the predicted price of coal on the day compared with the coal price of the previous day is Fcoal, then

[0165] PcoalFT=β*PcoalL+(1-β)*PcoalFL+Tcoal

[0166] Tcoal=γ*TcoalR+(1-γ)*TcoalL

[0167] TcoalR=β*(PcoalL-PcoalLB)+(1-β)*(PcoalFL-PcoalFLB)

[0168] Fcoal=(PcoalFT-PcoalL) / PcoalL

[0169] The actual price of coal on that day, PcoalAT, is taken from the beta distribution 530+103*beta(0.586, 0.507), that is, PcoalAT=530+103*beta(0.586, 0.507). This distribution is obtained by fitting the Bohai Rim Thermal Coal Price Index (BSPI) from January 9, 2013 to December 24, 2013 through the input analyzer of the simulation platform Arena (priced in RMB). The detailed information of the fitting results is as follows:

[0170] Distribution Summary

[0171] Distribution:Beta

[0172] Expression:530+103*BETA(0.586,0.507)

[0173] Square Error:0.013085

[0174] Chi Square Test

[0175] Number of intervals = 6

[0176] Degrees of freedom = 3

[0177] Test Statistic = 4.22

[0178] Corresponding p-value=0.242

[0179] Kolmogorov-Smirnov Test

[0180] Test Statistic = 0.155

[0181] Corresponding p-value>0.15

[0182] Data Summary

[0183] Number of Data Points = 47

[0184] Min DataValue = 530

[0185] Max DataValue = 633

[0186] Sample Mean=590

[0187] Sample Std Dev=35.2

[0188] Histogram Summary

[0189] HistogramRange = 530 to 633

[0190] NumberofIntervals=6

[0191] It can be seen that the t-value of the chi-square test is 4.22, and its corresponding p-value is 0.242; while the t-value of the KS test is 0.155, and its corresponding p-value is greater than 0.15. This distribution fit has passed the chi-square test.

[0192] The reason why the forecast price of international crude oil futures on the day is obtained by exponential smoothing forecasting method instead of trend exponential smoothing method is that from the historical trend of international crude oil futures prices, there is no obvious trend, such as Figure 2 The reason why the forecast price of coal for the day is obtained by trend exponential smoothing method instead of exponential smoothing prediction method is that there is an obvious trend in the historical trend of coal prices, such as Figure 3 shown.

[0193] (3) Carbon quota reference price and expected price

[0194] ①The first day of each month

[0195] Assume that the uncorrected expected carbon quota price is PcarbonU, the carbon quota reference price is PcarbonR, the expected price is PcarbonA, the smoothing constant is θ, the value in the model is 0.4, the carbon quota price on the previous day is PcarbonL, the initial value is 64.32, the predicted price of the carbon quota on the previous day is PcarbonFL, the initial value is 109, the influence coefficient of crude oil is Coil, the value in the model is 0.08, the influence coefficient of coal is Ccoal, the value in the model is 0.3, then:

[0196] PcarbonU=θ*PcarbonL+(1-θ)*PcarbonFL

[0197] PcarbonR=PcarbonU*(1+Ccoal*Fcoal)*(1+Coil*Foil)

[0198] PcarbonA=PcarbonR*Cd

[0199] Among them, Cd is the coefficient of variation. In the model, Cd for possible buyers is uniformly distributed uniform(0.9, 1.05), and Cd for possible sellers is uniformly distributed uniform(0.95, 1.1). These two distributions are our assumptions and may be adjusted as the research goes deeper.

[0200] ②Except the first day of each month

[0201] Assume that the expected price of the previous day is PcarbonAL, the price fluctuation of carbon quota is R, and the model is uniformly distributed uniform(0, 0.5). This distribution is what we assume, that is, we assume that the fluctuation does not exceed 0.5 yuan. As the research goes deeper, it may be adjusted:

[0202] For potential buyers: PcarbonA = PcarbonAL + R

[0203] For possible sellers: PcarbonA = PcarbonAL-R

[0204] The calculation process of self-purification cost is as follows: when the cumulative self-purification emission reduction is m, the self-purification emission reduction cost of self-purification emission reduction nm is:

[0205] Among them, C0 is the initial value of the unit self-purification emission reduction cost, λ is the difficulty coefficient of self-purification emission reduction, the variable qp represents the cumulative self-purification emission reduction, and Qp is the target total self-purification emission reduction during the planning period.

[0206] The penalty cost is calculated as follows:

[0207] (1) When the unfulfilled self-purification target, i.e. the unfulfilled mandatory emission reduction, accounts for less than 10% of the mandatory emission reduction planned by the government, the unit penalty cost shall be three times the average price of the carbon emission trading market quota for the six consecutive months before the current month;

[0208] (2) When the unfulfilled self-purification target, i.e., the unfulfilled mandatory emission reduction, accounts for more than 10% of the mandatory emission reduction planned by the government, the unit penalty cost shall be four times the average price of the carbon emission trading market quota for the six consecutive months before the current month.

[0209] The logic of the multi-source risk decision-making and analysis experimental system of the present invention is as follows Figure 4 shown.

[0210] 1.3 Possible Buyer’s Logic

[0211] Now let's look at the logic of possible buyers (this logic is a step-by-step decision-making logic that simulates the psychological decision-making process of ordinary people). When the emission demand Δe(t) of the enterprise intelligent body is positive, it is a possible buyer. Its logic flow chart is as follows: Figure 5 As shown. At the beginning of each month, the business agent, as a possible buyer, compares the expected price with the unit self-cleaning cost and the unit penalty:

[0212] (1) If the expected price is the lowest, then the logic of purchasing carbon will be adopted;

[0213] (2) If the unit self-purification cost is the lowest, then the logic of self-purification will be implemented;

[0214] (3) If the unit fine is the lowest, then the enterprise agent will choose to be fined by the management department;

[0215] As a possible buyer, the business agent compares the expected price with the unit penalty at the beginning of each month:

[0216] (1) If the expected price is low, then the logic of choosing to purchase carbon will be entered;

[0217] (2) If the unit fine is low, then the enterprise agent will choose to be fined by the management department.

[0218] The logic behind choosing to purchase carbon is as follows:

[0219] First, the enterprise intelligent body will judge whether the market system risk occurs. The so-called market system risk mainly refers to some unexpected risks that will cause carbon price fluctuations in the carbon emission rights trading market; secondly, if the market system risk does not occur, the enterprise intelligent body will enter the carbon emission rights trading market to purchase carbon at the expected price; if the market system risk occurs, the enterprise intelligent body will adjust the expected price, and then enter the carbon emission rights trading market to purchase carbon at the adjusted expected price.

[0220] 1.4 Calculation of relevant risks

[0221] The above logic of possible buyers involves two marginal risks of carbon purchase and one self-purification risk, namely: the first risk of carbon purchase - the risk of carbon price fluctuation (characterized by the risk that the actual carbon price significantly exceeds the average price of the past three days), the second risk of carbon purchase - the risk of failing to complete the remaining self-purification target on schedule due to carbon purchase, and the risk of self-purification - the risk that the actual completion time is significantly longer than the planned completion time. Let's explain these three in detail below.

[0222] (1) Carbon purchase risk assessment

[0223] ① Consider the first risk of purchasing carbon

[0224] Defining risk events Carbon price fluctuations are characterized by the actual carbon price significantly exceeding the average price of the past three days; let p(t) be the acceptable price in period t, ons be the standard deviation of the carbon price in the past three days, p(t) = the moving average carbon price of the past three days (t-1, t-2, t-3), then:

[0225] Table 1 Risk probability table of actual carbon price significantly exceeding the average price of the past three days

[0226]

[0227]

[0228] ② Consider the second risk of purchasing carbon

[0229] Defining risk events = (t) = (t) = (t) = (t) g is the time required to complete the remaining self-purification target (= remaining self-purification target amount / C, C is a coefficient (e.g. 12000), where it is assumed that there is a linear relationship between the time required to complete the remaining self-purification target and the remaining self-purification target amount), σ g The standard deviation of the time required to complete the remaining self-purification target (=CVSP*t g , assuming here that the coefficient of variation of the time required for self-purification (CVSP) is 0.2), then:

[0230] Table 2 Risk probability of not being able to complete the remaining self-purification targets on schedule due to carbon purchase

[0231] Risk Judgment Rules Risk Level Probability of risk occurrence <![CDATA[(T-t<t g ]]> Very high ＝100％ <![CDATA[t g ≤(T-t)<t g +σ g ]]> high ≤85％ <![CDATA[t g +σ g ≤(T-t)<t g +2σ g ]]> Higher ≤60％ <![CDATA[t g +2σ g ≤(T-t)<t g +3σ g ]]> Lower ≤15％ <![CDATA[t g +3σ g ≤(T-t)]]> Low ≤5％

[0232] ③ Consider two risks of carbon purchase at the same time and

[0233] When considering the combined risk of carbon purchase, the “average method” is used based on the probability of occurrence of the above two types of risks. The calculation formula is p ij =p ji =a i +b j / 2, a i for The upper limit of risk probability at different levels, b j for The probability upper limits of different risk levels are calculated to be the possibility of the occurrence of joint risks, which are organized into the following numerical form and semantic form. The numerical form can be simulated on a computer simulation platform, and the semantic form provides the logic of decision-making. When the higher The corresponding upper limit of risk probability is 0.60, and the corresponding upper limit of high risk probability is 0.85. According to the above formula, the joint risk probability is 0.725. This probability value is greater than 0.60 and less than 0.85. Therefore, the joint risk level is judged to be "high".

[0234] (2) Self-purification risk assessment

[0235] Consider the risks of self-purification Defining risk events The actual completion time is significantly greater than the planned completion time; let D,(q) be the mean of the equivalent duration of the self-cleaning project (= self-cleaning project equivalent / C, C is a coefficient (e.g. 12000)), the standard deviation of the equivalent duration of oa (= CVSP*D(q), assuming here that the coefficient of variation of the time required for self-cleaning is 0.2), and D(q) be the actual completion time, then:

[0236] Table 3 Risk probability table of actual completion time being significantly greater than planned completion time

[0237] Risk Judgment Rules Risk Level Probability of risk occurrence <![CDATA[D0(q)<D(q)]]> Very high ＝100％ <![CDATA[D(q)≤D0(q)<D(q)+σ d ]]> high ≤85％ <![CDATA[D(q)+σ d ≤D0(q)<D(q)+2σ d ]]> Higher ≤60％ <![CDATA[D(q)+2σ d ≤D0(q)<D(q)+3σ d ]]> Lower ≤15％ <![CDATA[D(q)+3σ d ≤D0(q)]]> Low ≤5％

[0238] 2. Attributes, functions and behavior logic design of government agents

[0239] The government agent has the functions of total control, quota allocation, reserve release, quota repurchase, auction, entry approval, verification, quota clearance and fines. The government agent determines the total carbon emissions of the region according to the grandfather principle and the baseline principle, that is, the budget quota, and then allocates quotas to the enterprise agent for free without exceeding the total amount. The government agent will comprehensively consider the budget quota, the emission reduction cost of the emission-controlled enterprise, the output value, the standard coal volume per 10,000 yuan of GDP, the reserved auction ratio, the reserved government reserve ratio, and the reserved new entrant ratio when allocating quotas to the enterprise agent for free. This free quota allocation plan is conducive to the survival of the fittest among enterprises, and also takes proper care of enterprises with high standard coal volume per 10,000 yuan of GDP. The government agent has its own set of reserve mechanisms. Once the market situation triggers the government's reserve mechanism, the government agent will release or repurchase quotas accordingly. The government agent will hold a carbon quota auction by first issuing an announcement and then conducting an auction. For enterprises that meet the entry conditions, the government agent will approve their entry. At the specified time, the government agent will verify the emissions of the enterprise agent and then clear its quota. The government agent will impose fines on corporate agents that fail to meet the requirements.

[0240] 2.1 Total quantity control mechanism and quota allocation mechanism

[0241] The calculation process of budget quota, self-purification emission reduction target and free quota is as follows: Figure 6As shown:

[0242] (1) Assuming that the actual regional emissions in the base year are Q0, the budget quota correction coefficient is d%, and the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, then the government's budget quota for this year can be obtained as QT = Q0*rT, where r T =r T-1 *(1-d%)*(1-R%)*(1+e%), r0=1; and the government's mandatory emission reduction target for this year, that is, the self-net emission reduction target, is: G T =Q T-1 *(1-d%)*R%*(1+e%), which does not include the self-purification emission reduction targets of new entrants.

[0243] (2) Assume that the unit self-purification emission reduction costs of each enterprise in the previous year are C1, C2, ... C n The output value scale is V1, V2...V n , the standard coal quantity per 10,000 yuan of GDP is E1, E2...E n , then, we can get the self-purification emission reduction targets of each enterprise this year: G i =(V i E i / C i ) / (V1E1 / C1+V2E2 / C2+…+V n E n / C n )*G T .

[0244] (3) Assuming that the reserved auction ratio of the region is a%, the reserved government reserve ratio is b%, and the reserved new entrant ratio is c%, then the free quota for each enterprise this year is: Q i =(V i E i / C i ) / (V1E1 / C1+V2E2 / C2+…+V n E n / C n )*Q T *(1-a%-b%-c%).

[0245] 2.2 New Entrant Mechanism

[0246] The calculation process of the initial quota of the reserve quota of the new entrant is as follows: Figure 7 As shown:

[0247] (1) Assuming that the actual regional emissions in the base year are Q0, the budget quota correction coefficient is d%, and the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, then the government's budget quota for this year can be obtained as Q T =Q0*r T , where r T =r T-1 *(1-d%)*(1-R%)*(1+e%), r0=1;

[0248] (2) Assuming that the proportion of the region's reserved new entrants is c%, then the reserved new entrants reserve for this year can be obtained as: QR = QT * c%;

[0249] The calculation process for updating the free quota for new entrants, the self-purification emission reduction target and the reserved reserve quota for new entrants is as follows: Figure 8 As shown:

[0250] Assuming that the regional budget quota correction coefficient is d%, the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, the emissions of the newly entered enterprises in the same period last year are Q T-1 , the reserved auction ratio of the region is a%, the reserved government reserve ratio is b%, and the reserved new entrant ratio is c%, then the free quota for new entrants is Q T =Q T-1 *(1-d%)*(1-R%)*(1+e%)*(1-a%-b%-c%), we can also get the mandatory emission reduction for new entrants: G T =Q T-1 *(1-d%)*R%*(1+e%), at this time, the reserved quota for new entrants is Q R Update according to the following formula: QR→QR-QT.

[0251] 2.3 Government carbon reserve mechanism

[0252] The calculation process of the initial quota of the government carbon reserve quota is as follows: Fig. 9 As shown:

[0253] 1. Assuming that the actual regional emissions in the base year are Q0, the budget quota correction coefficient is d%, the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, then the government's budget quota for this year can be obtained as Q T =Q0*r T , where r T =r T-1 *(1-d%)*(1-R%)*(1+e%), r0=1.

[0254] 2. Assuming that the proportion of reserved government reserves in the region is b%, then the reserved government reserves this year can be obtained as: Q b =Q T *b%.

[0255] The detailed government carbon reserve mechanism is as follows:

[0256] (1) Repurchase quota

[0257] 1) Repurchase conditions

[0258] A repurchase is taken when one of the following conditions is met:

[0259] ① When the market carbon price is lower than the given threshold P0 and is expected to continue to fall (the carbon price has fallen for three consecutive days and is expected to fall again today) and the supply is greater than the demand, a repurchase is adopted; P0 = P min -K1*σ, where P min represents the lowest carbon price in the previous year, σ represents the standard deviation of the carbon price in the previous year, K1=(CV Q +CV P ) / (1+CV Q +CV P ) represents the price-related buyback trigger correction factor, where CV Q The coefficient of variation of the previous year's trading volume, CV P represents the coefficient of variation of carbon prices in the previous year; ② The amount of carbon quotas in the carbon quota reserve is less than Q min =K*half of the highest daily trading volume in the previous year, and repurchase will be adopted.

[0260] 2) Repurchase quantity Q1

[0261] Considering that supply is greater than demand at this time, the repurchase quantity should be related to the changes in supply and demand. The repurchase quantity should make supply and demand tend to a balanced state. Let the supply quantity at this time be Q s1 , the demand is Q d1 , let ΔQ1=Q s1 -Q d1 , K2 is the adjustment coefficient; repurchase quantity Q1 = K2*ΔQ1, where K2 = (CV q +CV p ) / (1+CV q +CV p ) represents the repurchase correction coefficient, where CV q The coefficient of variation of the trading volume in the previous three days, CV p represents the coefficient of variation of carbon prices in the previous three days.

[0262] 3) Buyback price

[0263] ① The repurchase price for stabilizing the market is P1*

[0264] P1*=Mt-1+θ*σ 3(t-1) , where M t-1 represents the market carbon price of the previous day, σ 3(t-1) is the standard deviation of the market carbon price in the previous three days, θ=(CV q +CV p ) / (1+CV q +CV p ) represents the repurchase price correction factor, where CV q Indicates the coefficient of variation of the trading volume in the previous three days, CV p Represents the coefficient of variation of the market carbon price in the previous three days.

[0265] ②The repurchase price for filling the carbon reserve is P2*

[0266] At this time, the repurchase department is similar to a market participant, and the repurchase price P2* = C* reference price.

[0267] (2) Release of reserves

[0268] 1) Conditions for placement

[0269] The market carbon price is higher than the given threshold P M , and it is expected to continue to rise and when the market demand is greater than the supply, the release is adopted. M =P max +K3*σ, where P max represents the highest carbon price in the previous year, σ represents the standard deviation of the carbon price in the previous year, K3=(CV Q +CV P ) / (1+CV Q +CV P ) represents the price-related trigger correction coefficient, where CV Q The coefficient of variation of the previous year's trading volume, CV P represents the coefficient of variation of carbon prices in the previous year.

[0270] 2) Quantity of delivery Q2

[0271] Considering that the market demand is greater than the supply at this time, the quantity of supply should be related to the changes in supply and demand, and the quantity of supply should make the supply and demand tend to a balanced state. Suppose the supply at this time is Q s2 , the demand is Q d2 , let ΔQ2=Q s2 -Q d2 , K4 is the adjustment coefficient. The quantity of delivery Q2 = K4*ΔQ2, where K4 = (CV q +CV p ) / (1+CV q +CV p ) represents the correction factor of the delivery amount, where CVq The coefficient of variation of the trading volume in the previous three days, CV p represents the coefficient of variation of carbon prices in the previous three days.

[0272] 3) The price of the advertisement

[0273] The repurchase price for market stabilization is P3*

[0274] P3*=M t-1 -θ*σ 3(t-1) , where M t-1 represents the market carbon price of the previous day, σ 3(t-1) is the standard deviation of the market carbon price in the previous three days, θ=(CV q +CV p ) / (1+CV q +CV p ) represents the price correction factor, where CV q The coefficient of variation of the trading volume in the previous three days, CV p Represents the coefficient of variation of the market carbon price in the previous three days.

[0275] 2.4 Auction Mechanism

[0276] The calculation process of the initial quota amount of auctionable quota is as follows: Fig.10 As shown:

[0277] 1. Assuming that the actual regional emissions in the base year are Q0, the budget quota correction coefficient is d%, the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, then the government's budget quota for this year can be obtained as Q T =Q0*r T , where r T =r T-1 *(1-d%)*(1-R%)*(1+e%), r0=1.

[0278] 2. Assuming that the proportion of the reserved auction quota in the region is a%, then the reserved auction quota for this year can be obtained as: Q A =Q T *a%.

[0279] 2.5 Penalty Mechanism

[0280] When the unfulfilled self-purification target, that is, the unfulfilled mandatory emission reduction accounts for less than 10% of the government's planned mandatory emission reduction, the unit penalty cost is three times the average price of the carbon emission trading market quota for the six consecutive months before that month.

[0281] When the unfulfilled self-purification target, that is, the unfulfilled mandatory emission reduction, accounts for more than 10% of the government's planned mandatory emission reduction, the unit penalty cost shall be four times the average price of the carbon emission trading market quota for the six consecutive months before that month.

[0282] 3. Attributes, functions and behavior logic design of exchange agents

[0283] The exchange agent has the functions of facilitating transactions, creating enterprise accounts, updating enterprise information, deleting enterprise information, publishing market information, updating market information and deleting market information. The transaction method is mainly splittable pricing point selection, the logic flow chart is as follows Fig.11 shown.

[0284] (1) Sort the sellers by selling price from low to high.

[0285] (2) Starting from the seller with the lowest price, a buyer is randomly selected from the set of buyers who can accept the price to trade. If: ① the buyer fails to complete the transaction in full, the seller continues to wait for the transaction, leaves the trading market, and the seller set is updated; ② the seller fails to complete the transaction in full, the buyer continues to wait for the transaction, leaves the trading market, and the buyer set is updated; ③ the seller and the buyer both complete the transaction in full, and both the seller and the buyer leave the trading market, and the seller set and the buyer set are updated.

[0286] (3) This cycle continues until all tradable activities for the day are completed, leaving only: ① buyers who failed to complete a transaction, whose expected purchase price is less than all current selling prices; and ② sellers who failed to complete a transaction, whose expected selling price is greater than all current purchase prices.

[0287] (4) Sellers and buyers who failed to complete a transaction adjust their expected prices and prepare to enter the next trading day.

[0288] 4 Interactions between Agents

[0289] The interaction between agents in the present invention is as follows Fig.12 As shown:

[0290] The emission trading system, which is composed of the government, carbon emission rights exchange and many emission-controlled enterprises, is such a system. It has many levels, and each level has its own structure, such as regions, governments, carbon emission rights exchanges, industries and many emission-controlled enterprises. Individuals will interact with each other, for example, enterprises that sell carbon quotas and enterprises that buy carbon quotas trade in the trading market, and carbon exchanges facilitate transactions between buyers and sellers. When the government reserves are released, the government releases government reserves in the carbon emission rights trading market. When the government repurchases carbon quotas, the government repurchases carbon quotas in the carbon emission rights trading market. When new emission-controlled enterprises join the carbon emission rights trading system, the government issues free quotas to the emission-controlled enterprises. The government holds carbon quota auctions for emission-controlled enterprises, and emission-controlled enterprises report their willingness to bid to the government. The government will allocate quotas to each emission-controlled enterprise. Each emission-controlled enterprise must summarize its annual output value, carbon emissions, etc. to the government every year, and so on. The various interactions between individuals combined together are like a complex giant web. Every individual in this giant web will be directly or indirectly affected by other individuals or influence other individuals. It can be said that a single move can affect the entire body.

[0291] The present invention adopts a multi-agent modeling and simulation method to establish a model for the complex system of regional carbon emission trading system (that is, the carbon emission trading system in reality that includes the government, carbon emission exchange and many emission-controlled enterprises is abstracted into a virtual system model based on simulation), so as to analyze the dynamic performance of the system under the influence of total control and trading mechanism and related risks (such as the risk of carbon price fluctuations, the risk of failure to complete the self-purification emission reduction targets required by the government on schedule, and the risk of self-purification emission reduction period exceeding the budget), the emergence phenomenon caused by the decentralized decision-making of many individuals (that is, the government, emission exchange, and each emission-controlled enterprise), and the response strategies of emission-controlled enterprises, and conduct relevant policy summarization, description, characterization and evaluation, as well as description, characterization and analysis of the relevant functions of the carbon emission trading market.

[0292] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0293] The above description is only a description of the preferred embodiments of the present invention, and is not intended to limit the scope of the present invention. Any changes or modifications made by a person skilled in the art in the field of the present invention based on the above disclosure shall fall within the scope of protection of the claims.

Claims

1. A regional carbon emission trading system modeling based on multi-agent simulation, characterized in that: Including the attributes, functions and behavior logic design of enterprise intelligent agents, the attributes, functions and behavior logic design of government intelligent agents, the attributes, functions and behavior logic design of exchange intelligent agents and the interaction between intelligent agents; The enterprise intelligent body has the functions of meeting the carbon gap, handling carbon surplus, no need for decision-making, viewing account information and updating account information. When the enterprise intelligent body has a carbon gap, it can meet its own carbon gap by purchasing carbon, self-purification and being fined. When there is a carbon surplus, it can choose to sell or hold. When it has neither a carbon gap nor a carbon surplus, it does not need to make any decision from purchasing carbon, self-purification, being fined, selling and holding. Its logic includes the top-level logic of a single enterprise intelligent body, the calculation of expected price, self-purification cost and fine cost, the logic of possible buyers, and the calculation of related risks. The government agent has the functions of total control, quota allocation, reserve release, quota repurchase, auction, entry approval, verification, quota clearance and fines. The government agent determines the total carbon emissions of the region according to the grandfather principle and the baseline principle, that is, the budget quota, and then allocates quotas to the enterprise agent for free without exceeding the total amount. The government agent has its own reserve mechanism. Once the market situation triggers the government's reserve mechanism, the government agent will release or repurchase quotas accordingly. The government agent will hold a carbon quota auction by first issuing an announcement and then auctioning. For enterprises that meet the entry conditions, the government agent will approve their entry. At the specified time, the government agent will verify the emissions of the enterprise agent and then clear its quota. For enterprise agents that do not meet the requirements, the government agent will impose fines on them. Its logic includes total control mechanism and quota allocation mechanism, new entrant mechanism, government carbon reserve mechanism, auction mechanism, and punishment mechanism. The exchange agent has the functions of facilitating transactions, creating enterprise accounts, updating enterprise information, deleting enterprise information, publishing market information, updating market information and deleting market information. Its logical process is as follows: (1) Sort the sellers by selling price from low to high; (2) Starting from the seller with the lowest price, a buyer is randomly selected from the set of buyers who can accept the price to trade with. If: ① the buyer fails to complete the transaction in full, the seller continues to wait for the transaction, leaves the trading market, and updates the seller set; ② the seller fails to complete the transaction in full, the buyer continues to wait for the transaction, leaves the trading market, and updates the buyer set; ③ the seller and the buyer both complete the transaction in full, and both leave the trading market, and update the seller set and the buyer set; (3) This cycle continues until all tradable activities for the day are completed, leaving only: ① buyers who failed to complete a transaction, whose expected buying prices are less than all current selling prices; ② sellers who failed to complete a transaction, whose expected selling prices are greater than all current buying prices; (4) Sellers and buyers who failed to complete a transaction adjust their respective expected prices and prepare to enter the next trading day; The interaction between intelligent entities includes transactions between companies selling carbon quotas and companies purchasing carbon quotas in the trading market. The carbon exchange facilitates transactions between buyers and sellers. When the government reserve release is triggered, the government releases the government reserve in the carbon emission rights trading market. When the government repurchases carbon quotas, the government repurchases carbon quotas in the carbon emission rights trading market. When new emission-controlled enterprises join the carbon emission rights trading system, the government issues free quotas to the emission-controlled enterprises. The government holds carbon quota auctions for emission-controlled enterprises, and emission-controlled enterprises report their willingness to bid to the government. The government will allocate quotas to each emission-controlled enterprise, and each emission-controlled enterprise must summarize its annual output value, carbon emissions, etc. to the government every year.

2. According to claim 1, a regional carbon emission trading system modeling based on multi-agent simulation is characterized in that: The top-level logic of the single enterprise agent is: At the beginning of the month, the enterprise agent will first estimate its excess emission demand Δe(t). If Δe(t) is positive, it means there is a carbon gap, and the enterprise agent must purchase carbon or perform self-purification or wait to be fined by the management department. The enterprise agent then enters the logic of possible buyers. If Δe(t) is negative, it means there is a carbon surplus, and the enterprise agent must decide whether to sell the remaining carbon quota or hold it without selling it. The enterprise agent then enters the logic of possible sellers. After exiting the logic of possible buyers or possible sellers, the relevant variables are updated, and then the next period is entered, repeating the same logic until the simulation ends. The calculation process of the emission demand Δe(t) when the enterprise agent makes production plans at the beginning of each month is as follows: (1) Estimate the market demand of the product d(t)~Uniform(dl,du), where dl represents the lower bound of the market demand and du represents the upper bound of the market demand; (2) Determine the output q(t) to meet market demand and expand production; (3) Estimate total output value TP(t) = q(t) × pd, where pd = unit product price; (4) Convert TP(t) to standard coal equivalent: TP(t) → STC(t) = TP(t) × α1, where α1 is the coefficient for converting total output value to standard coal equivalent; (5) Convert standard coal equivalent STC(t) to carbon dioxide equivalent e(t): STC(t)→e(t)=STC(t)×α2, where α2 is the coefficient for converting standard coal equivalent to carbon dioxide equivalent; (6) Calculate the emission demand Δe(t) = e(t) - EQ(t), where EQ(t) = SQ(t-1) + EQA / 12, SQ(t-1) is the remaining quota in period t-1, and EQA is the annual carbon quota allocated free of charge by the government. It is assumed that the corporate intelligent agent divides the annual carbon quota allocated free of charge by the government equally into each month.

3. According to claim 1, a regional carbon emission trading system modeling based on multi-agent simulation is characterized in that: The calculation of the expected price, self-purification cost and penalty cost includes: ①The first day of each month Assume that the uncorrected expected carbon quota price is PcarbonU, the carbon quota reference price is PcarbonR, the expected price is PcarbonA, the smoothing constant is θ, the value in the model is 0.4, the carbon quota price on the previous day is PcarbonL, the initial value is 64.32, the predicted price of the carbon quota on the previous day is PcarbonFL, the initial value is 109, the influence coefficient of crude oil is Coil, the value in the model is 0.08, the influence coefficient of coal is Ccoal, the value in the model is 0.3, then PcarbonU=θ*PcarbonL+(1-θ)*PcarbonFL PcarbonR=PcarbonU*(1+Ccoal*Fcoal)*(1+Coil*Foil) PcarbonA=PcarbonR*Cd Where Cd is the coefficient of variation. In the model, Cd for possible buyers is uniformly distributed (uniform(0.9, 1.05)), and Cd for possible sellers is uniformly distributed (uniform(0.95, 1.1)). These two distributions are our assumptions and may be adjusted as the research progresses. ②Except the first day of each month Assume that the expected price of the previous day is PcarbonAL, the price fluctuation of carbon quota is R, and the model is uniformly distributed uniform(0, 0.5). This distribution is what we assume, that is, the fluctuation does not exceed 0.5 yuan. As the research goes deeper, it may be adjusted. For potential buyers: PcarbonA = PcarbonAL + R For possible sellers: PcarbonA = PcarbonAL-R The calculation process of self-purification cost is as follows: when the cumulative self-purification emission reduction is m, the self-purification emission reduction cost of self-purification emission reduction nm is: Among them, C0 is the initial value of the unit self-purification emission reduction cost, λ is the difficulty coefficient of self-purification emission reduction, the variable qp represents the cumulative self-purification emission reduction, and Qp is the total target self-purification emission reduction during the planning period; The penalty cost is calculated as follows: (1) When the unfulfilled self-purification target, i.e. the unfulfilled mandatory emission reduction, accounts for less than 10% of the mandatory emission reduction planned by the government, the unit penalty cost shall be three times the average price of the carbon emission trading market quota for the six consecutive months before the current month; (2) When the unfulfilled self-purification target, i.e., the unfulfilled mandatory emission reduction, accounts for more than 10% of the mandatory emission reduction planned by the government, the unit penalty cost shall be four times the average price of the carbon emission trading market quota for the six consecutive months before the current month.

4. According to claim 1, a regional carbon emission trading system modeling based on multi-agent simulation is characterized in that: The logic of the possible buyer is: At the beginning of each month, the expected price is compared with the unit self-cleaning cost and the unit penalty: (1) If the expected price is the lowest, then the logic of purchasing carbon will be adopted; (2) If the unit self-purification cost is the lowest, then the logic of self-purification will be implemented; (3) If the unit fine is the lowest, then the enterprise agent will choose to be fined by the management department; As a possible buyer, the business agent compares the expected price with the unit penalty at the beginning of each month: (1) If the expected price is low, then the logic of choosing to purchase carbon will be entered; (2) If the unit fine is low, then the enterprise agent will choose to be fined by the management department; The logic behind choosing to purchase carbon is as follows: First, the enterprise intelligence will determine whether the market system risk occurs; Secondly, if the market system risk does not occur, the corporate intelligent agent will enter the carbon emission trading market to purchase carbon at the originally expected futures price; if the market system risk occurs, the corporate intelligent agent will adjust the originally expected expected price, and then enter the carbon emission trading market to purchase carbon at the adjusted expected price.

5. The modeling of a regional carbon emission trading system based on multi-agent simulation according to claim 1 is characterized in that: The calculation of the relevant risk is: There are two marginal risks of carbon purchase and one self-purification risk involved, namely: the first risk of carbon purchase - the risk of carbon price fluctuations, the second risk of carbon purchase - the risk of failing to complete the remaining self-purification targets on schedule due to carbon purchases, and the risk of self-purification - the risk that the actual completion time is significantly longer than the planned completion time: (1) Carbon purchase risk assessment ① Consider the first risk of purchasing carbon Defining risk events Carbon price fluctuations are characterized by the actual carbon price significantly exceeding the average price of the past three days; Let p(t) be the acceptable price in period t, ons be the standard deviation of carbon prices in the past three days, p(t) = the moving average carbon price in the past three days (t-1, t-2, t-3), the probability of risk occurrence of p3(t)<p(t) is 100%, p(t)≤p3(t)<p(t)+σ M3 The probability of risk occurrence is 85%, p(t)+σ M3 ≤p3(t)<p(t)+2σ M3 The probability of risk occurrence is 60%, p(t)+2σ M3 ≤p3(t)<p(t)+3σ M3 The probability of risk occurrence is ≤15%, p(t)+3σ M3 ≤p3(t) The probability of risk occurrence is ≤5%; ② Consider the second risk of purchasing carbon Defining risk events Because: due to carbon purchase, the remaining self-purification target cannot be completed on schedule. Let the remaining time be rt), t g The time required to complete the remaining self-purification target = the remaining self-purification target amount / C, where C is a coefficient, σ g The standard deviation of the time required to complete the remaining self-purification target = CVSP*t g , (Tt<t g The probability of risk occurrence is 100%, t g ≤(Tt)<t g +σ g The probability of risk occurrence is 85%, t g +σ g ≤(Tt)<t g +2σ g The probability of risk occurrence is ≤ 60%, t g +2σ g ≤(Tt)<t g +3σ g The probability of risk occurrence is ≤15%, t g +3σ g ≤(Tt) The probability of risk occurrence is ≤5%; ③ Consider two risks of carbon purchase at the same time and When considering the combined risk of carbon purchase, the "average method" is used according to the probability of occurrence of the above two types of risks. The calculation formula is p ij =p ji =a i +b j / 2, a i for The upper limit of risk probability at different levels, b is The probability upper limits of different risk levels are calculated to be the possibility of the occurrence of joint risks, which are organized into the following numerical form and semantic form. The numerical form can be simulated on a computer simulation platform, and the semantic form provides the logic of decision-making. When the higher The corresponding upper limit of risk probability is 0.60, and the corresponding upper limit of risk probability is 0.

85. According to the above formula, the joint risk probability is 0.

725. This probability value is greater than 0.60 and less than 0.85, so the joint risk level is judged to be "high"; (2) Self-purification risk assessment Consider the risks of self-purification Defining risk events The actual completion time is significantly longer than the planned completion time; Let D,(q) be the mean of the equivalent duration of the self-cleaning project = self-cleaning project equivalent / C, C is a coefficient, the standard deviation of the equivalent duration of oa = CVSP*D(q), D(q) is the actual completion time, the probability of the risk of D0(q)<D(q) is = 100%, D(q)≤D0(q)<D(q)+σ d The probability of risk occurrence is ≤85%, D(q)+σ d ≤D0(q)<D(q)+2σ d The probability of risk occurrence is ≤60%, D(q)+2σ d ≤D0(q)<D(q)+3σ d The probability of risk occurrence is ≤15%, D(q)+3σ d The probability of occurrence of ≤D0(q) risk is ≤5%.

6. The modeling of a regional carbon emission trading system based on multi-agent simulation according to claim 1 is characterized in that: The calculation process of the total amount control mechanism and quota allocation mechanism is as follows: (1) Assuming that the actual regional emissions in the base year are Q0, the budget quota correction coefficient is d%, and the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, then the government's budget quota for this year can be obtained as QT = Q0*rT, where r T =r T-1 *(1-d%)*(1-R%)*(1+e%), r0=1; and the government's mandatory emission reduction target for this year, that is, the self-net emission reduction target, is: G T =Q T-1 *(1-d%)*R%*(1+e%), which does not include the self-net emission reduction targets of new entrants; (2) Assume that the unit self-purification emission reduction costs of each enterprise in the previous year are C1, C2, ... C n The output value scale is V1, V2...V n , the standard coal quantity per 10,000 yuan of GDP is E1, E2...E n , then, we can get the self-purification emission reduction targets of each enterprise this year: G i =(V i E i / C i ) / (V1E1 / C1+V2E2 / C2+…+V n E n / C n )*G T ; (3) Assuming that the reserved auction ratio of the region is a%, the reserved government reserve ratio is b%, and the reserved new entrant ratio is c%, then the free quota for each enterprise this year is: Q i =(V i E i / C i ) / (V1E1 / C1+V2E2 / C2+…+V n E n / C n )*Q T *(1-a%-b%-c%).

7. The modeling of a regional carbon emission trading system based on multi-agent simulation according to claim 1 is characterized in that: The calculation process of the initial quota amount of the new entrant reserve quota is as follows: (1) Assuming that the actual regional emissions in the base year are Q0, the budget quota correction coefficient is d%, and the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, then the government's budget quota for this year can be obtained as Q T =Q0*r T , where r T =r T-1 *(1-d%)*(1-R%)*(1+e%), r0=1; (2) Assuming that the proportion of the region's reserved new entrants is c%, then the reserved new entrants reserve for this year can be obtained as: QR = QT * c%; The calculation process for updating the free quota for new entrants, the self-purification emission reduction target and the reserved reserve quota for new entrants is as follows: Assuming that the regional budget quota correction coefficient is d%, the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, the emissions of the newly entered enterprises in the same period last year are Q T-1 , the reserved auction ratio of the region is a%, the reserved government reserve ratio is b%, and the reserved new entrant ratio is c%, then the free quota for new entrants is Q T =Q T-1 *(1-d%)*(1-R%)*(1+e%)*(1-a%-b%-c%), we can also get the mandatory emission reduction for new entrants: G T =Q T-1 *(1-d%)*R%*(1+e%), at this time, the reserved quota for new entrants is Q R Update according to the following formula: Q R →Q R -Q T .

8. The modeling of a regional carbon emission trading system based on multi-agent simulation according to claim 1 is characterized in that: The calculation process of the initial quota of the government carbon reserve quota is as follows:

1. Assuming that the actual regional emissions in the base year are Q0, the budget quota correction coefficient is d%, the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, then the government's budget quota for this year can be obtained as Q T =Q0*r T , where r T =r T-1 *(1-d%)*(1-R%)*(1+e%), r0=1; 2. Assuming that the proportion of reserved government reserves in the region is b%, then the reserved government reserves this year can be obtained as: Q b =Q T * b%; The detailed government carbon reserve mechanism is as follows: (1) Repurchase quota 1) Repurchase conditions A repurchase is taken when one of the following conditions is met: ① When the market carbon price is lower than the given threshold P0 and is expected to continue to fall (the carbon price has fallen for three consecutive days and is expected to fall again today), and the supply exceeds the demand, a repurchase is adopted; P0=P min -K1*σ, where P min represents the lowest carbon price in the previous year, σ represents the standard deviation of the carbon price in the previous year, K1=(CV Q +CV P ) / (1+CV Q +CV P ) represents the price-related buyback trigger correction factor, where CV Q Represents the coefficient of variation of the previous year's trading volume, CV P represents the coefficient of variation of carbon prices in the previous year; ② The amount of carbon quota in the carbon quota reserve is less than Q min = K*half of the highest daily trading volume in the previous year, repurchase; 2) Repurchase quantity Q1 Considering that supply is greater than demand at this time, the repurchase quantity should be related to the changes in supply and demand. The repurchase quantity should make supply and demand tend to a balanced state. Let the supply quantity at this time be Q s1 , the demand is Q d1 , let ΔQ1=Q s1 -Q d1 , K2 is the adjustment coefficient; Repurchase quantity Q1 = K2*ΔQ1, where K2 = (CV q +CV p ) / (1+CV q +CV p ) represents the repurchase correction coefficient, where CV q The coefficient of variation of the trading volume in the previous three days, CV p represents the coefficient of variation of carbon prices in the previous three days; 3) Buyback price ① The repurchase price for stabilizing the market is P1* P1*=Mt-1+θ*σ 3(t-1) , where M t-1 represents the market carbon price of the previous day, σ 3(t-1) is the standard deviation of the market carbon price in the previous three days, θ=(CV q +CV p ) / (1+CV q +CV p ) represents the repurchase price correction factor, where CV q The coefficient of variation of the trading volume in the previous three days, CV p represents the coefficient of variation of the market carbon price in the previous three days; ②The repurchase price for filling the carbon reserve is P2* At this point, the repurchase department is similar to a market participant, and the repurchase price P2* = C* reference price; (2) Release of reserves 1) Conditions for placement The market carbon price is higher than the given threshold P M , and it is expected to continue to rise and the market demand is greater than the supply at this time, so release is adopted; P M =P max +K3*σ, where P max represents the highest carbon price in the previous year, σ represents the standard deviation of the carbon price in the previous year, K3=(CV Q +CV P ) / (1+CV Q +CV P ) represents the price-related trigger correction coefficient, where CV Q Represents the coefficient of variation of the previous year's trading volume, CV P represents the coefficient of variation of carbon prices in the previous year; 2) Quantity of delivery Q2 Considering that the market demand is greater than the supply at this time, the quantity of supply should be related to the changes in supply and demand, and the quantity of supply should make the supply and demand tend to a balanced state. Suppose the supply at this time is Q s2 , the demand is Q d2 , let ΔQ2=Q s2 -Q d2 , K4 is the adjustment coefficient; The quantity of delivery Q2 = K4*ΔQ2, where K4 = (CV q +CV p ) / (1+CV q +CV p ) represents the correction factor of the delivery amount, where CV q The coefficient of variation of the trading volume in the previous three days, CV p represents the coefficient of variation of carbon prices in the previous three days; 3) The price of the advertisement The repurchase price for market stabilization is P3* P3*=M t-1 -θ*σ 3(t-1) , where M t-1 represents the market carbon price of the previous day, σ 3(t-1) is the standard deviation of the market carbon price in the previous three days, θ=(CV q +CV p ) / (1+CV q +CV p ) represents the price correction factor, where CV q The coefficient of variation of the trading volume in the previous three days, CV p Represents the coefficient of variation of the market carbon price in the previous three days.

9. The modeling of a regional carbon emission trading system based on multi-agent simulation according to claim 1 is characterized in that: The calculation process of the initial quota amount of the auction mechanism is as follows:

1. Assuming that the actual regional emissions in the base year are Q0, the budget quota correction coefficient is d%, the government's planned emission reduction coefficient is R%, and the expected economic growth this year is e%, then the government's budget quota for this year can be obtained as Q T =Q0*r T , where r T =r T-1 *(1-d%)*(1-R%)*(1+e%), r0=1; 2. Assuming that the proportion of the reserved auction quota in the region is a%, then the reserved auction quota for this year can be obtained as: Q A =Q T *a%.

10. The modeling of a regional carbon emission trading system based on multi-agent simulation according to claim 1 is characterized in that: The penalty mechanism is: When the unfulfilled self-purification target, i.e. the unfulfilled mandatory emission reduction, accounts for less than 10% of the mandatory emission reduction planned by the government, the unit penalty cost shall be three times the average price of the carbon emission trading market quota for the six consecutive months before the current month; When the unfulfilled self-purification target, that is, the unfulfilled mandatory emission reduction, accounts for more than 10% of the government's planned mandatory emission reduction, the unit penalty cost shall be four times the average price of the carbon emission trading market quota for the six consecutive months before that month.

Images