Economic Dispatch Method for Grid-Connected CCHP Microgrid System Based on Load Demand Response and Two-Layer Adjustable Robust Optimization

Abstract

The present invention discloses an optimal scheduling method for a combined cooling, heating and power (CCHP) microgrid based on load demand response and two-layer adjustable robust optimization. First, a load demand response strategy is used to model the microgrid system. Then, with the objective of minimizing the system operation cost under the worst operating conditions of renewable energy and load, according to the strong duality theory, the two-layer adjustable robust optimization model is decomposed into two linear models, and the column constraint generation algorithm is used to solve and formulate the optimal economic scheduling strategy of the system. In addition, by selecting an appropriate robust adjustment coefficient, the robustness of the scheduling plan can be appropriately adjusted to balance the economy and security of system operation. The method proposed by the present invention can effectively improve the energy utilization rate of the system and reduce the economic cost of system operation while reducing the impact of the uncertainty of renewable energy and load.

Description

Technical Field

[0001] The invention belongs to the technical field of power systems, and particularly relates to an economic dispatch method for a grid-connected combined cooling, heating and power microgrid system based on load demand response and two-layer adjustable robust optimization. Background Art

[0002] With the gradual depletion of resources such as oil and coal, countries currently pay great attention to the development and utilization of renewable energy. Due to the advantages of high energy utilization efficiency and high power supply reliability, the grid-connected combined cooling, heating and power microgrid system containing renewable energy has received extensive attention. This kind of microgrid system integrates distributed power sources, energy storage and loads, not only provides multiple energy supplies for load demands, but also is one of the effective means to improve the penetration rate of renewable energy. For the economic dispatch of such a system, the system is often modeled as a traditional deterministic optimization model. However, the prediction errors of renewable energy and loads pose challenges to the economic and safe operation of the system. Therefore, to ensure the economic operation of such a system, the uncertain impacts of renewable energy and loads must be minimized. Summary of the Invention

[0003] In view of the above problems, the invention proposes an economic dispatch method for a grid-connected combined cooling, heating and power microgrid system based on load demand response and two-layer adjustable robust optimization, which can reduce the uncertain impacts of renewable energy and loads, improve the energy utilization efficiency of the system, effectively reduce the system operation cost, and improve the stability of system operation.

[0004] The invention adopts the following technical solutions:

[0005] An economic dispatch method for a grid-connected combined cooling, heating and power microgrid system based on load demand response and two-layer adjustable robust optimization, and the specific design scheme is as follows:

[0006] Step (1) Model the grid-connected combined cooling, heating and power microgrid system;

[0007] Step (2) Optimize the electrical load by using load demand response and obtain time-of-use electricity price data;

[0008] Step (3) Establish a two-layer adjustable robust optimization model and decompose it into two linear models;

[0009] Step (4) Use the column constraint generation algorithm for iterative solution.

[0010] Furthermore, in the step (1), the grid-connected combined cooling, heating and power microgrid system includes a renewable energy system, an energy storage device, a heat storage device, a micro gas turbine, a waste heat recovery device, an electric boiler, a gas boiler, an absorption chiller and an electric chiller, and the waste heat recovery device, the electric boiler, the absorption chiller and the electric chiller are modeled as follows:

[0011] Waste heat recovery device: The waste heat recovery device can recover the waste heat gas generated when the micro gas turbine generates electricity to supply the heat load demand.

[0012]

[0013] Among them, η mt , η loss , COP mt and They are the power generation efficiency, loss rate, performance coefficient and power generation of the micro gas turbine respectively; and η hr are the heating power and heating efficiency of the waste heat recovery device respectively;

[0014] Electric boiler: Electric boiler can convert electrical energy into thermal energy to supply heat load demand.

[0015]

[0016] Among them, η ed , COP eb , and They are the efficiency, coefficient of performance, heating power and power consumption of the electric boiler respectively;

[0017] Electric Chiller: Electric chiller converts electrical energy into cooling energy to supply cooling load demand.

[0018]

[0019] Among them, COP ec , and are the coefficient of performance, cooling power and power consumption of the electric refrigerator respectively;

[0020] Absorption Chiller: An absorption chiller absorbs heat energy and converts it into cooling energy to supply the cooling load demand.

[0021]

[0022] Among them, COP ac , and They are the coefficient of performance, cooling power and heat consumption of the absorption refrigerator respectively;

[0023] Furthermore, in step (2), load demand response is used to optimize the electric load. The steps are as follows:

[0024] First, the membership function is introduced to represent the characteristics of the electric load, and the peak, valley and normal periods of the electric load are divided according to the fuzzy cluster analysis method:

[0025]

[0026]

[0027] Among them, is the electrical load before optimization, P old,min and P old,max are the minimum and maximum values of the electrical load before optimization. and are the membership functions of the electrical load during peak and valley periods.

[0028] Subsequently, the transitive closure matrix R λ is obtained by the absolute value subtraction method. Taking the classification tree λ = 3, the peak-valley-flat classification period data of the electrical load is obtained.

[0029] Then, a multi-objective optimization model is established with the minimum peak-valley difference of the electrical load and the maximum comprehensive satisfaction of electricity consumption as the objective function, and the time-of-use electricity price is obtained by solving with the multi-objective genetic algorithm. The multi-objective optimization model is as follows:

[0030]

[0031] Among them, q i represents the load during the peak (f), valley (g), and flat (p) periods, p i,min and p i,max are the minimum and maximum electricity prices during the three periods, P new,min and P new,max are the minimum and maximum values of the electrical load after optimization, S l and S p are the satisfaction of user habit change and the satisfaction of user expenditure, S l,min , S l,max , S p,min and S p,max are the minimum and maximum values of the satisfaction of user habit change and the minimum and maximum values of the satisfaction of user expenditure, λ l and λ p are the weight coefficients, Δq i , Δp i , p i are the load change amount, electricity price change amount, and electricity price data during the peak-valley-flat periods.

[0032] Finally, by introducing the demand elasticity coefficient, an electrical load response electricity price model is established based on the time-of-use electricity price data, and the optimized electrical load is obtained. The specific model is as follows:

[0033]

[0034] Among them, p0 is the initial electricity price.

[0035] Furthermore, the double robust optimization model in step (3) is specifically as follows:

[0036]

[0037] Among them, \(c\) is the coefficient matrix, \(y\) is the matrix representing the output power of the system source-side equipment, \(T\) represents the matrix transpose, \(u\) is the matrix describing the interaction state of the energy storage system and the interaction state between the system and the power grid, is the uncertainty variable describing the renewable energy and the load, is the uncertainty set.

[0038]

[0039] Y i =\{P mt ,P dis ,P chr ,P grid ,P excess ,P ec, P eb ,H b

[0040] H dis ,H chr ,H ac ,P pv ,P wt ,P load ,H h ,Q c \}

[0041]

[0042] U i =\{U bat,dis ,U grid ,U tst,dis \}

[0043]

[0044]

[0045]

[0046] Among them, \(P mt ,P dis ,P chr ,P grid ,P excess ,P ec ,P eb ,H b ,H dis ,H chr ,H ac ,P pv, P wt , P load , H h and Q c respectively represent the power generation power of the micro gas turbine, the charge and discharge power of the electrical energy storage system, the power of purchasing and selling electricity to the power grid, the power consumption of the electric refrigeration, the power consumption of the electric boiler, the heating power of the boiler, the charge and discharge heat power of the thermal energy storage system, the heat consumption power of the absorption refrigeration machine, the photovoltaic power generation, the wind power generation, and the three loads of electricity, heat, and cold. U bat,dis , U grid and U tst,dis represent the discharge state of the electrical energy storage system, the state of purchasing electricity from the power grid, and the heat release state of the thermal energy storage system;

[0047] The double-layer robust optimization model is divided into an outer-layer min structure model and an inner-layer max-min structure model. Among them, the inner-layer model is based on the strong duality theory, transforms the min structure into a max structure, then combines the two max structures and linearizes the model. The structures of the inner and outer layer models are specifically as follows:

[0048] Outer-layer model:

[0049]

[0050] Among them, c, D, K, F, G, I, d, and h are coefficient matrices. The outer-layer model takes minimizing the system operation cost as the objective function and u as the optimization variable.

[0051] Inner-layer model:

[0052]

[0053] Among them, γ, λ, v, π are dual variables, O 120 is a 120×1 zero matrix, E 120 is a 120th-order identity matrix, is the upper bound of π. When is large enough, the inner-layer model is a linear model, reducing the operation complexity of the algorithm. x and Δx are the prediction data of renewable energy and load in the deterministic optimization model and the maximum deviation value under the worst operating conditions of the system. Γ i is the robust adjustment coefficient; B is a diagonal matrix with 0-1 state variables as elements, and each element represents whether the renewable energy and load can take the interval maximum value in the uncertainty set during the corresponding time period. The form of B is as follows:

[0054]

[0055] B i ={B pv , B wt ,, B load , Bh , B c ,}

[0056] Among them, B pv , B wt , B load , B h and B c are the judgment matrices for the severe cases of uncertainty factors, which are respectively used to judge whether the photovoltaic output, wind power output, and the three loads of electricity, heat, and cold reach the interval boundary values of the uncertainty concentration.

[0057] System severe operating conditions:

[0058]

[0059]

[0060]

[0061]

[0062] X i = {P pv , P wt , P load , H h ,, Q c ,}

[0063] Furthermore, the "tunable" in the tunable robust optimization model described in step (3) means that this two-layer robust optimization model introduces a robust adjustment coefficient Γ to adjust the robustness of the system model in order to balance the economy and stability of system operation. Specifically:

[0064]

[0065] Γ i = {Γ pv , Γ wt , Γ load , Γ h , Γ c}

[0066] Among them, Γ is the robust adjustment coefficient, and its value range is [1, 24]. When the value of Γ is larger, the robustness of the system is stronger. When the value of Γ is 0, this robust model is equivalent to the traditional deterministic optimization model.

[0067] Furthermore, the column constraint generation algorithm and its solution steps described in step (4) are specifically as follows:

[0068] Main idea of the column constraint generation algorithm: The column constraint generation algorithm decomposes the original problem into two independent and coupled sub-problems, and transfers the coupling parameters of the two sub-problems during the solution process, and continuously iteratively alternates to solve to finally obtain the optimal solution of the original problem.

[0069] Solution steps of the column constraint generation algorithm:

[0070] Step (4.1): Set the upper and lower bounds of the day-ahead operating cost to UB = +∞ and LB = 0 respectively, the number of iterations to k = 1, ε = 5, and given the initial bad operating conditions of renewable energy and load demand

[0071] Step (4.2): Input the bad operating conditions of renewable energy and load demand Solve the outer-layer optimization model to obtain the economic optimal planned output y of each unit of the system k , the state of the energy storage device, the state of the interaction between the system and the power grid u k and the minimum operating cost LB of the system k , and update the lower bound of the operating cost to LB = LB k .

[0072] Step (4.3): Input u obtained in step (4.2) k , solve the inner-layer optimization model, check whether the energy storage device and the interaction device between the system and the power grid can cope with the fluctuations of the uncertainty variables, and obtain the uncertainty variables under the new bad operating conditions of the system and the minimum operating cost UB of the system k , and update the upper bound of the operating cost UB = min{UB, UB k}.

[0073] Step (4.4): Judge whether the condition UB - LB ≤ ε is satisfied. If it is satisfied, stop the iteration and return the optimal day-ahead scheduling plan y k ; otherwise, input the obtained in the inner-layer optimization model in step (4.3) into the outer-layer optimization model, let k = k + 1, and jump to step (4.2) to continue iterative optimization until the algorithm converges.

[0074] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0075] (1) The present invention establishes a two-layer robust optimization model with load demand response to solve the uncertainty problem of renewable energy and load in the grid-connected combined cooling, heating and power microgrid system.

[0076] (2) The present invention introduces a robust adjustment coefficient to adjust the robustness of the scheduling plan and improve the economic benefits of system operation.

[0077] (3) The present invention optimizes the electrical load of the system through load demand response, forms a reasonable power consumption guidance for users, thereby reducing the peak-valley difference of the electrical load and improving the reliability of the system operation. Description of the Drawings

[0078] Figure 1 It is the structure diagram of the grid-connected cooling, heating and power cogeneration microgrid system in the embodiment of the present invention;

[0079] Figure 2 It is the flow chart of the system scheduling strategy in the embodiment of the present invention;

[0080] Figure 3 It is the classification result diagram of the load demand response in the embodiment of the present invention;

[0081] Figure 4 It is the Pareto result diagram of the load demand response in the embodiment of the present invention;

[0082] Figure 5 It is the comparison diagram of the electrical load before and after optimization in the embodiment of the present invention. Detailed Embodiment

[0083] The following further clarifies the present invention in combination with specific implementation cases and drawings. The present invention proposes an economic scheduling method for a grid-connected cooling, heating and power cogeneration microgrid system based on load demand response and two-layer adjustable robust optimization. The structure of the grid-connected cooling, heating and power cogeneration microgrid system is as Figure 1 shown. To improve the economic operation efficiency of the system, a system scheduling strategy is established. The flow of the system scheduling strategy is as Figure 2 shown. The specific implementation steps are as follows:

[0084] (1) Model the grid-connected cooling, heating and power cogeneration microgrid system

[0085] The structure diagram of the grid-connected cooling, heating and power cogeneration microgrid system is as Figure 1 shown. The system components include a renewable energy system, an energy storage device, a heat storage device, a micro gas turbine, a waste heat recovery device, an electric boiler, a gas boiler, an absorption chiller and an electric chiller. The following introduces several key devices.

[0086] Waste heat recovery device: The waste heat recovery device can recover the waste heat gas generated during the power generation of the micro gas turbine to supply the heat load demand.

[0087]

[0088] Among them, η mt 、η loss 、COP mt and are respectively the power generation efficiency, loss rate, performance coefficient and power generation power of the micro gas turbine; and η hrare the heating power and heating efficiency of the waste heat recovery device respectively;

[0089] Electric boiler: The electric boiler can convert electrical energy into heat energy to supply the heat load demand.

[0090]

[0091] Among them, η eb , COP eb , and are the efficiency, coefficient of performance, heating power and power consumption of the electric boiler respectively;

[0092] Electric refrigerating machine: The electric refrigerating machine converts electrical energy into cooling energy to supply the cooling load demand.

[0093]

[0094] Among them, COP ec , and are the coefficient of performance, refrigerating power and power consumption of the electric refrigerating machine respectively;

[0095] Absorption refrigerating machine: The absorption refrigerating machine can absorb heat energy and convert it into cooling energy to supply the cooling load demand.

[0096]

[0097] Among them, COP ac , and are the coefficient of performance, refrigerating power and heat consumption of the absorption refrigerating machine respectively.

[0098] (2) Optimize the electrical load by using load demand response and obtain time-of-use electricity price data

[0099] The purpose of load demand response is to form a reasonable power consumption guidance for users, thereby reducing the peak-valley difference of the electrical load and improving the reliability of system operation:

[0100] First, introduce the membership function to represent the characteristics of the electrical load, and divide the peak-valley normal periods of the electrical load according to the fuzzy clustering analysis method:

[0101]

[0102]

[0103] Among them, is the electrical load before optimization, P old,min and P old,max are the minimum and maximum values of the electrical load before optimization, and It is the membership function of the electrical load during peak and valley periods.

[0104] Subsequently, the transitive closure matrix R is obtained by the absolute value subtraction method λ , taking the classification tree λ = 3, the peak-valley-flat classification period data of the electrical load is obtained.

[0105] Next, a multi-objective optimization model is established with the minimum peak-valley difference of the electrical load and the maximum comprehensive satisfaction of electricity consumption as the objective function, and the time-of-use electricity price is obtained by solving with the multi-objective genetic algorithm. The multi-objective optimization model is as follows:

[0106]

[0107] Among them, q i represents the load during the three periods of peak (f), valley (g), and flat (p), p i,min and p i,max are the minimum and maximum electricity prices during the three periods, P new,min and P new,max are the minimum and maximum values of the optimized electrical load, S l and S p are the satisfaction of user habit change and the satisfaction of user expenditure, S l,min , S l,max , S p,min and S p,max are the minimum and maximum values of the satisfaction of user habit change and the minimum and maximum values of the satisfaction of user expenditure, λ l and λ p are the weight coefficients, Δq i , Δp i , p i are the load change amount, electricity price change amount, and electricity price data during the peak-valley-flat periods.

[0108] Finally, by introducing the demand elasticity coefficient, an electrical load response electricity price model is established based on the time-of-use electricity price data. The specific model is as follows:

[0109]

[0110] Among them, p0 is the initial electricity price.

[0111] (3) Establish a two-layer robust optimization model based on the strong duality principle

[0112]

[0113] Among them, c is the coefficient matrix, y is the matrix representing the output power of the system source-side equipment, T represents the matrix transpose, u is the matrix describing the interaction state of the energy storage system and the interaction state between the system and the power grid, is the uncertainty variable describing the renewable energy and the load, is the uncertainty set.

[0114]

[0115] Y i = {P mt , P dis , P chr , P grid , P excess , P ec , P eb , H b

[0116] H dis , H chr , H ac , P pv , P wt , P load , H h , Q c}

[0117]

[0118] U i = {U bat,dis , U grid , U tst,dis}

[0119]

[0120]

[0121]

[0122] Among them, P mt , P dis , P chr , P grid , P excess , P ec , P eb , H b , H dis , H chr , H ac , P pv , P wt , P load , H h and Q c respectively represent the power generation power of the micro gas turbine, the charge and discharge power of the electric energy storage system, the power purchased and sold to the power grid, the power consumption of the electric refrigeration, the power consumption of the electric boiler, the heating power of the boiler, the charge and discharge heat power of the thermal energy storage system, the heat consumption power of the absorption refrigeration machine, the power generation power of the photovoltaic power generation, the wind power generation power, and the three loads of electricity, heat, and cold. U bat,dis , U grid and Utst,dis Indicates the discharge state of the electrical energy storage system, the state of purchasing electricity from the power grid, and the heat release state of the thermal energy storage system;

[0123] The two-layer robust optimization model is decomposed into an outer-layer min-structure model and an inner-layer max-min structure model. Among them, based on the strong duality theory, the inner-layer model transforms the min structure into a max structure, then combines the two max structures and linearizes the model. The structures of the outer and inner-layer models are specifically as follows:

[0124] Outer-layer model:

[0125]

[0126] Among them, c, D, K, F, G, I, d, and h are coefficient matrices. The outer-layer model takes minimizing the system operation cost as the objective function and u as the optimization variable.

[0127] Inner-layer model:

[0128]

[0129]

[0130] Among them, γ, λ, v, π are dual variables, O 120 is a 120×1 zero matrix, E 120 is a 120th-order identity matrix, is the upper bound of π. When gets large enough, the inner-layer model is a linear model, reducing the operation complexity of the algorithm. x and Δx are the predicted data of renewable energy and load in the deterministic optimization model and the maximum deviation values under the worst operating conditions of the system. Γ i is the robust adjustment coefficient; B is a diagonal matrix with 0-1 state variables as elements, and each element indicates whether renewable energy and load can take the interval maximum value in the uncertainty set during the corresponding time period. The form of B is as follows:

[0131]

[0132] B i ={B pv , B wt , B load , B h , B c ,}

[0133] Among them, B pv , B wt , B load , B h and B cIt is a determination matrix for severe cases of uncertainty factors, which are respectively used to determine whether the photovoltaic output, wind power output, and the three loads of electricity, heat, and cold reach the interval boundary values of the uncertainty concentration;

[0134] System severe operating conditions:

[0135]

[0136]

[0137]

[0138]

[0139] X i ={P pv ,P wt ,P load ,H h ,,Q c ,}

[0140] (4) Establish an adjustable robust optimization model. Here, "adjustable" means that this two-layer robust optimization model introduces a robust adjustment coefficient Γ to adjust the robustness of the system model to balance the economy and stability of system operation.

[0141]

[0142] Γ i ={Γ pv ,Γ wt ,Γ load ,Γ h ,Γ c}

[0143] Among them, Γ is the robust adjustment coefficient, and its value range is [1, 24]. When the value of Γ is larger, the robustness of the system is stronger. When the value of Γ is 0, this robust model is equivalent to the traditional deterministic optimization model.

[0144] (5) Run the column constraint generation algorithm to solve the above two-layer adjustable robust optimization model, which specifically includes the following steps:

[0145] The main idea of the column constraint generation algorithm: The column constraint generation algorithm decomposes the original problem into two independent and coupled sub-problems, and transmits the coupling parameters of the two sub-problems during the solution process, and continuously iteratively alternates to solve to finally obtain the optimal solution of the original problem.

[0146] The solution steps of the column constraint generation algorithm:

[0147] Step 1: Set the upper and lower bounds of the daily operating cost as UB = +∞ and LB = 0, the number of iterations as k = 1, ε = 5, and given the initial poor operating conditions of renewable energy and load demand.

[0148] Step 2: Input the poor operating conditions of renewable energy and load demand. Solve the outer-layer optimization model to obtain the economic optimal planned output y of each unit in the system, k the state of energy storage devices, the state of interaction between the system and the power grid u, k and the minimum operating cost LB of the system. k Update the lower bound of the operating cost to LB = LB. k .

[0149] Step 3: Input u obtained in Step 2. k Solve the inner-layer optimization model to check whether the energy storage devices and the interaction devices between the system and the power grid can cope with the fluctuations of uncertain variables, and obtain the uncertain variables under the new poor operating conditions of the system and the minimum operating cost UB of the system. k And update the upper bound of the operating cost UB = min{UB, UB k}.

[0150] Step 4: Judge whether the condition UB - LB ≤ ε is satisfied. If it is satisfied, stop the iteration and return the optimal daily dispatch plan y. k ; Otherwise, input the obtained in the inner-layer optimization model in Step 3 into the outer-layer optimization model, let k = k + 1, and jump to Step 2 to continue iterative optimization until the algorithm converges.

[0151] To verify the effectiveness of the proposed economic dispatch method for the grid-connected cooling, heating and power cogeneration microgrid system, the economic costs of the following five different operating strategies are compared and analyzed.

[0152] a. Operating strategy 1: A grid-connected cooling, heating and power cogeneration microgrid system with load demand response and double-layer adjustable robust optimization under normal daily prediction errors and Γ = 6.

[0153] b. Operating strategy 2: A grid-connected cooling, heating and power cogeneration microgrid system with load demand response and double-layer adjustable robust optimization under small daily prediction errors and Γ = 6.

[0154] c. Operating strategy 3: A grid-connected cooling, heating and power cogeneration microgrid system with load demand response and double-layer adjustable robust optimization under large daily prediction errors and Γ = 6.

[0155] d. Operating strategy 4: A grid-connected cooling, heating and power cogeneration microgrid system with load demand response and double-layer adjustable robust optimization under small daily prediction errors and Γ = 3.

[0156] e. Operation Strategy 5: Grid-connected combined cooling, heat and power microgrid system with load demand response and traditional deterministic optimization under normal day-ahead prediction error.

[0157] First, the classification process of electricity loads using fuzzy clustering analysis method is as Figure 3 shown. As Figure 3 can be seen, taking the number of classifications as 3, the classification information of the system's electricity loads can be obtained, and the classification information is shown in Table 1.

[0158] Subsequently, time-of-use electricity price data is obtained by solving the multi-objective load demand response model. The Pareto diagram obtained after solving is as Figure 4 shown. Taking point A of Figure 4 as the result, the time-of-use electricity price result is shown in Table 2.

[0159] Finally, by introducing the demand elasticity coefficient, the optimized curve is obtained. The comparison diagram of electricity loads before and after optimization is as Figure 5 shown. As Figure 5 can be seen, load demand response is adopted to achieve peak shaving and valley filling of users' initial electricity loads, improving the operation stability of the system.

[0160] Table 3 shows the system operation costs under five operation strategies. Among them, the real-time operation cost is an important indicator to measure the economic operation benefit of the system, while the day-ahead plan cost and adjustment cost more illustrate the robustness level of the system.

[0161] As can be seen from Table 3, comparing Operation Strategy 1 and Operation Strategy 5, under the same prediction error, compared with the system with traditional deterministic optimization, the system with double-layer adjustable robust optimization reduces the real-time operation cost by 6.16%. Under the same scheduling strategy, the optimization effect of the double-layer adjustable robust optimization strategy gradually improves with the improvement of prediction accuracy. However, when the prediction error is small, the adjustment cost of the system shows a negative number, indicating that the robustness of the system is too high at this time, and the robust adjustment coefficient needs to be adjusted to reduce the robustness of the system. Therefore, the robust adjustment coefficient of the system is reduced to 3 when the prediction error is small, and Operation Strategy 4 is obtained. As can be seen from Table 3, the day-ahead plan cost of Operation Strategy 4 is significantly reduced compared with that of Operation Strategy 2, which indicates that the robustness of the system has decreased at this time, and the real-time operation cost at this time is also lower than that of Operation Strategy 2, indicating that the economic benefit of the system has been improved after correcting the robust adjustment coefficient.

[0162] In summary, in the grid-connected combined cooling, heat and power microgrid system based on load demand response and double-layer adjustable robust optimization, the uncertainty problems of renewable energy and load can be effectively handled, the system operation cost can be reduced, and the economy and stability of the system operation can be improved.

[0163] Table 1 Classification Results of Peak-Valley-Flat Periods of Electric Load

[0164]

[0165] Table 2 Time-of-Use Electricity Price Results

[0166]

[0167] Table 3 Comparison of System Operating Costs under Five Operating Strategies

[0168]

[0169] The above examples are used to explain the present invention rather than limit it. Any modifications and changes made to the present invention within the spirit and scope of the claims of the present invention fall within the protection scope of the present invention.

Claims

1. An optimal dispatching method for a combined cooling, heating and power microgrid based on load demand response and double-layer adjustable robust optimization, characterized in that: It includes the following steps: Step 1: Model the combined cooling, heat and power (CCHP) microgrid system; Step 2: Optimize the electrical load using load demand response and obtain time-of-use electricity price data; Step 3: Establish a two-layer adjustable robust optimization model and decompose it into two linear models; Step 4: Use the column constraint generation algorithm for iterative solution; In the two-layer robust optimization model in Step 3, it is specifically as follows: Among them, \(c\) is the coefficient matrix, \(y\) is the matrix representing the output power of the system source-side equipment, \(T\) represents the matrix transpose, and \(u\) is the matrix describing the interaction state of the energy storage system and the interaction state between the system and the power grid. is the uncertainty variable describing renewable energy and load, is the uncertainty set; y = [Y i t＝1 ,..., Y i t＝24 T ​ Y i = {P mt , P dis , P chr , P grid , P excess , P ec , P eb , H b H dis ,H chr ,H ac ,P pv ,P wt ,P load ,H h ,Q c} U i = {U bat,dis , U grid , U tst,dis} Among them, P mt 、P dis 、P chr 、P grid 、P excess 、P ec 、P eb 、H b 、H dis 、H chr 、H ac 、P pv 、P wt 、P load 、H h and Q c respectively represent the power generation power of the micro gas turbine, the charge and discharge power of the electric energy storage system, the power purchased and sold to the power grid, the power consumption of the electric refrigeration, the power consumption of the electric boiler, the heating power of the boiler, the charge and heat release power of the thermal energy storage system, the heat consumption power of the absorption chiller, the power generation power of the photovoltaic power generation, the wind power generation power, and the three loads of electricity, heat, and cold. U bat，dis 、U grid and U tst，dis represent the discharge state of the electric energy storage system, the state of purchasing electricity from the power grid, and the heat release state of the thermal energy storage system. and represent the fluctuation ranges of the renewable energy output of the system and the three loads of cold, heat, and electricity. The two-layer robust optimization model is divided into an outer-layer min structure model and an inner-layer max-min structure model. Among them, based on the strong duality theory, the inner-layer model transforms the min structure into a max structure, then combines the two max structures and linearizes the model. The structures of the inner and outer layer models are specifically as follows: Outer-layer model: Among them, D, K, F, G, I, d, and h are coefficient matrices; the outer-layer model takes minimizing the system operation cost as the objective function and u as the optimization variable; Inner-layer model: where γ, λ, v, and π are dual variables, O 120 is a 120×1 zero matrix, E 120 is a 120th-order identity matrix, is the upper bound of π. When is large enough, the inner-layer model is a linear model. x and Δx are the predicted data of renewable energy and load in the deterministic optimization model and the maximum deviation values under severe system operating conditions; Γ i is the robust regulation coefficient; B is a diagonal matrix with 0-1 state variables as elements, and each element represents whether the renewable energy and load can take the interval maximum value in the uncertainty set during the corresponding time period. The form of B is as follows: B i = {B pv , B wt , B load , B h , B c ,} Among them, B pv , B wt , B load , B h and B c are the judgment matrices for the severe cases of uncertain factors, which are respectively used to judge whether the photovoltaic output, wind power output, and the three loads of electricity, heat, and cold reach the interval boundary values of the uncertainty concentration; System in severe operating conditions: X i = {P pv , P wt , P load , H h , Q c ,}; In the adjustable robust optimization model in Step 3, "adjustable" means that this two-layer robust optimization model introduces a robust adjustment coefficient Γ to adjust the robustness of the system model, specifically as follows: Γ i = {Γ pv , Γ wt , Γ load , Γ h , Γ c} When the value of Γ is larger, the robustness of the system is stronger. When the value of Γ is 0, this robust model is equivalent to the traditional deterministic optimization model.

2. The optimal dispatching method for a combined cooling, heating and power (CCHP) microgrid based on load demand response and two-layer adjustable robust optimization according to claim 1, characterized in that: In Step 1, the CCHP microgrid equipment includes a renewable energy system, an energy storage device, a heat storage device, a micro gas turbine, a waste heat recovery device, an electric boiler, a gas boiler, an absorption chiller, and an electric chiller. Among them, the waste heat recovery device, the electric boiler, the absorption chiller, and the electric chiller are modeled as follows: Waste heat recovery device: The waste heat recovery device recovers the waste heat gas generated during the power generation of the micro gas turbine to supply the heat load demand; Among them, η mt , η loss , COP mt and are respectively the power generation efficiency, loss rate, coefficient of performance, and power generation power of the micro gas turbine; and η hr are respectively the heating power and heating efficiency of the waste heat recovery device; Electric boiler: The electric boiler converts electrical energy into heat energy to supply the heat load demand; Among them, η eb , COP eb , and are the efficiency, coefficient of performance, heating power and power consumption of the electric boiler respectively; Electric chiller: The electric chiller converts electrical energy into cooling energy to supply the cooling load demand; Among them, COP ec , and are the coefficient of performance, refrigerating power, and power consumption of the electric refrigerating machine, respectively; Absorption chiller: The absorption chiller absorbs heat energy and converts it into cooling energy to supply the cooling load demand; Among them, COP ac , and are the coefficient of performance, refrigeration power, and heat consumption power of the absorption chiller, respectively.

3. The optimal dispatching method for a combined cooling, heating and power microgrid based on load demand response and two-layer adjustable robust optimization according to claim 1, wherein: In Step 2, optimizing the electrical load using load demand response is specifically as follows: Step 2.1: Introduce a membership function to represent the characteristics of the electrical load, and divide the peak, valley, and normal periods of the electrical load according to the fuzzy clustering analysis method: Among them, is the electrical load before optimization, Po l d,min and Po l d,max are the minimum and maximum values of the electrical load before optimization, and are the membership functions of the electrical load during peak and valley periods; Step 2.2: Obtain the transitive closure matrix R by the absolute value subtraction method λ , take the classification tree λ = 3 to obtain the peak-valley-flat classification period data of the electrical load Step 2.3: Establish a multi-objective optimization model with the minimum peak-valley difference of the electrical load and the maximum comprehensive electricity consumption satisfaction as the objective function, and use the multi-objective genetic algorithm to solve for the time-of-use electricity price; the multi-objective optimization model is: min(P new，max -P new，min ) 2 -(λ l S l +λ p S p ) Among them, q i represents the load under three time periods of peak (f), valley (g), and flat (p). p i,min and p i,max are the minimum and maximum electricity prices under the three time periods. P new,min and P new,max are the minimum and maximum values of the optimized electricity load. S l and S p are the satisfaction degrees of user habit change and user expenditure satisfaction. S l，min , S l，max , S p，min and S p，max are the minimum and maximum values of the satisfaction degree of user habit change and the minimum and maximum values of the satisfaction degree of user expenditure. λ l and λ p are the weight coefficients. Δq i , Δp i , p i are the load change amount, electricity price change amount, and electricity price data under the peak-valley-flat time periods; Step 2.4: By introducing the demand elasticity coefficient, establish an electrical load response electricity price model based on the time-of-use electricity price data to obtain the optimized electrical load; the electrical load response electricity price model is specifically: i, j ∈ {f, p, g} Among them, p0 is the initial electricity price.

4. The economic dispatch method of the grid-connected cooling, heating and power cogeneration microgrid system based on load demand response and double-layer adjustable robust optimization according to claim 1, wherein: The column constraint generation algorithm and its solution steps in Step 4 are specifically as follows: Step 4.1: Set the upper and lower bounds of the operating cost before a certain date as UB = +∞ and LB = 0, the number of iterations as k = 1, ε = 5, and given the initial bad operating conditions of renewable energy and load demand Step 4.2: Input the severe operating conditions of renewable energy and load demand Solve the outer optimization model to obtain the economically optimal planned output y of each unit in the system k , the state of energy storage devices, the state of interaction between the system and the power grid u k and the minimum operating cost LB of the system k , update the lower bound of the operating cost to LB = LB k ; Step 4.3: Input the u obtained in Step 4.2 k , solve the inner-layer optimization model, check whether the energy storage device, system and grid interaction device can cope with the fluctuations of uncertain variables, and obtain the uncertain variables under the new system's severe operating conditions and the minimum operating cost UB of the system k , and update the upper bound of the operating cost UB = min{UB, UB k}; Step 4.4: Determine whether the condition UB - LB ≤ ε is satisfied. If it is satisfied, stop the iteration and return the optimal day-ahead scheduling plan y k ; otherwise, input the obtained in the inner-layer optimization model in Step 4.3 into the outer-layer optimization model, let k = k + 1, and jump to Step 4.2 to continue the iterative optimization until the algorithm converges.

Images