Multi-time scale optimization scheduling method for multi-agent integrated energy system

Abstract

The invention discloses multi-time-scale optimal scheduling of a multi-agent integrated energy system. The multi-time-scale optimal scheduling comprises the following steps: constructing a multi-agent scheduling model and a A day-ahead optimization stage: inputting day-ahead wind, light and load prediction data, establishing a multi-subject integrated energy system day-ahead scheduling optimization model, solving to obtain a unit start-stop state, a power purchasing and selling state of each subject and a load calling amount, and issuing a day-ahead scheduling instruction to the intra-day optimization stage; an intra-day optimization stage: updating intra-day wind and light output and load change, establishing an intra-day scheduling optimization model of the multi-subject integrated energy system, solving to obtain a load calling amount, and issuing an intra-day scheduling instruction to the real-time optimization stage; and a real-time optimization stage: updating real-time data of wind and light output and load electricity consumption, establishing a real-time scheduling optimization model of the multi-subject integrated energy system, and solving real-time output of each subject of the multi-subject integrated energy system according to real-time scheduling. According to the method, electric energy sharing among subjects can be realized, the rapid adjustment capability of DR resources is better utilized, and the overall economic benefit of the system is improved.

Description

Technical Field

[0001] The present invention relates to the field of integrated energy optimization scheduling, and in particular to a multi-agent integrated energy system multi-time scale optimization scheduling method. Background Art

[0002] With the development of energy Internet technology, integrated energy systems with multiple energy networks coupled have emerged. Multiple integrated energy systems are connected to the same regional distribution network to form an integrated energy system group, and the systems interact with each other through energy interconnection lines. At this stage, the multi-subject and multi-park joint model is developing rapidly. With the power market as the core, it is particularly important to achieve multi-market integration and coordinate the interests of various subjects.

[0003] At present, most of the research on multi-park integrated energy systems achieves the optimal economy of multiple subjects through game mechanisms and energy pricing strategies. Or, on the basis of considering demand response (DR), the economic benefits of multiple parks are improved with the goal of maximizing the consumption of new energy and maximizing benefits. However, the above studies all ignore the multi-time scale characteristics of demand response (DR) resources. In fact, the time characteristics and rapid adjustment capabilities of demand response (DR) resources will affect the accuracy of the forecast data, thereby affecting the final scheduling cost and alliance benefits.

[0004] Existing research on the optimization and dispatching of multi-agent integrated energy systems mostly focuses on formulating day-ahead dispatching strategies and achieving optimal economy through game mechanisms. However, this strategy does not consider the impact of uncertain factors such as wind power, photovoltaics, and demand response on the system. In fact, the random volatility of renewable energy output and the uncertainty of demand response will affect the accuracy of the day-ahead dispatching plan formulated by the power grid. Moreover, most studies ignore the joint consideration of both random volatility of sources and loads and demand response, especially the lack of joint research on their different characteristics on multiple time scales. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a multi-time scale optimization scheduling method for a multi-agent integrated energy system, which can realize the sharing of electricity among the subjects, better utilize the rapid adjustment capability of demand response DR resources, and improve the overall economic benefits of the system.

[0006] The technical solution adopted by the present invention is:

[0007] The multi-agent integrated energy system multi-time scale optimization scheduling method includes the following steps:

[0008] Step 1: Construct a multi-agent scheduling model and each agent's profit model;

[0009] Step 2: Day-ahead optimization stage: input the day-ahead wind and solar power and load forecast data, establish a day-ahead dispatch optimization model for a multi-agent integrated energy system, solve the start-stop status of the unit, the power purchase and sales status of each entity, and the load call amount, and issue the day-ahead dispatch instruction to the intraday optimization stage;

[0010] Step 3: Intraday optimization stage: Update the wind and solar power output and load changes within the day, establish a multi-agent integrated energy system intraday scheduling optimization model, solve the load call amount, and issue the intraday scheduling instructions to the real-time optimization stage;

[0011] Step 4: Real-time optimization stage: Update the real-time data of wind and solar power output and load power consumption, establish a real-time scheduling optimization model for the multi-agent integrated energy system, and solve the real-time output of each subject in the multi-agent integrated energy system based on real-time scheduling.

[0012] In step 1, the multi-agent scheduling model is expressed as follows:

[0013]

[0014]

[0015] Where: C pro represents the multi-park benefits, C cost represents the multi-park operation and maintenance cost; N represents the number of parks, which is 3 in the present invention; T is 24h, t represents the time, and i and j both represent the park; It represents the power purchased by park i from the distribution network at time t; and They represent the power purchased and sold by park i to park j at time t respectively; They represent the discharging and charging power of the energy storage power station to park i respectively; The power of gas purchased from the gas network by park i at that moment; is the unit operation and maintenance power of the equipment contained in park i; represents the wind power output of park i at time t; It represents the different types of load demands of park i at time t; represents the purchase price of wind power at time t; is the unit price of electricity sold by the distribution network at time t; Indicates the interactive electricity price of the park; Respectively represent the electricity sales and purchase prices of energy storage power stations; is the natural gas price at time t; is the unit power operation and maintenance cost of the equipment contained in park i; It represents the energy selling price corresponding to the load demand at time t.

[0016] In step 1, the energy supply strategy of each entity is analyzed, and the profit model of each entity is established;

[0017] The energy supply strategies of each entity are that there is two-way power interaction between the energy storage power station and each park, and there is two-way power interaction between the parks. The distribution network and wind farm transmit power to each park, and the wind farm transmits power to the energy storage power station and the power grid.

[0018] The main subjects of the revenue model include:

[0019] 1) Energy storage power station revenue model:

[0020]

[0021] Where: C ess represents the profit of energy storage power station; P t grid,ch , P t wind,ch They represent the power purchased by the energy storage power station from the distribution network and the wind farm at time t respectively; Represents the maintenance cost per unit power of the energy storage power station; They represent the discharging and charging power of the energy storage power station to park i respectively; Indicates the price of electricity purchased and sold by the energy storage power station; is the electricity price of the distribution network at time t; It represents the purchase price of wind power corresponding to time t.

[0022] 2) Wind farm revenue model:

[0023]

[0024] Where: C wind represents the profit of wind farm; P t wind,grid represents the power purchased by the power grid from the wind farm at time t, represents the maintenance cost per unit power of the wind farm; Represents the wind power output of park i at time t.

[0025] 3) Grid power purchase model:

[0026]

[0027] Where: C grid It represents the cost of each entity purchasing electricity from the grid.

[0028] In step 2, demand response (DR) is divided into incentive-based demand response (IDR) and price-based demand response (PDR). Price-based demand response PDR is usually determined in the day-ahead stage; incentive-based demand response IDR is divided into: PDR according to the length of response time. IDRA , P IDRB , P IDRC , P IDRD ;P IDRA , P IDRB , P IDRC , P IDRD represents the incentive demand response corresponding to different response times, P IDRA Confirm 24h in advance; IDRB The time is determined to be 15min to 2h; P IDRC The time is determined to be 5min to 15min; IDRD Determined in real time.

[0029] In step 2, the objective function of the multi-agent integrated energy system day-ahead dispatch optimization model is expressed as:

[0030] The three parks and the external power purchasing system including the distribution network, energy storage power station and wind farm form an alliance. The goal of the alliance is to maximize the overall benefits of the three parks, energy storage power station and wind farm. The objective function is as follows:

[0031] maxC=(C pro +C ess +C wind )-(C cost +C grid +C load )

[0032]

[0033] Where: C represents the total revenue of the alliance; C load represents the load call cost; k IDRA , k IDRB Respectively represent P IDRA and P IDRB Cost coefficient, △P IDRA,i,t |, △P IDRB,i,t |respectively represent the corresponding IDR call volume at time t.

[0034] The constraints of the above objective function include:

[0035] 1) Electricity, heat and cold balance constraints:

[0036]

[0037] Where: Indicates the power consumption of the air conditioner; Indicates photovoltaic output; the power output of the CHP unit in Park 2 and the CCHP unit in Park 3 of the present invention are both expressed as express; Indicates the cooling output of CCHP unit; Indicates the cooling output of the air conditioner; Represents the cooling load of park i.

[0038] 2) Inter-park power transmission constraints:

[0039]

[0040] Where: P i grid,buy,max represents the upper limit of electricity sold by the distribution network to park i at time t; They represent the upper limit of electricity purchase and electricity sales from park i to park j at time t respectively; The state variable representing the distribution network's electricity sales behavior to park i at time t, The state variable representing the electricity purchase behavior of park i from park j at time t, The state variable representing the electricity sales behavior of park i to park j at time t.

[0041] 3) Constraints on the interaction between each entity and the park’s power:

[0042]

[0043] Where: They represent the states of charging and discharging behavior of the energy storage power station to park i respectively; Indicates the status of wind farm selling electricity to park i; The state variable representing the distribution network's electricity sales behavior to park i at time t.

[0044] 4) Coupling equipment output constraints:

[0045]

[0046] Where: P represents the output power of the equipment in park i at time t, i B,max P represents the upper limit of the output power of the equipment in park i at time t. i B,min It represents the lower limit of the output power of the equipment in park i at time t. The state variable representing the output of equipment i in the park at time t, The state variable representing the output of equipment i in the park at the last moment t, They respectively represent the start and stop behavior state variables of equipment i in the park at time t.

[0047] 5) Energy storage power station operation constraints:

[0048]

[0049] Where: S t represents the charge state of the energy storage power station at time t, S max Represents the upper limit of the energy storage power station, S min Indicates the lower limit of energy storage power station power. Represents the state variable of the energy storage power station charging behavior, Represents the state variable of the energy storage power station discharge behavior, S t-1 Indicates the charge state of the energy storage power station at the last moment t, S t＝1 It represents the state of charge of the energy storage power station when t is 1, S t＝1 Indicates that t is the charge state of the energy storage power station at time T.

[0050] 6) DR resource constraints:

[0051] The day-ahead optimization phase needs to consider PDR and P IDRA and P IDRB Constraints:

[0052]

[0053] Where: They represent the upper and lower limits of the PDR call volume in park i respectively; They represent the increase in IDR of category A and B in park i respectively; They represent the reduction of Class A and Class B IDR in park i. PDR,i,t represents the call volume of PDR in park i; It represents the upper limit of the increase of Class A IDR in park i; It represents the upper limit of the reduction of Class A IDR in park i; It represents the upper limit of the increase of Class B IDR in park i; It represents the upper limit of the reduction of Class B IDR in park i.

[0054] The multi-agent integrated energy system day-ahead dispatch optimization model is solved and the following is obtained:

[0055] a. Start and stop status of the unit; b. Power purchase and sales status of each entity; c. PDR call volume, P IDRA Call volume; the obtained day-ahead scheduling instructions are issued to the intraday optimization stage.

[0056] In step 3, the intraday optimization scheduling is to feed back the current measured power generation of wind and solar power and the measured power consumption of load to the intraday optimization stage, and obtain the optimal solution by predicting the source and load output with a time scale of 15 minutes within the next 4 hours;

[0057] a. Objective function:

[0058] The goal of intraday optimization scheduling is the same as that of day-ahead scheduling, which is to maximize the profit of the alliance. Compared with the day-ahead scheduling optimization model, the only thing that changes in the intraday scheduling optimization model is C. load , P IDRA The call volume has been determined in the day-ahead scheduling phase, so C load Change to P IDRB Call cost and P IDRC The sum of the call costs; the objective function is as follows:

[0059] maxC=(C pro +C ess +C wind )-(C cost +C grid +C load )

[0060]

[0061] Where: k IDRC represents the cost coefficient of Class C IDR, △|P IDRC,i,t | represents the number of calls of class C IDR at time t.

[0062] Constraints:

[0063] The constraints of the intraday optimization phase are basically the same as those of the day-ahead phase. IDRC The additional constraints are as follows:

[0064]

[0065] Where: represents the increase in Class C IDR in park i, It represents the reduction of Class C IDR in park i; It indicates the upper limit of the increase of Class C IDR in park i; It indicates the upper limit of the reduction of Class C IDR in park i.

[0066] The optimization phase has determined the start and stop status of the units, the power purchase and sales status of each entity, the PDR call volume, and the P IDRA The call volume, the intraday optimization stage determines P based on this IDRB The call volume and the intraday scheduling instructions are issued to the real-time optimization stage.

[0067] In step 4, the real-time optimization stage is to find the optimal solution by predicting the source load output with a future time scale of 5 minutes;

[0068] Objective function:

[0069] maxC=(C pro +C ess +C wind )-(C cost +C grid +C load )

[0070]

[0071] Where: k IDRD represents the cost coefficient of Class D IDR, △|P IDRD,i,t | represents the call volume of class D IDR at time t.

[0072] Constraints:

[0073] The constraints in the real-time optimization phase are basically the same as those in the intraday optimization phase. IDRD The additional constraints are as follows:

[0074]

[0075] Where: It represents the increase of Class D IDR in park i; It represents the increase of Class C IDR in park i; It represents the upper limit of the increase of Class D IDR in park i; It represents the upper limit of the reduction of Class D IDR in park i.

[0076] By optimizing the real-time dispatch optimization model of the multi-agent integrated energy system, the start-stop status and output of the unit can be obtained; IDRC Call volume and P IDRD Call volume.

[0077] The present invention provides a multi-agent integrated energy system multi-time scale optimization scheduling method, the technical effects are as follows:

[0078] 1) In step 1 of the present invention, the multi-agent scheduling model constructed in step 1 comprehensively considers multiple revenue items and cost items. The revenue items include multi-park revenue, energy storage power station revenue and wind farm revenue. The cost items include the operation and maintenance costs of multiple parks, the cost of each entity purchasing electricity from the power grid, and the call volume cost of IDR loads. This comprehensiveness helps to more comprehensively reflect the overall profit of the multi-agent integrated energy system. There is power interaction between the entities. The distribution network, wind farm and energy storage power station form an alliance with the park to optimize the output of each unit and improve the overall economic benefits of the alliance.

[0079] 2) In steps 2 to 4 of the present invention, a day-ahead scheduling plan is formulated according to the multi-time scale characteristics of the demand response resources, and the results are regulated in the intraday and real-time stages. Multi-time scale optimization scheduling gives full play to the rapidity of DR resource regulation and improves the accuracy of multi-agent integrated energy system data. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 It is a schematic diagram of energy flow of the multi-agent integrated energy system of the present invention.

[0081] Figure 2 This is a flow chart for solving the multi-time scale optimization scheduling model proposed in the present invention.

[0082] Figure 3 This is the predicted and measured curve of the power load in Park 1.

[0083] Figure 4 This is the predicted and measured curve of the power load in Park 2.

[0084] Figure 5 This is the predicted and measured curve of the electrical load in Park 3.

[0085] Figure 5 This is the predicted and measured curve of the electrical load in Park 3.

[0086] Figure 6 This is a wind and solar prediction curve. DETAILED DESCRIPTION

[0087] Figure 1 Schematic diagram of energy flow in a multi-agent integrated energy system. The present invention constructs three parks, and there is electric energy interaction between the parks. The distribution network, wind farm and energy storage power station form an alliance with the park. The three parks are a photovoltaic park, a combined heat and power park, and a combined heat and power park. Park 1 is a photovoltaic park, representing the electricity consumption of urban and suburban residents. It is a surplus power park and supplies power to Parks 2 and 3. Park 2 is a combined heat and power park, representing the electricity consumption of industrial users. Park 3 is a combined heat and power park, representing the electricity consumption of urban residents. There is a two-way electric energy interaction between the energy storage power station and each park, and between the parks. The distribution network and wind farm transmit electricity to each park, and the wind farm transmits electricity to the energy storage power station and the power grid.

[0088] Figure 2 This is the model solution flow chart of the present invention. First, initialization is performed and the day-ahead wind and load forecast data are input. The day-ahead wind and load forecast is as follows: Figure 6 The day-ahead load forecasts for the three parks are shown in Figure 3 , Figure 4 , Figure 5Secondly, a multi-agent integrated energy system day-ahead dispatch model is established to solve the day-ahead dispatch plan and some specific quantities, and the day-ahead dispatch instructions are issued to the intraday optimization stage.

[0089] Then, a multi-agent integrated energy system intraday dispatch model is established, the wind and solar power output and load changes are updated, the intraday dispatch plan is solved and issued to the real-time stage. Then a multi-agent integrated energy system real-time dispatch optimization model is established, and the real-time output of each subject in the multi-agent integrated energy system is solved according to the real-time dispatch.

[0090] And solve the real-time output of each device in the system; finally, determine whether the real-time optimization period reaches the maximum of the whole day. If it reaches it, output the system scheduling plan; if not, the time window continues to roll forward.

[0091] Figure 3 , Figure 4 , Figure 5 The following are the load forecast and measured curves of the three parks. The corresponding forecast errors of the load in the three situations of day-ahead, intraday and real-time are 3%, 1% and 0.5% respectively. Three strategies are set to compare the costs of each park and the total benefits of the alliance under different strategies.

[0092] Strategy 1: Each park operates independently, and all scheduling plans are day-ahead scheduling plans;

[0093] Strategy 2: Multiple agents jointly operate, and all scheduling plans are day-ahead scheduling plans;

[0094] Strategy three: Multiple agents work together to adopt a multi-time scale scheduling plan.

[0095] Table 1 Costs of each park and total alliance revenue under different strategies

[0096]

[0097] From Table 1 above, we know that by comparing Strategy 1 and Strategy 2, the total cost of the park is reduced by 7124 yuan and the total income of the alliance is increased by 169 yuan compared with the independent operation of multi-agent joint operation. This shows that the joint operation of multiple subjects promotes the sharing of electricity and improves the overall economic benefits of the alliance. Strategy 3 adopts a multi-time scale scheduling plan based on the joint operation of multiple subjects, which effectively reduces the total cost of the park and increases the total income of the alliance. Compared with Strategy 2 which does not adopt a multi-time scale scheduling plan, the total cost of the park decreases by 4.7% and the total income of the alliance increases by 3.9%. The research results show that the multi-time scale optimization scheduling method of the multi-agent integrated energy system proposed in this invention is feasible.

Claims

1. Multi-agent integrated energy system multi-time scale optimization scheduling method, characterized by The following steps are involved: Step 1: Construct a multi-agent scheduling model and each agent's profit model; Step 2: Day-ahead optimization stage: input the day-ahead wind and solar power and load forecast data, establish a day-ahead dispatch optimization model for a multi-agent integrated energy system, solve the start-stop status of the unit, the power purchase and sales status of each entity, and the load call amount, and issue the day-ahead dispatch instruction to the intraday optimization stage; Step 3: Intraday optimization stage: Update the wind and solar power output and load changes within the day, establish a multi-agent integrated energy system intraday scheduling optimization model, solve the load call amount, and issue the intraday scheduling instructions to the real-time optimization stage; Step 4: Real-time optimization stage: Update the real-time data of wind and solar power output and load power consumption, establish a real-time scheduling optimization model for the multi-agent integrated energy system, and solve the real-time output of each subject in the multi-agent integrated energy system based on real-time scheduling.

2. The multi-agent integrated energy system multi-time scale optimization scheduling method according to claim 1 is characterized by: In step 1, the multi-agent scheduling model is expressed as follows: Where: C pro represents the multi-park benefits, C cost represents the multi-park operation and maintenance cost; N represents the number of parks; T is 24 hours, t represents the time, and i and j both represent the park; It represents the power purchased by park i from the distribution network at time t; and They represent the power purchased and sold by park i to park j at time t respectively; They represent the discharging and charging power of the energy storage power station to park i respectively; The power of gas purchased from the gas network by park i at that moment; is the unit operation and maintenance power of the equipment contained in park i; represents the wind power output of park i at time t; It represents the different types of load demands of park i at time t; represents the purchase price of wind power at time t; is the unit price of electricity sold by the distribution network at time t; Indicates the interactive electricity price of the park; Respectively represent the electricity sales and purchase prices of energy storage power stations; is the natural gas price at time t; is the unit power operation and maintenance cost of the equipment contained in park i; It represents the energy selling price corresponding to the load demand at time t.

3. The multi-time scale optimization scheduling method for a multi-agent integrated energy system according to claim 2 is characterized in that: In step 1, the energy supply strategy of each entity is analyzed, and the profit model of each entity is established; The energy supply strategies of each entity include two-way power exchange between energy storage power stations and various parks, two-way power exchange between parks, distribution networks and wind farms transmitting power to various parks, and wind farms transmitting power to energy storage power stations and power grids; The main subjects of the revenue model include: 1) Energy storage power station revenue model: Where: C ess represents the profit of energy storage power station; P t grid,ch , P t wind,ch They represent the power purchased by the energy storage power station from the distribution network and the wind farm at time t respectively; Represents the maintenance cost per unit power of the energy storage power station; They represent the discharging and charging power of the energy storage power station to park i respectively; Indicates the price of electricity purchased and sold by the energy storage power station; is the electricity price of the distribution network at time t; represents the purchase price of wind power at time t; 2) Wind farm revenue model: Where: C wind represents the profit of wind farm; P t wind,grid represents the power purchased by the power grid from the wind farm at time t, represents the maintenance cost per unit power of the wind farm; represents the wind power output of park i at time t; 3) Grid power purchase model: Where: C grid It represents the cost of each entity purchasing electricity from the grid.

4. The multi-agent integrated energy system multi-time scale optimization scheduling method according to claim 1 is characterized by: In step 2, demand response (DR) is divided into incentive-based demand response (IDR) and price-based demand response (PDR). The price-based demand response PDR is determined in the day-ahead stage; the incentive-based demand response IDR is divided into: PDR according to the length of the response time. IDRA , P IDRB , P IDRC , P IDRD ;P IDRA , P IDRB , P IDRC , P IDRD represents the incentive demand response corresponding to different response times, P IDRA Confirm 24h in advance; IDRB The time is determined to be 15min to 2h; P IDRC The time is determined to be 5min to 15min; IDRD Determined in real time.

5. The multi-time scale optimization scheduling method for a multi-agent integrated energy system according to claim 4 is characterized in that: In step 2, the objective function of the multi-agent integrated energy system day-ahead dispatch optimization model is expressed as: The three parks and the external power purchasing system including the distribution network, energy storage power station and wind farm form an alliance; the goal of the alliance is to maximize the overall benefits of the three parks, energy storage power station and wind farm. The objective function is as follows: maxC=(C pro +C ess +C wind )-(C cost +C grid +C load ) Where: C represents the total revenue of the alliance; C load represents the load call cost; k IDRA , k IDRB Respectively represent P IDRA and P IDRB The cost coefficient, △|P IDRA,i,t |、△|P IDRB,i,t |respectively represent the corresponding IDR call volume at time t.

6. The multi-agent integrated energy system multi-time scale optimization scheduling method according to claim 5 is characterized by: The constraints of the objective function include: 1) Electricity, heat and cold balance constraints: Where: Indicates the power consumption of the air conditioner; Indicates photovoltaic output; the CHP units in Park 2 and the CCHP units in Park 3 are both used for power output express; Indicates the cooling output of CCHP unit; Indicates the cooling output of the air conditioner; represents the cooling load of park i; 2) Inter-park power transmission constraints: Where: P i grid,buy,max represents the upper limit of electricity sold by the distribution network to park i at time t; They represent the upper limit of electricity purchase and electricity sales from park i to park j at time t respectively; The state variable representing the distribution network's electricity sales behavior to park i at time t, The state variable representing the electricity purchase behavior of park i from park j at time t, The state variable representing the electricity sales behavior of park i to park j at time t; 3) Constraints on the interaction between each entity and the park’s power: Where: They represent the states of charging and discharging behavior of the energy storage power station to park i respectively; Indicates the status of wind farm selling electricity to park i; The state variable representing the distribution network's electricity sales behavior to park i at time t; 4) Coupling equipment output constraints: Where: represents the output power of the equipment in park i at time t, P i B,max P represents the upper limit of the output power of the equipment in park i at time t. i B,min It represents the lower limit of the output power of the equipment in park i at time t. The state variable representing the output of equipment i in the park at time t, The state variable representing the output of equipment i in the park at the last moment t, They represent the start and stop behavior state variables of equipment i in the park at time t respectively; 5) Energy storage power station operation constraints: Where: S t represents the charge state of the energy storage power station at time t, S max Represents the upper limit of the energy storage power station, S min Indicates the lower limit of energy storage power station power. Represents the charging behavior state variable of the energy storage power station, Represents the state variable of the energy storage power station discharge behavior, S t-1 Indicates the charge state of the energy storage power station at the last moment t, S t＝1 Indicates the state of charge of the energy storage power station when t is 1, S t＝1 Indicates that t is the charge state of the energy storage power station at time T; 6) DR resource constraints: The day-ahead optimization phase considers PDR and P IDRA and P IDRB Constraints: Where: They represent the upper and lower limits of the PDR call volume in park i respectively; They represent the increase in IDR of category A and B in park i respectively; They represent the reduction of IDR of category A and B in park i respectively; P PDR,i,t represents the call volume of PDR in park i; It represents the upper limit of the increase of Class A IDR in park i; It represents the upper limit of the reduction of Class A IDR in park i; It represents the upper limit of the increase of Class B IDR in park i; It represents the upper limit of the reduction of Class B IDR in park i; The multi-agent integrated energy system day-ahead dispatch optimization model is solved to obtain: a. unit start-up and shutdown status; b. power purchase and sales status of each entity; c. PDR call volume, P IDRA Call volume; the obtained day-ahead scheduling instructions are issued to the intraday optimization stage.

7. The multi-agent integrated energy system multi-time scale optimization scheduling method according to claim 4 is characterized by: In step 3, the intraday optimization scheduling is to feed back the current measured power generation of wind and solar power and the measured power consumption of load to the intraday optimization stage, and obtain the optimal solution by predicting the source and load output with a time scale of 15 minutes within the next 4 hours; a. Objective function: The goal of intraday optimization scheduling is the same as that of day-ahead scheduling, which is to maximize the profit of the alliance. Compared with the day-ahead scheduling optimization model, the only thing that changes in the intraday scheduling optimization model is C. load , P IDRA The call volume has been determined in the day-ahead scheduling phase, so C load Change to P IDRB Call cost and P IDRC The sum of the call costs; the objective function is as follows: maxC=(C pro +C ess +C wind )-(C cost +C grid +C load ) Where: k IDRC represents the cost coefficient of Class C IDR, △|P IDRC,i,t | represents the number of calls of class C IDR at time t.

8. The multi-agent integrated energy system multi-time scale optimization scheduling method according to claim 7 is characterized by: The constraints of the objective function are as follows: The constraints in the intraday optimization phase are increased by P IDRC The additional constraints are as follows: Where: represents the increase in Class C IDR in park i, It represents the reduction of Class C IDR in park i; It indicates the upper limit of the increase of Class C IDR in park i; It represents the upper limit of the reduction of Class C IDR in park i; The optimization phase has determined the start and stop status of the units, the power purchase and sales status of each entity, the PDR call volume, and the P IDRA The call volume, the intraday optimization stage determines P based on this IDRB The call volume and the intraday scheduling instructions are issued to the real-time optimization stage.

9. The multi-agent integrated energy system multi-time scale optimization scheduling method according to claim 4 is characterized by: In step 4, the real-time optimization stage is to find the optimal solution by predicting the source load output with a future time scale of 5 minutes; Objective function: maxC=(C pro +C ess +C wind )-(C cost +C grid +C load ) Where: k IDRD represents the cost coefficient of Class D IDR, △|P IDRD,i,t | represents the call volume of class D IDR at time t.

10. The multi-agent integrated energy system multi-time scale optimization scheduling method according to claim 9 is characterized in that: The constraints of the objective function are as follows: The constraints in the real-time optimization phase are similar to those in the intraday optimization phase. IDRD The additional constraints are as follows: Where: It represents the increase of Class D IDR in park i; It represents the increase of Class C IDR in park i; It represents the upper limit of the increase of Class D IDR in park i; It represents the upper limit of the reduction of Class D IDR in park i; By optimizing the real-time dispatch optimization model of the multi-agent integrated energy system, the start-stop status and output of the unit can be obtained; IDRC Call volume and P IDRD Call volume.

Images