An economical dispatch method for multi-energy systems in hydrogen-containing buildings under extreme weather events

Abstract

This invention discloses an economic scheduling method for multi-energy systems in hydrogen-containing buildings under extreme weather events, involving the intersection of energy systems and artificial intelligence. The method includes: establishing a multi-timescale economic scheduling problem for risk perception in multi-energy systems of hydrogen-containing buildings; reducing the source-load related scenarios and decomposing the reduced-scenario risk-perception economic scheduling problem into binary variable decision search and inner-layer continuous variable linear programming solution; generating a particle swarm related to the binary variables of system components and evaluating to obtain the globally optimal particle; setting several iterations, evaluating the fitness value of each particle and updating the globally optimal particle; decoding the position of the globally optimal particle to obtain the optimal system component operating binary variables, obtaining continuous decisions for electric-hydrogen mobile resource leasing; rounding the obtained continuous decisions for electric-hydrogen mobile resource leasing, and solving the inner-layer linear programming again to output the final economic scheduling scheme. This invention achieves lower risk-perception economic scheduling costs.

Description

Technical Field

[0001] This invention relates to the intersection of energy systems and artificial intelligence, specifically to an economic dispatch method for multi-energy systems in hydrogen-containing buildings under extreme weather events. Background Technology

[0002] With the rapid development of renewable energy, hydrogen-based multi-energy building systems have become an important pathway to achieve deep decarbonization and enhance energy resilience. This system integrates multiple energy sources, including electricity, heat, and hydrogen, enabling flexible equipment coupling and cross-seasonal, cross-regional dispatch capabilities. Furthermore, hydrogen's unique black-start capability enhances emergency recovery capabilities. However, frequent extreme weather events bring complexity and high-risk challenges to system operation. Simultaneously, renewable energy generation and electricity load are susceptible to weather fluctuations, while hydrogen production and storage have relatively slow dynamic responses, requiring coordinated operation with rapidly changing electrical and thermal processes. These multi-timescale coupling characteristics and the high frequency of extreme weather lead to significant uncertainties and operational risks for the system. Therefore, researching economic dispatch methods for hydrogen-based multi-energy building systems under extreme weather conditions is of great significance.

[0003] Currently, numerous studies have focused on the economic dispatch of multi-energy systems in hydrogen-containing buildings, with particular emphasis on electricity-heat synergistic optimization, hydrogen production and utilization, and renewable energy integration. Early studies primarily relied on deterministic dispatch to improve system operational economics. To address the uncertainty of renewable energy and load, subsequent studies proposed stochastic risk-neutral dispatch methods, including hierarchical model predictive control, Lyapunov optimization, and deep reinforcement learning strategies. Although these methods consider uncertainty, they are mostly limited to risk-neutral decision-making. To optimize operational risks caused by extreme weather events, recent studies have introduced robust optimization and methods that consider conditional risk value constraints. While the above studies have made some progress, they still have the following limitations: (1) Most studies model the heterogeneous energy subsystems of electricity, heat, and hydrogen as a single time scale, failing to fully consider their cross-time scale dynamic coupling characteristics; (2) There is insufficient research on the collaborative operation mechanism of fixed and mobile energy resources under extreme weather events, limiting the flexible allocation capability of emergency resources. Therefore, it is urgent to study multi-time scale economic dispatch methods for multi-energy systems in hydrogen-containing buildings that consider electricity-hydrogen mobile resource leasing under extreme weather events.

[0004] The above research faces the following challenges: (1) the existence of multi-timescale dynamic coupling characteristics of the electric-thermal-hydrogen subsystem; (2) the existence of spatiotemporal coupling constraints related to energy storage systems and energy balance; and (3) the existence of large-scale mixed integer decision variables. To address these challenges, this invention proposes a risk-aware multi-timescale economic scheduling method for electric-hydrogen mobile resources that combines farthest-first traversal clustering with improved quantum behavior particle swarm optimization. Through scenario dimensionality reduction and decomposition strategies based on the improved quantum behavior particle swarm optimization algorithm, the economic efficiency and safety of hydrogen-containing building multi-energy systems under extreme weather events are optimized. Summary of the Invention

[0005] To address the shortcomings of existing research, this invention provides an economic scheduling method for hydrogen-containing building multi-energy systems under extreme weather events. Its purpose is to realize the risk-aware multi-timescale economic scheduling problem of hydrogen-containing building multi-energy systems considering electric-hydrogen mobile resources under extreme weather events. This method compensates for the lack of research on the dynamic coupling characteristics across time scales and the collaborative operation mechanism of electric-hydrogen fixed resources and electric-hydrogen mobile resources under extreme weather events in existing research on hydrogen-containing building multi-energy systems. Furthermore, by proposing a decomposition strategy combining farthest-first traversal clustering and an improved quantum behavior particle swarm optimization algorithm, the large-scale mixed integer decision variable problem in the model is effectively solved.

[0006] To achieve the above objectives, the present invention employs the following technical solution:

[0007] This invention provides an economic dispatch method for multi-energy systems in hydrogen-containing buildings under extreme weather events, comprising:

[0008] Step 1: Establish a multi-timescale economic dispatch problem for risk perception of hydrogen-containing buildings and multi-energy systems under extreme weather events, considering the leasing of electric and hydrogen mobile resources;

[0009] Step 2: Use the farthest first traversal method to reduce the source load related scenarios and decompose the risk perception economic scheduling problem after scenario reduction into two stages: binary variable decision search and inner continuous variable linear programming solution.

[0010] Step 3: Generate a particle swarm related to the binary variables of system components and evaluate the fitness value of each particle to obtain the globally optimal particle;

[0011] Step 4: Set up several iterations. In each iteration, use the global best particle to update the position of each particle, evaluate the fitness value of each particle, and update the global best particle.

[0012] After the iterations of steps 5 and 4 are completed, the global optimal particle position is decoded to obtain the optimal system component running binary variable. The inner linear programming problem is solved by fixing this binary variable to obtain the continuous decision of electric-hydrogen mobile resource leasing.

[0013] Step 6: After rounding the continuous decision of electric-hydrogen mobile resource leasing obtained in Step 5, solve the inner-layer linear programming again to output the final economic scheduling scheme.

[0014] Furthermore, the risk perception multi-timescale economic scheduling problem of hydrogen-containing building multi-energy system considering electric-hydrogen mobile resources established in step 1 includes constraints, decision variables, and objective functions.

[0015] First, the constraints include multi-timescale decision-making models, photovoltaic power generation models, electric-hydrogen mobile resource leasing models, hydrogen energy storage system models, ground source heat pump models, electric boiler models, hot water tank models, energy balance models, generation of uncertain scenario sets models, operating cost models, and risk perception models.

[0016] The multi-timescale decision-making model is as follows:

[0017] (1),

[0018] (2),

[0019] (3),

[0020] (4),

[0021] (5),

[0022] (6),

[0023] (7),

[0024] (8),

[0025] Wherein: Formulas (1)-(3) represent the electrical energy decision time slot, thermal energy decision time slot and hydrogen energy decision time slot models, respectively; , and These represent the sets of time slots for electrical energy decisions, thermal energy decisions, and hydrogen energy decisions, respectively. , and These represent the time slot indexes for electrical energy decisions, thermal energy decisions, and hydrogen energy decisions, respectively. , and The total electrical energy decision time slot, the total thermal energy decision time slot, and the total hydrogen energy decision time slot are respectively represented; Formulas (4) to (6) respectively represent the relationship between the electrical energy decision time slot and the thermal energy decision time slot, the relationship between the electrical energy decision time slot and the hydrogen energy decision time slot, and the relationship between the thermal energy decision time slot and the hydrogen energy decision time slot; and These represent the normal operating slots and the operating slots during extreme weather events for hydrogen-containing building multi-energy systems, respectively. , as well as These represent the time slot lengths for electrical energy decision-making, thermal energy decision-making, and hydrogen energy decision-making, respectively.

[0026] The photovoltaic power generation model is modeled as follows:

[0027] (9),

[0028] (10)

[0029] in: For the first Photovoltaic power reduction per electrical time slot; For the first Maximum photovoltaic power generation per electrical time slot; For the first Actual output power of photovoltaic cells in each electrical time slot.

[0030] To enhance the resilience of the system under extreme weather events, operators of hydrogen-based multi-energy building systems can pre-lease mobile resources from the market to supply energy shortages. The model for their electricity-hydrogen mobile resource leasing is as follows:

[0031] (11),

[0032] (12)

[0033] (13)

[0034] (14)

[0035] (15)

[0036] (16)

[0037] (17)

[0038] (18)

[0039] (19)

[0040] (20)

[0041] (twenty one),

[0042] in: and These represent the number of electric vehicles rented on a one-time basis and the number of hydrogen fuel cell vehicles rented on a one-time basis, respectively. and These represent the maximum number of electric vehicles and the maximum number of hydrogen fuel cell vehicles that can be rented in a single transaction, respectively. Represents the set of positive integers; and The first The output power of the electric time slot tram and the first Output power of hydrogen fuel cell vehicle in one electrical time slot; and The first The number of trams powered by each time slot and the first The number of hydrogen fuel cell vehicles supplied with hydrogen per electrical time slot; and These are the maximum output power of electric vehicles and the maximum output power of hydrogen fuel cell vehicles, respectively. and They were respectively in the second The total electrical energy storage of all trams in each time slot and the first The total hydrogen storage capacity of all hydrogen fuel cell vehicles in each hydrogen time slot; and These represent the total electrical energy storage of all electric vehicles in the initial electric time slot and the total hydrogen storage of all hydrogen fuel cell vehicles in the initial hydrogen time slot, respectively. and These are the storage capacities for a single electric vehicle and a single hydrogen fuel cell vehicle, respectively. Indicates the first The thermal energy output of a hydrogen fuel cell vehicle in a thermal time slot; and These refer to the hydrogen-electric efficiency and the thermal-electric efficiency of hydrogen fuel cell vehicles, respectively. For the discharge efficiency of the tram, This refers to the heat recovery efficiency of hydrogen fuel cell vehicles.

[0043] The hydrogen energy storage system is modeled as follows:

[0044] (twenty two),

[0045] (twenty three),

[0046] (twenty four),

[0047] (25)

[0048] (26)

[0049] (27)

[0050] (28)

[0051] (29)

[0052] (30)

[0053] (31),

[0054] (32),

[0055] (33),

[0056] (34),

[0057] (35),

[0058] (36)

[0059] (37)

[0060] in: and The first The input power of the first time slot electrolytic cell and the first The input power of a hydrogen time-slot electrolyzer. This is the maximum input power of the electrolytic cell. This represents the maximum ramp rate of the electrolytic cell's power. For the first Hydrogen production per hydrogen time-slot electrolyzer, The electro-hydrogen conversion efficiency of the electrolyzer; Indicates the first Each hydrogen time slot is used to purchase hydrogen from the hydrogen market in the absence of extreme weather events. To maximize the amount of hydrogen purchased; For the first The input power of a single-slot fuel cell, This represents the maximum output power of the fuel cell. This represents the maximum ramp rate for fuel cell power. For the first Hydrogen consumption per hydrogen time slot fuel cell and These are the hydrogen-electric efficiency and thermal-electric efficiency of the fuel cell, respectively. For the heat recovery efficiency of fuel cells, For the first The thermal energy generated by the thermal time-slot fuel cell; Equations (32)-(34) describe why the electrolyzer and fuel cell cannot operate simultaneously. Using binary values, this avoids the low round-trip efficiency caused by short energy cycles ("electrical energy to hydrogen energy, then back to electrical energy or heat energy") within the same operating window. A strict mutual exclusion operating strategy is implemented. If the electrolyzer is not working, the cell will not work; conversely, if the cell is not working, the fuel cell will not work. It is a very large positive integer; For the first The storage capacity of each hydrogen time slot hydrogen tank This is the maximum storage capacity of the hydrogen tank. This indicates the relationship between the electrical time slot length and the hydrogen time slot length. and They represent the first time. In the hydrogen time slot, the first The output power of the fuel cell in the current time slot and the first In the hydrogen time slot, the first The input power of the electrolytic cell in each time slot; the inequality constraint described by formula (37) reflects a non-first-come, first-served operating logic, that is, in the first time slot... The first hydrogen time slot At the end of each power time slot, the cumulative fuel cell consumption must not exceed the available storage capacity.

[0061] The ground source heat pump model is modeled as follows:

[0062] (38),

[0063] (39)

[0064] in: For the first Input power of the ground source heat pump in each time slot; This is the maximum input power of the ground source heat pump; The electro-thermal conversion efficiency of a ground source heat pump; For the first The output heat energy of a thermal time slot ground source heat pump.

[0065] The electric boiler model is modeled as follows:

[0066] (40)

[0067] (41),

[0068] in: Indicates the first Input power of the electric boiler in each time slot; This refers to the maximum input power of the electric boiler. The electro-thermal conversion efficiency of the electric boiler; The first The output heat energy of a single-time slot electric boiler.

[0069] The hot water tank model is shown below.

[0070] (42),

[0071] (43),

[0072] (44),

[0073] (45)

[0074] (46)

[0075] in: For the first Storage capacity of each hot water tank in a hot time slot; This is the maximum storage capacity of the hot water tank. and These are the input efficiency and output efficiency of the hot water tank, respectively. For the heat retention rate of the hot water tank, and The first The input thermal power of the hot water tank in the first thermal time slot and the first Each hot water tank in a hot time slot outputs heat power. and These are the maximum input heat power and the maximum output heat power of the hot water tank, respectively. For the first The first hot-slot binary variable is used to avoid the second hot-slot binary variable. Each hot water tank in a heat time slot simultaneously charges / releases heat, when... At that time, the hot water tank is not allowed to release heat. At that time, it is not allowed to heat the hot water tank.

[0076] The energy balance model is as follows:

[0077] (47)

[0078] (48)

[0079] (49)

[0080] (50),

[0081] in: and The first Electrical time slots have priority Electrical loads and those with priority The amount of electricity load reduction, and The first The heat load of the first thermal time slot and the first Heat load reduction per thermal time slot.

[0082] The uncertainty scenario set generated includes uncertainties related to photovoltaic power generation, electrical load, heat load, and the duration of extreme weather events. The specific modeling method is as follows:

[0083] (51),

[0084] (52),

[0085] (53),

[0086] (54),

[0087] (55),

[0088] (56),

[0089] (57),

[0090] (58),

[0091] (59),

[0092] (60)

[0093] (61),

[0094] (62),

[0095] (63),

[0096] (64),

[0097] in: , and The first Predicted value of maximum output power of photovoltaic power in each time slot, the first The predicted value of the heat load in the first thermal time slot and the first Predicted values ​​of electrical load per electrical time slot; , and The scenarios are extreme events following a uniform distribution. Influencing factors of photovoltaic, heat load and electrical load; For the scene Next Photovoltaic power generation per electrical time slot; For the scene Next Heat load demand per thermal slot; For the scene Next The priority of each time slot is The demand for electrical load; For the first A fixed proportion of the demand for priority electrical loads in the total demand for electrical loads; It follows a normal distribution; , and For the scene Next Prediction error of maximum photovoltaic output power in individual time slots, and scenarios Next Prediction error of heat load under each thermal time slot and scenarios Next Prediction error of electrical load under individual electrical time slots; and These are the minimum and maximum standard deviations of the maximum output power of photovoltaic power, respectively. and These are the minimum and maximum standard deviations of the electrical load, respectively. and These are the minimum and maximum standard deviations of the heat load, respectively. and Scenes The start and end times of the following extreme events and The start and end times of the extreme event windows are known. and All are random discrete jitter quantities; To generate each scene The probability of.

[0098] The operating cost of this system comprises five components: photovoltaic load reduction cost, equipment depreciation cost, multi-source load reduction cost, hydrogen procurement cost, and mobile resource leasing cost. The operating cost model is as follows:

[0099] (65)

[0100] in: , , , as well as These are the photovoltaic cost reduction coefficient, the electrolyzer depreciation cost coefficient, the fuel cell depreciation cost coefficient, the electricity load reduction cost, and the heat load reduction cost coefficient, respectively. For the price of hydrogen; and These are the rental costs for a single electric vehicle and the rental costs for a single hydrogen fuel cell vehicle, respectively. This represents an index of large-scale extreme weather event scenarios, where: and Separate the scene set and the total number of scenes; , , , , and Scenes The following correspondence , , , , and The value.

[0101] The risk perception model is modeled as follows.

[0102] (66),

[0103] (67)

[0104] (68)

[0105] in: To calculate the highest Average operating costs; For when Confidence level is Risk value at time, auxiliary variable Used to measure each scenario Costs exceeding The part.

[0106] Secondly, the decision variables include the number of trolleys rented in a single transaction. Number of hydrogen fuel cell vehicles rented at one time , No. Actual output power of photovoltaic power in each time slot , No. The output power of the electric time slot tram , No. The output power of a hydrogen fuel cell vehicle in a single time slot , No. Input power of a single-slot fuel cell , No. Input power of individual electric boiler time slot , No. The input power of the ground source heat pump in the individual time slot , No. The priority of each time slot is Electricity load reduction , No. The hot water tank input heat power in each hot time slot , No. Each hot water tank outputs heat power in a specific time slot. , No. Heat load reduction per thermal time slot , No. Input power of each hydrogen time-slot electrolyzer , No. Each hydrogen time slot purchases hydrogen from the hydrogen market in the absence of extreme weather events. .

[0107] Finally, the objective function is modeled as follows:

[0108] (69)

[0109] in: For the scene The set of all variables specifically includes the number of trams rented in a single transaction. Number of hydrogen fuel cell vehicles rented at one time , No. Input power of individual time-slot electrolytic cells , No. The output power of a single-time-slot fuel cell , No. The input power of the ground source heat pump in each time slot , No. Input power of individual electric boiler time slot , No. The output power of the electric time-slot tram , No. The output power of a hydrogen fuel cell vehicle in a single time slot , No. Photovoltaic power reduction in individual time slots , No. Electrical time slots have priority Electricity load reduction , No. Total electrical energy storage of all trams in each time slot , No. The output heat energy of the time-slot ground source heat pump , No. The output heat energy of the electric boiler with a time slot , No. The heat energy generated by a time-slot fuel cell , No. The heat energy output by a hydrogen fuel cell vehicle in a single time slot , No. The hot water tank input heat power in each hot time slot , No. Each hot water tank outputs heat power in a specific time slot. , No. Binary variables related to the operation of the hydrogen time slot and the electrolyzer and fuel cell , No. Heat load reduction per thermal time slot , No. Hot water tank storage capacity per hot time slot , No. Input power of each hydrogen time-slot electrolyzer , No. Binary variables related to the filling and discharging of hot water tanks in each hot time slot , No. Each hydrogen time slot purchases hydrogen from the hydrogen market in the absence of extreme weather events. , No. Hydrogen production per hydrogen time slot electrolyzer , No. Hydrogen consumption per hydrogen time slot fuel cell , No. Storage capacity of each hydrogen time slot hydrogen tank , No. Total hydrogen storage capacity of all hydrogen fuel cell vehicles in each hydrogen time slot ; express The weighting coefficients.

[0110] Furthermore, the steps in step 2, which employ the farthest point first traversal algorithm to reduce uncertain scenarios and extract representative scenarios and filter scenarios related to the duration of extreme weather, are as follows:

[0111] (1) Input the generated large-scale uncertainty scenario set and their corresponding probability distribution ,in: , , , and These are the weighting coefficients for photovoltaic power generation, electrical load, and heat load, respectively. Total electrical time slots for the operation of hydrogen-containing multi-energy systems;

[0112] (2) Each extreme weather event scenario The multi-timescale sequence data of photovoltaic power generation, electrical load, and heat load are first mapped to their corresponding feature vectors. Secondly, initialize a representative set of scenarios. That is, to calculate the feature vectors of each scene. The norm of the scene with the largest norm will be indexed. As the first representative scene and Then, for all scenarios and scene Constructing the vector distance matrix Finally, for all scenarios Initialize distance array ,in: Euclidean distance;

[0113] (3) Among the remaining scenes, select the representative scene set that is most relevant to the current scene. The farthest scene ,in: This is an index for a representative set of scenarios extracted from a set of large-scale uncertain scenarios. The number of representative scenes will be concentrated; subsequently, Add representative scene set Finally, update the distance array. Finally, repeat this process until a representative set of scenes is reached. The number of medium scenes equals the number of target representative scenes. ;

[0114] (4) Based on all scenarios With representative scene set Construct scene assignment vectors Then all scenes Assigned to the representative scene closest to it This leads to clustering to calculate each representative scenario. probability and obtain each scene start time of extreme weather events and stop time ;

[0115] (5) Output the representative scene set extracted Its corresponding probability and representative scenes Corresponding extreme event start and end time parameters and .

[0116] Furthermore, in step 2, the risk-aware economic scheduling problem after scenario reduction is decomposed into two stages: binary variable decision search and inner-layer continuous variable linear programming solution. The system components run the outer-layer decision related to binary variable optimization, including scenario optimization. Below are all binary variables related to the operation of the hydrogen time slot, electrolyzer, and fuel cell. and scene All binary variables related to the filling and discharging of hot water tanks in the following hot time slots Based on this, the inner-level linear programming problem related to continuous variable decision-making is established as follows:

[0117] (70)

[0118] (71),

[0119] (72),

[0120] (73),

[0121] (74),

[0122] in: Represents the set of real numbers; For the scene Next Each hydrogen time slot is a binary variable related to the operation of the electrolyzer and fuel cell; For the scene Next Binary variables related to the filling and discharging of a hot water tank in a specific thermal time slot; and These represent the start / stop state matrices for the electrolyzer / fuel cell and the start / stop state matrices for the hot water tank charging / discharging, respectively.

[0123] Furthermore, the specific steps in step 3 for generating the particle swarm related to the binary variables of system components and evaluating the fitness value of each particle to obtain the globally optimal particle are as follows:

[0124] (1) Let the particle swarm The number of particles is The original time-series electrolyzer / fuel cell binary variables and hot water tank charge / discharge variables are compressed into several consecutive blocks representing the start / stop modes of the equipment within corresponding time periods using block coding. and Their sizes are respectively and ;

[0125] (2) Let and These represent the number of hydrogen time slots in each electrolyzer / fuel cell start-stop block and the number of thermal time slots in each hot water tank charge / discharge start-stop block, respectively. The number of start-stop blocks is as follows: and This encoding method compresses the exploration space of each particle, avoiding unnecessary frequent device start-ups and shutdowns;

[0126] (3) The random generation size is Particle swarm ,in: , and This is achieved by adjusting the original particle binary block structure, i.e. and Their sizes are respectively and , Each particle represents two probability blocks concatenated row-by-row in the extreme weather event scenario; and the relevant binary encoded blocks within the extreme window of each scenario. Set to 1, which is used to operate the fuel cell only during extreme windows, consuming hydrogen to generate electricity and heat.

[0127] (4) For each particle, the probability block is repeatedly mapped to the full time domain according to the block length, i.e. and ;

[0128] (5) For each particle, a threshold function is used to obtain the corresponding binary start-stop matrix. and ,Right now .

[0129] (6) Decision matrix based on the outer binary variables of each particle The fitness value of each particle is obtained by solving P2 using the interior-point method. And save the optimal value for each individual particle. ;

[0130] (7) Obtain the globally optimal particle With group diversity .

[0131] Furthermore, in step 4, the specific steps for updating the position of each particle using the globally optimal particle and evaluating the fitness value of each particle in each iteration, as well as updating the globally optimal particle, are as follows:

[0132] (1) Calculate the current iteration number Group diversity And set the quantum step size and particle swarm center The specific formula is as follows:

[0133] (75),

[0134] (76)

[0135] in: It is a very small positive number; and These represent the range of the quantum potential well contraction-expansion factor and the quantum step size, respectively. according to and The relative relationship between them is adaptively determined when As the quantum step size increases, This increases in size, thereby enhancing its global exploration capabilities;

[0136] (2) Sample quantum update parameters, i.e.: , as well as ;

[0137] (3) Calculate the quantum attractor to generate the reference point to which each particle in the particle swarm is pulled. The specific formula is as follows:

[0138] (77),

[0139] in: This is the element-wise multiplication operator;

[0140] (4) Update the position of each particle based on the quantum behavior update mechanism, that is:

[0141] (78),

[0142] in: It is a random symbol vector. for The update mechanism uses uniformly distributed random numbers within the interval. This mechanism makes the position update of each particle in the search space more random, thus helping to enhance the algorithm's global search capability.

[0143] (5) For each scenario Force the binary code block to be used within the time window of its corresponding extreme weather event. Set to 1;

[0144] (6) Decision matrix for binary variables Perform block repetition and threshold function decoding, and output the decoding results. The input is fed into the inner linear programming problem P2 and solved to obtain the fitness values ​​of each particle. ;

[0145] (7) Update the historical best for each particle, if the fitness value of the current particle at its new position is... Less than the particle's historical best fitness value Then update the optimal position of the particle. And update the optimal fitness value of the particle. ;

[0146] (8) After all particles have completed their individual optimal updates, select the particle with the best fitness as the current global optimal particle. And update the global optimal fitness. .

[0147] Furthermore, in step 5, after the iteration is completed, the binary variables of the system components are obtained by decoding the globally optimal particle positions, and the inner-layer linear programming problem is solved to obtain continuous decision-making for the electric-hydrogen mobile resource leasing. The specific steps are as follows:

[0148] (1) After step 4 is completed, the binary variables of the system components are obtained by decoding the position of the globally optimal particle, that is, the binary variable decision matrix of the globally optimal particle. After performing block repetition and threshold function decoding, the decoded result is: ;

[0149] (2) Fixed binary variable decision matrix after decoding The input is then used to solve the inner linear programming problem P2 to obtain the continuous decision-making process for electric-hydrogen mobility resource leasing, namely, the number of electric vehicles to be leased. and the number of leased hydrogen fuel cell vehicles .

[0150] Furthermore, in step 6, the continuous decision-making process for electric-hydrogen mobile resource leasing is rounded and fixed, and the inner-layer linear programming is solved again to output the final economic scheduling scheme. The specific steps are as follows:

[0151] (1) The obtained continuous decision variables for electric-hydrogen mobile resource leasing Rounding to the nearest integer yields the integer form of the decision variables for electric-hydrogen mobile resource leasing. ;

[0152] (2) Fixed binary variable decision matrix after decoding and the rounded-down decision variables for electric-hydrogen mobile resource leasing Solve the inner linear programming problem again;

[0153] (3) Output the economic scheduling scheme, including the binary variable decision matrix. Rounded-down decision variables for electric-hydrogen mobile resource leasing and the optimal target value .

[0154] Compared to existing technologies, the beneficial effects of this invention are as follows: This invention combines farthest-first traversal clustering with an improved quantum behavior particle swarm optimization algorithm to achieve multi-timescale risk-aware economic scheduling of hydrogen-containing building multi-energy systems under the condition of considering electric-hydrogen mobile resource leasing. Furthermore, the beneficial effects achieved by this invention are as follows:

[0155] (1) Compared with the prior art, the present invention considers the cross-timescale coupling characteristics of heterogeneous energy sources such as electricity, heat, and hydrogen, and realizes the synergistic optimization of fixed and mobile resources of electricity and hydrogen. Compared with the method of ignoring the leasing of mobile resources of electricity and hydrogen, the present invention performs better in terms of operating cost and conditional risk value;

[0156] (2) Compared with the traditional particle swarm optimization algorithm, this invention introduces block coding and knowledge-assisted hot start mechanism, combined with quantum behavior particle swarm algorithm, to achieve efficient solution on representative scenarios obtained by farthest priority traversal clustering, further reducing operating cost and condition risk value. Attached Figure Description

[0157] Figure 1 This is a flowchart of an economic dispatch method for hydrogen-containing building multi-energy systems under extreme weather events, as proposed in this invention.

[0158] Figure 2 This is a comparison chart of the optimal leasing quantity decision of electric-hydrogen mobile resources by the method of the present invention and the comparative scheme;

[0159] Figure 3 This is a comparison chart of the actual target values ​​of economic scheduling between the method of this invention and the comparative scheme;

[0160] Figure 4 This is a performance comparison chart of the method of the present invention and the case where the electric-hydrogen mobility resources are ignored. Detailed Implementation

[0161] The present invention will now be described in further detail with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely for illustrating the technical solutions of the present invention more clearly, and should not be construed as limiting the scope of protection of the present invention.

[0162] Example:

[0163] like Figure 1 As shown, the flowchart of the design of an economic dispatch method for hydrogen-containing building multi-energy system under extreme weather events provided by the present invention includes the following steps:

[0164] The established problem of risk perception and multi-timescale economic scheduling of hydrogen-containing building multi-energy systems considering electric-hydrogen mobile resources under extreme weather events includes constraints, decision variables, and objective functions.

[0165] First, the constraints mainly include multi-timescale decision-making models, photovoltaic power generation models, electric-hydrogen mobile resource leasing models, hydrogen energy storage system models, ground source heat pump models, electric boiler models, hot water tank models, energy balance models, generation of uncertain scenario sets models, operating cost models, and risk perception models.

[0166] Specifically, the multi-timescale decision-making model is as follows:

[0167] (1),

[0168] (2),

[0169] (3),

[0170] (4),

[0171] (5),

[0172] (6),

[0173] (7),

[0174] (8),

[0175] Wherein: Formulas (1)-(3) represent the electrical energy decision time slot, thermal energy decision time slot and hydrogen energy decision time slot models, respectively; , and These represent the sets of time slots for electrical energy decisions, thermal energy decisions, and hydrogen energy decisions, respectively. , and These represent the time slot indexes for electrical energy decisions, thermal energy decisions, and hydrogen energy decisions, respectively. , and The total electrical energy decision time slot, the total thermal energy decision time slot, and the total hydrogen energy decision time slot are respectively represented; Formulas (4) to (6) respectively represent the relationship between the electrical energy decision time slot and the thermal energy decision time slot, the relationship between the electrical energy decision time slot and the hydrogen energy decision time slot, and the relationship between the thermal energy decision time slot and the hydrogen energy decision time slot. Specifically, each thermal energy decision time slot contains two electrical energy decision time slots, each hydrogen energy decision time slot contains four electrical energy decision time slots, and each hydrogen energy decision time slot contains two thermal energy decision time slots. and These represent the normal operating slots and the operating slots during extreme weather events for hydrogen-containing building multi-energy systems, respectively; , as well as These represent the lengths of the electrical energy decision time slot, the thermal energy decision time slot, and the hydrogen energy decision time slot, respectively. The lengths of the electrical energy decision time slot, the thermal energy decision time slot, and the hydrogen energy decision time slot are set to 15 minutes, 30 minutes, and 60 minutes, respectively.

[0176] The photovoltaic power generation model is modeled as follows:

[0177] (9),

[0178] (10)

[0179] in: For the first Photovoltaic power reduction per electrical time slot; For the first Maximum photovoltaic power generation per electrical time slot; Actual output power of photovoltaic power in each time slot .

[0180] To enhance the resilience of the system under extreme weather events, operators of hydrogen-based multi-energy building systems can pre-lease mobile resources from the market to supply energy shortages. The model for their electricity-hydrogen mobile resource leasing is as follows:

[0181] (11),

[0182] (12)

[0183] (13)

[0184] (14)

[0185] (15)

[0186] (16)

[0187] (17)

[0188] (18)

[0189] (19)

[0190] (20)

[0191] (twenty one),

[0192] in: and These represent the number of electric vehicles rented on a one-time basis and the number of hydrogen fuel cell vehicles rented on a one-time basis, respectively. and These represent the maximum number of electric vehicles and the maximum number of hydrogen fuel cell vehicles that can be rented in a single transaction, respectively. Represents the set of positive integers; and The first The output power of the electric time slot tram and the first Output power of hydrogen fuel cell vehicle in one electrical time slot; and The first The number of trams powered by each time slot and the first The number of hydrogen fuel cell vehicles supplied with hydrogen per electrical time slot; and These are the maximum output power of electric vehicles and the maximum output power of hydrogen fuel cell vehicles, respectively. and They were respectively in the second The total electrical energy storage of all trams in each time slot and the first The total hydrogen storage capacity of all hydrogen fuel cell vehicles in each hydrogen time slot; and These represent the total electrical energy storage of all electric vehicles in the initial electric time slot and the total hydrogen storage of all hydrogen fuel cell vehicles in the initial hydrogen time slot, respectively. and These are the storage capacities for a single electric vehicle and a single hydrogen fuel cell vehicle, respectively. Indicates the first The thermal energy output of a hydrogen fuel cell vehicle in a thermal time slot; and These refer to the hydrogen-electric efficiency and the thermal-electric efficiency of hydrogen fuel cell vehicles, respectively. For the discharge efficiency of the tram, This refers to the heat recovery efficiency of hydrogen fuel cell vehicles.

[0193] The hydrogen energy storage system is modeled as follows:

[0194] (twenty two),

[0195] (twenty three),

[0196] (twenty four),

[0197] (25)

[0198] (26)

[0199] (27)

[0200] (28)

[0201] (29)

[0202] (30)

[0203] (31),

[0204] (32),

[0205] (33),

[0206] (34),

[0207] (35),

[0208] (36)

[0209] (37)

[0210] in: and The first The input power of the first time slot electrolytic cell and the first The input power of a hydrogen time-slot electrolyzer. This is the maximum input power of the electrolytic cell. This represents the maximum ramp rate of the electrolytic cell's power. For the first Hydrogen production per hydrogen time-slot electrolyzer, The electro-hydrogen conversion efficiency of the electrolyzer; Indicates the first Each hydrogen time slot is used to purchase hydrogen from the hydrogen market in the absence of extreme weather events. To maximize the amount of hydrogen purchased; For the first The input power of a single-slot fuel cell, This represents the maximum output power of the fuel cell. This represents the maximum ramp rate for fuel cell power. For the first Hydrogen consumption per hydrogen time slot fuel cell and These are the hydrogen-electric efficiency and thermal-electric efficiency of the fuel cell, respectively. For the heat recovery efficiency of fuel cells, For the first The thermal energy generated by the thermal time-slot fuel cell; Equations (32)-(34) describe why the electrolyzer and fuel cell cannot operate simultaneously. Using binary values, this avoids the low round-trip efficiency caused by short energy cycles ("electrical energy to hydrogen energy, then back to electrical energy or heat energy") within the same operating window. A strict mutual exclusion operating strategy is implemented. If the electrolyzer is not working, the cell will not work; conversely, if the cell is not working, the fuel cell will not work. It is a very large positive integer; For the first The storage capacity of each hydrogen time slot hydrogen tank This is the maximum storage capacity of the hydrogen tank. This indicates the relationship between the electrical time slot length and the hydrogen time slot length. and They represent the first time. In the hydrogen time slot, the first The output power of the fuel cell in the current time slot and the first In the hydrogen time slot, the first The input power of the electrolytic cell in each time slot; the inequality constraint described by formula (37) reflects a non-first-come, first-served operating logic, that is, in the first time slot... The first hydrogen time slot At the end of each power time slot, the cumulative fuel cell consumption must not exceed the available storage capacity. The hydrogen energy storage system adopts an hourly batch scheduling strategy for electrolyzers and a flexible 15-minute scheduling scheme for fuel cells. The former can stabilize the hydrogen production plan and reduce the complexity of the problem, while the latter can maintain a rapid response capability to adapt to fluctuations in photovoltaic output and load changes.

[0211] The ground source heat pump model is modeled as follows:

[0212] (38),

[0213] (39)

[0214] in: For the first Input power of the ground source heat pump in each time slot; This is the maximum input power of the ground source heat pump; The electro-thermal conversion efficiency of a ground source heat pump; For the first The output heat energy of a thermal time slot ground source heat pump.

[0215] The electric boiler model is modeled as follows:

[0216] (40)

[0217] (41),

[0218] in: Indicates the first Input power of the electric boiler in each time slot; This refers to the maximum input power of the electric boiler. The electro-thermal conversion efficiency of the electric boiler; The first The output heat energy of a single-time slot electric boiler.

[0219] The hot water tank model is modeled as follows:

[0220] (42),

[0221] (43),

[0222] (44),

[0223] (45)

[0224] (46)

[0225] in: For the first Storage capacity of each hot water tank in a hot time slot; This is the maximum storage capacity of the hot water tank. and These are the input efficiency and output efficiency of the hot water tank, respectively. For the heat retention rate of the hot water tank, and The first The input thermal power of the hot water tank in the first thermal time slot and the first Each hot water tank in a hot time slot outputs heat power. and These are the maximum input heat power and the maximum output heat power of the hot water tank, respectively. For the first The first hot-slot binary variable is used to avoid the second hot-slot binary variable. Each hot water tank in a heat time slot simultaneously charges / releases heat, when... At that time, the hot water tank is not allowed to release heat. At that time, it is not allowed to heat the hot water tank.

[0226] The energy balance model is as follows:

[0227] (47)

[0228] (48)

[0229] (49)

[0230] (50),

[0231] in: and The first Electrical time slots have priority Electrical loads and those with priority The amount of electricity load reduction, and The first The heat load of the first thermal time slot and the first Heat load reduction per thermal time slot.

[0232] The uncertainty scenario set generated includes uncertainties related to photovoltaic power generation, electrical load, heat load, and the duration of extreme weather events. The specific modeling method is as follows:

[0233] (51),

[0234] (52),

[0235] (53),

[0236] (54),

[0237] (55),

[0238] (56),

[0239] (57),

[0240] (58),

[0241] (59),

[0242] (60)

[0243] (61),

[0244] (62),

[0245] (63),

[0246] (64),

[0247] in: , and The first Predicted value of maximum output power of photovoltaic power in each time slot, the first The predicted value of the heat load in the first thermal time slot and the first Predicted values ​​of electrical load per electrical time slot; , and The scenarios are extreme events following a uniform distribution. Influencing factors of photovoltaic, heat load and electrical load; For the scene Next Photovoltaic power generation per electrical time slot; For the scene Next Heat load demand per thermal slot; For the scene Next The priority of each time slot is The demand for electrical load; For the first A fixed proportion of the demand for priority electrical loads in the total demand for electrical loads; It follows a normal distribution; , and For the scene Next Prediction error of maximum photovoltaic output power in individual time slots, and scenarios Next Prediction error of heat load under each thermal time slot and scenarios Next Prediction error of electrical load under individual electrical time slots; and These are the minimum and maximum standard deviations of the maximum output power of photovoltaic power, respectively. and These are the minimum and maximum standard deviations of the electrical load, respectively. and These are the minimum and maximum standard deviations of the heat load, respectively. and Scenes The start and end times of the following extreme events and The start and end times of the extreme event windows are known. and All are random discrete jitter quantities; To generate each scene The probability of.

[0248] The operating cost of this system comprises five components: photovoltaic load reduction cost, equipment depreciation cost, multi-source load reduction cost, hydrogen procurement cost, and mobile resource leasing cost. The operating cost model is as follows:

[0249] (65)

[0250] in: , , , as well as These are the photovoltaic cost reduction coefficient, the electrolyzer depreciation cost coefficient, the fuel cell depreciation cost coefficient, the electricity load reduction cost, and the heat load reduction cost coefficient, respectively. For the price of hydrogen; and These are the rental costs for a single electric vehicle and the rental costs for a single hydrogen fuel cell vehicle, respectively. This represents an index of large-scale extreme weather event scenarios, where: and Separate the scene set and the total number of scenes; , , , , and Scenes The following correspondence , , , , and The value.

[0251] The risk perception model is modeled as follows.

[0252] (66),

[0253] (67)

[0254] (68)

[0255] in: To calculate the highest Average operating costs; For when Confidence level is Risk value at time, auxiliary variable Used to measure each scenario Costs exceeding The part.

[0256] Secondly, the decision variables include the number of trolleys rented in a single transaction. Number of hydrogen fuel cell vehicles rented at one time , No. Actual output power of photovoltaic power in each time slot , No. The output power of the electric time-slot tram , No. The output power of a hydrogen fuel cell vehicle in a single time slot , No. Input power of a single-slot fuel cell , No. Input power of individual electric boiler time slot , No. The input power of the ground source heat pump in each time slot , No. The priority of each time slot is Electricity load reduction , No. The hot water tank input heat power in each hot time slot , No. Each hot water tank outputs heat power in a specific time slot. , No. Heat load reduction per thermal time slot , No. Input power of each hydrogen time-slot electrolyzer , No. Each hydrogen time slot purchases hydrogen from the hydrogen market in the absence of extreme weather events. .

[0257] Finally, the objective function is modeled as follows:

[0258] (69)

[0259] in: For the scene The set of all variables specifically includes the number of trams rented in a single transaction. Number of hydrogen fuel cell vehicles rented at one time , No. Input power of individual time-slot electrolytic cells , No. The output power of a single-time-slot fuel cell , No. The input power of the ground source heat pump in each time slot , No. Input power of individual electric boiler time slot , No. The output power of the electric time-slot tram , No. The output power of a hydrogen fuel cell vehicle in a single time slot , No. Photovoltaic power reduction in individual time slots , No. Electrical time slots have priority Electricity load reduction , No. Total electrical energy storage of all trams in each time slot , No. The output heat energy of the time-slot ground source heat pump , No. The output heat energy of the electric boiler with a time slot , No. The heat energy generated by a time-slot fuel cell , No. The heat energy output by a hydrogen fuel cell vehicle in a single time slot , No. The hot water tank input heat power in each hot time slot , No. Each hot water tank outputs heat power in a specific time slot. , No. Binary variables related to the operation of the hydrogen time slot and the electrolyzer and fuel cell , No. Heat load reduction per thermal time slot , No. Hot water tank storage capacity per hot time slot , No. Input power of each hydrogen time-slot electrolyzer , No. Binary variables related to the filling and discharging of hot water tanks in each hot time slot , No. Each hydrogen time slot purchases hydrogen from the hydrogen market in the absence of extreme weather events. , No. Hydrogen production per hydrogen time slot electrolyzer , No. Hydrogen consumption per hydrogen time slot fuel cell , No. Storage capacity of each hydrogen time slot hydrogen tank , No. Total hydrogen storage capacity of all hydrogen fuel cell vehicles in each hydrogen time slot ; express The weighting coefficients.

[0260] Furthermore, the steps in step 2, which employ the farthest point first traversal algorithm to reduce uncertain scenarios and extract representative scenarios and filter scenarios related to the duration of extreme weather, are as follows:

[0261] (1) Input the generated large-scale uncertainty scenario set and their corresponding probability distribution ,in: , , , and These are the weighting coefficients for photovoltaic power generation, electrical load, and heat load, respectively. Total electrical time slots for the operation of hydrogen-containing multi-energy systems;

[0262] (2) Each extreme weather event scenario The multi-timescale sequence data of photovoltaic power generation, electrical load, and heat load are first mapped to their corresponding feature vectors. Secondly, initialize a representative set of scenarios. That is, to calculate the feature vectors of each scene. The norm of the scene with the largest norm will be indexed. As the first representative scene and Then, for all scenarios and scene Constructing the vector distance matrix Finally, for all scenarios Initialize distance array ,in: Euclidean distance;

[0263] (3) Among the remaining scenes, select the representative scene set that is most relevant to the current scene. The farthest scene ,in: This is an index for a representative set of scenarios extracted from a set of large-scale uncertain scenarios. The number of representative scenes will be concentrated; subsequently, Add representative scene set Finally, update the distance array. Finally, repeat this process until a representative set of scenes is reached. The number of medium scenes equals the number of target representative scenes. ;

[0264] (4) Based on all scenarios With representative scene set Construct scene assignment vectors Then all scenes Assigned to the representative scene closest to it This leads to clustering to calculate each representative scenario. probability and obtain each scene start time of extreme weather events and stop time ;

[0265] (5) Output the representative scene set extracted Its corresponding probability and representative scenes Corresponding extreme event start and end time parameters and .

[0266] Furthermore, in step 2, the risk-aware economic scheduling problem after scenario reduction is decomposed into two stages: binary variable decision search and inner-layer continuous variable linear programming solution. The system components run the outer-layer decision related to binary variable optimization, including the scenario... Below are all binary variables related to the operation of the hydrogen time slot, electrolyzer, and fuel cell. and scene All binary variables related to the filling and discharging of hot water tanks in the following hot time slots Based on this, the inner-level linear programming problem related to continuous variable decision-making is established as follows:

[0267] (70)

[0268] (71),

[0269] (72),

[0270] (73),

[0271] (74),

[0272] in: Represents the set of real numbers; For the scene Next Each hydrogen time slot is a binary variable related to the operation of the electrolyzer and fuel cell; For the scene Next Binary variables related to the filling and discharging of a hot water tank in a specific thermal time slot; and These represent the start / stop state matrices for the electrolyzer / fuel cell and the start / stop state matrices for the hot water tank charging / discharging, respectively.

[0273] Furthermore, the specific steps for generating the particle swarm related to the binary variables of the system components and evaluating the fitness value of each particle to obtain the globally optimal particle are as follows:

[0274] (1) Let the particle swarm The number of particles is The original time-series electrolyzer / fuel cell binary variables and hot water tank charge / discharge variables are compressed into several consecutive blocks representing the start / stop modes of the equipment within corresponding time periods using block coding. and Their sizes are respectively and ;

[0275] (2) Let and These represent the number of hydrogen time slots in each electrolyzer / fuel cell start-stop block and the number of thermal time slots in each hot water tank charge / discharge start-stop block, respectively. The number of start-stop blocks is as follows: and This encoding method compresses the exploration space of each particle, avoiding unnecessary frequent device start-ups and shutdowns;

[0276] (3) The random generation size is Particle swarm ,in: , and This is achieved by adjusting the original particle binary block structure, i.e. and Their sizes are respectively and , Each particle represents two probability blocks concatenated row-by-row in the extreme weather event scenario; and the relevant binary encoded blocks within the extreme window of each scenario. Set to 1, which is used to operate the fuel cell only during extreme windows, consuming hydrogen to generate electricity and heat.

[0277] (4) For each particle, the probability block is repeatedly mapped to the full time domain according to the block length, i.e. and ;

[0278] (5) For each particle, a threshold function is used to obtain the corresponding binary start-stop matrix. and ,Right now .

[0279] (6) Decision matrix based on the outer binary variables of each particle The fitness value of each particle is obtained by solving P2 using the interior-point method. And save the optimal value for each individual particle. ;

[0280] (7) Obtain the globally optimal particle With group diversity .

[0281] The specific steps for setting up several iterations, updating the position of each particle using the globally optimal particle in each iteration, evaluating the fitness value of each particle, and updating the globally optimal particle are as follows:

[0282] (1) Calculate the current iteration number Group diversity And set the quantum step size and particle swarm center The specific formula is as follows:

[0283] (75),

[0284] (76)

[0285] in: It is a very small positive number; and These represent the range of the quantum potential well contraction-expansion factor and the quantum step size, respectively. according to and The relative relationship between them is adaptively determined when As the quantum step size increases, This increases in size, thereby enhancing its global exploration capabilities;

[0286] (2) Sample quantum update parameters, i.e.: , as well as ;

[0287] (3) Calculate the quantum attractor to generate the reference point to which each particle in the particle swarm is pulled. The specific formula is as follows:

[0288] (77),

[0289] in: This is the element-wise multiplication operator;

[0290] (4) Update the position of each particle based on the quantum behavior update mechanism, that is:

[0291] (78),

[0292] in: It is a random symbol vector. for The update mechanism uses uniformly distributed random numbers within the interval. This mechanism makes the position update of each particle in the search space more random, thus helping to enhance the algorithm's global search capability.

[0293] (5) For each scenario Force the binary code block to be used within the time window of its corresponding extreme weather event. Set to 1;

[0294] (6) Decision matrix for binary variables Perform block repetition and threshold function decoding, and output the decoding results. The input is fed into the inner linear programming problem P2 and solved to obtain the fitness values ​​of each particle. ;

[0295] (7) Update the historical best for each particle, if the fitness value of the current particle at its new position is... Less than the particle's historical best fitness value Then update the optimal position of the particle. And update the optimal fitness value of the particle. ;

[0296] (8) After all particles have completed their individual optimal updates, select the particle with the best fitness as the current global optimal particle. And update the global optimal fitness. .

[0297] After iteration, the optimal system component operation binary variables are obtained by decoding the globally optimal particle position, and the inner linear programming problem is solved to obtain continuous decision-making for electric-hydrogen mobile resource leasing. The specific steps are as follows:

[0298] (1) After step 4 is completed, the binary variables of the system components are obtained by decoding the position of the globally optimal particle, that is, the binary variable decision matrix of the globally optimal particle. After performing block repetition and threshold function decoding, the decoded result is: ;

[0299] (2) Fixed binary variable decision matrix after decoding The input is then used to solve the inner linear programming problem P2 to obtain the continuous decision-making process for electric-hydrogen mobility resource leasing, namely, the number of electric vehicles to be leased. and the number of leased hydrogen fuel cell vehicles .

[0300] Furthermore, after rounding down the continuous decision-making process for electric-hydrogen mobile resource leasing, the inner-layer linear programming is solved again to output the final economic scheduling scheme. The specific steps are as follows:

[0301] (1) The obtained continuous decision variables for electric-hydrogen mobile resource leasing Rounding to the nearest integer yields the integer form of the decision variables for electric-hydrogen mobile resource leasing. ;

[0302] (2) Fixed binary variable decision matrix after decoding and the rounded-down decision variables for electric-hydrogen mobile resource leasing Solve the inner linear programming problem again;

[0303] (3) Output the economic scheduling scheme, including the binary variable decision matrix. Rounded-down decision variables for electric-hydrogen mobile resource leasing and the optimal target value .

[0304] To demonstrate the effectiveness of the method of the present invention, four comparative schemes were introduced.

[0305] Compared with Scheme 1, which is based on the decomposition framework adopted in step 4 of this invention, the traditional particle swarm optimization algorithm is used to solve the multi-timescale economic scheduling problem of risk perception of hydrogen-containing building multi-energy system considering electric-hydrogen mobile resource leasing under extreme weather events.

[0306] In contrast, based on the decomposition framework adopted in step 4 of this invention, the second scheme uses the traditional quantum behavior particle swarm optimization algorithm to solve the multi-timescale economic scheduling problem of risk perception of hydrogen-containing building multi-energy system considering electric-hydrogen mobile resource leasing under extreme weather events.

[0307] Comparative scheme three, based on GAMS, directly solves the multi-timescale economic scheduling problem of hydrogen-containing building multi-energy systems considering electric-hydrogen mobile resource leasing under extreme weather events using a built-in commercial solver. This comparative scheme can serve as a reference for the performance upper limit of the method of this invention.

[0308] Compared with the method of this invention, Scheme 4 uses GAMS to directly solve the economic scheduling problem of hydrogen-containing building multi-energy system under extreme weather events, ignoring the leasing of electric-hydrogen mobile resources.

[0309] Figure 2 and Figure 3 This is a performance comparison chart of the method of the present invention and the three comparative schemes mentioned above. Figure 2This is a comparison chart of the optimal rental quantity decision for electric-hydrogen mobile resources between the method of this invention and the comparative scheme. The method proposed in this invention outperforms both comparative scheme 1 and comparative scheme 2 in terms of the rental quantity of electric vehicles and hydrogen fuel cell vehicles, and is closer to the results of comparative scheme 3. Specifically, compared to comparative scheme 1, the method of this invention reduces the rental quantity of electric vehicles by 20.20% and the rental quantity of hydrogen fuel cell vehicles by 59.39%; compared to comparative scheme 2, the rental quantity of electric vehicles is reduced by 20.20% and the rental quantity of hydrogen fuel cell vehicles is reduced by 50.75%; furthermore, the rental quantity of the method proposed in this invention is closer to that of comparative scheme 3. The above results show that the method of this invention can obtain a near-optimal electric-hydrogen mobile resource rental strategy.

[0310] Furthermore, Figure 2 The rounded value of the continuous lease decision is used to calculate the economic scheduling target value of each scheme. For example... Figure 3 The figure shown is a comparison of the actual target values ​​of economic scheduling between the method of this invention and the comparative scheme. From... Figure 3 As can be seen, the method proposed in this invention achieves lower target values ​​compared to Comparative Scheme 1 and Comparative Scheme 2, and is quite close to the results of Scheme 3. Specifically, compared to Comparative Scheme 1 and Comparative Scheme 2, the method of this invention can reduce the target value by 35.52% and 22.98%, respectively. Furthermore, the relative performance difference between the method proposed in this invention and Comparative Scheme 3 is 2.76%. These results demonstrate that the method proposed in this invention is effective in solving this economic scheduling problem.

[0311] like Figure 4 The figure shown is a performance comparison chart of the method of the present invention and the case where the electric-hydrogen mobility resources are ignored. From... Figure 4 As can be seen, the method proposed in this invention achieves superior performance in terms of expected cost, conditional risk, and target value compared to comparative scheme four. Specifically, compared to comparative scheme four, the method of this invention can reduce the expected cost, conditional risk, and target value by 57.16%, 82.95%, and 80.86%, respectively. These results demonstrate that the method proposed in this invention has significant advantages in improving system economics and risk perception regarding the consideration of electric-hydrogen mobile resources.

Claims

1. An economic dispatch method for multi-energy systems in hydrogen-containing buildings under extreme weather events, characterized in that, Includes the following steps: Step 1: Establish a multi-timescale economic dispatch problem for risk perception of hydrogen-containing buildings and multi-energy systems under extreme weather events, considering the leasing of electric and hydrogen mobile resources; Step 2: Use the farthest first traversal method to reduce the source load related scenarios and decompose the risk perception economic scheduling problem after scenario reduction into two stages: binary variable decision search and inner continuous variable linear programming solution. Step 3: Generate a particle swarm related to the binary variables of system components and evaluate the fitness value of each particle to obtain the globally optimal particle. The specific steps are as follows: (5.1) Let the particle swarm The number of particles is The original time-series electrolyzer / fuel cell binary variables and hot water tank charge / discharge variables are compressed into several consecutive blocks representing the start / stop modes of the equipment within corresponding time periods using block coding. and Their sizes are respectively and ; (5.2) Let and These represent the number of hydrogen time slots in each electrolyzer / fuel cell start-stop block and the number of thermal time slots in each hot water tank charge / discharge start-stop block, respectively. The number of start-stop blocks is as follows: and ; (5.3) The random generation size is Particle swarm ,in: , and This is achieved by adjusting the original particle binary block structure, i.e. and Their sizes are respectively and , This represents the dimension of two probability blocks concatenated row by row for each particle, representing extreme weather event scenarios; relevant blocks within the extreme window of each scenario are directly set to 1; (5.4) For each particle, the probability block is repeatedly mapped to the full time domain according to the block length, i.e. and ; (5.5) For each particle, a threshold function is used to obtain the corresponding binary start-stop matrix. and ,Right now ; (5.6) Decision matrix based on the outer binary variables of each particle The fitness value of each particle is obtained by solving P2 using the interior-point method. And save the optimal value for each individual particle. ; (5.7) Obtain the globally optimal particle With group diversity ; Step 4: Set up several iterations. In each iteration, use the global best particle to update the position of each particle, evaluate the fitness value of each particle, and update the global best particle. After the iterations of steps 5 and 4 are completed, the global optimal particle position is decoded to obtain the optimal system component running binary variable. The inner linear programming problem is solved by fixing this binary variable to obtain the continuous decision of electric-hydrogen mobile resource leasing. Step 6: After rounding the continuous decision of electric-hydrogen mobile resource leasing obtained in Step 5, solve the inner-layer linear programming again to output the final economic scheduling scheme.

2. The method according to claim 1, characterized in that, The risk perception multi-timescale economic dispatch problem of hydrogen-containing building multi-energy system considering electric-hydrogen mobile resources established in step 1 includes constraints, decision variables, and objective function: First, the constraints include multi-timescale decision-making models, photovoltaic power generation models, electric-hydrogen mobile resource leasing models, hydrogen energy storage system models, ground source heat pump models, electric boiler models, hot water tank models, energy balance models, generation of uncertain scenario sets models, operating cost models, and risk perception models. The multi-timescale decision-making model is as follows: （1）， （2）， （3）， （4）， （5）， （6）， （7）， （8）， Wherein: Formulas (1)-(3) represent the electrical energy decision time slot, thermal energy decision time slot and hydrogen energy decision time slot models, respectively; , and These represent the sets of time slots for electrical energy decisions, thermal energy decisions, and hydrogen energy decisions, respectively. , and These represent the time slot indexes for electrical energy decisions, thermal energy decisions, and hydrogen energy decisions, respectively. , and The total electrical energy decision time slot, the total thermal energy decision time slot, and the total hydrogen energy decision time slot are respectively represented; Formulas (4) to (6) respectively represent the relationship between the electrical energy decision time slot and the thermal energy decision time slot, the relationship between the electrical energy decision time slot and the hydrogen energy decision time slot, and the relationship between the thermal energy decision time slot and the hydrogen energy decision time slot; and These represent the normal operating slots and the operating slots during extreme weather events for hydrogen-containing building multi-energy systems, respectively. , as well as These represent the time slot lengths for electrical energy decision-making, thermal energy decision-making, and hydrogen energy decision-making, respectively. The photovoltaic power generation model is modeled as follows: （9）， （10）， in: For the first Photovoltaic power reduction per electrical time slot; For the first Maximum photovoltaic power generation per electrical time slot; For the first Actual output power of photovoltaic power in each electrical time slot; The electric-hydrogen mobile resource leasing model is modeled as follows: （11）， （12）， （13）， （14）， （15）， （16）， （17）， （18）， （19）， （20）， （21）， in: and These represent the number of electric vehicles rented on a one-time basis and the number of hydrogen fuel cell vehicles rented on a one-time basis, respectively. and These represent the maximum number of electric vehicles and the maximum number of hydrogen fuel cell vehicles that can be rented in a single transaction, respectively. Represents the set of positive integers; and The first The output power of the electric time slot tram and the first Output power of hydrogen fuel cell vehicle in one electrical time slot; and The first The number of trams powered by each time slot and the first The number of hydrogen fuel cell vehicles supplied with hydrogen per electrical time slot; and These are the maximum output power of electric vehicles and the maximum output power of hydrogen fuel cell vehicles, respectively. and They were respectively in the second The total electrical energy storage of all trams in each time slot and the first The total hydrogen storage capacity of all hydrogen fuel cell vehicles in each hydrogen time slot; and These represent the total electrical energy storage of all electric vehicles in the initial electric time slot and the total hydrogen storage of all hydrogen fuel cell vehicles in the initial hydrogen time slot, respectively. and These are the storage capacities for a single electric vehicle and a single hydrogen fuel cell vehicle, respectively. Indicates the first The thermal energy output of a hydrogen fuel cell vehicle in a thermal time slot; and These refer to the hydrogen-electric efficiency and the thermal-electric efficiency of hydrogen fuel cell vehicles, respectively. For the discharge efficiency of the tram, For the heat recovery efficiency of hydrogen fuel cell vehicles; The hydrogen energy storage system is modeled as follows: （22）， （23）， （24）， （25）， （26）， （27）， （28）， （29）， （30）， （31）， （32）， （33）， （34）， （35）， （36）， （37）， in: and The first The input power of the first time slot electrolytic cell and the first The input power of a hydrogen time-slot electrolyzer. This is the maximum input power of the electrolytic cell. This represents the maximum ramp rate of the electrolytic cell's power. For the first Hydrogen production per hydrogen time-slot electrolyzer, The electro-hydrogen conversion efficiency of the electrolyzer; Indicates the first Each hydrogen time slot is used to purchase hydrogen from the hydrogen market in the absence of extreme weather events. To maximize the amount of hydrogen purchased; For the first The input power of a single-slot fuel cell, This represents the maximum output power of the fuel cell. This represents the maximum ramp rate for fuel cell power. For the first Hydrogen consumption per hydrogen time slot fuel cell and These refer to the hydrogen-electric efficiency and the thermal-electric efficiency of a fuel cell, respectively. For the heat recovery efficiency of fuel cells, For the first The thermal energy generated by the thermal time-slot fuel cell; Equations (32)-(34) describe why the electrolyzer and fuel cell cannot operate simultaneously. It is a binary value. It is a very large positive integer; For the first The storage capacity of each hydrogen time slot hydrogen tank This is the maximum storage capacity of the hydrogen tank. This indicates the relationship between the electrical time slot length and the hydrogen time slot length. and They represent the first time. In the hydrogen time slot, the first The output power of the fuel cell in the current time slot and the first In the hydrogen time slot, the first Input power of the electrolytic cell in each time slot; The ground source heat pump model is modeled as follows: （38）， （39）， in: For the first Input power of the ground source heat pump in each time slot; This represents the maximum input power of the time-slot ground source heat pump. The electro-thermal conversion efficiency of a ground source heat pump; For the first The output heat energy of a thermal time slot ground source heat pump; The electric boiler model is modeled as follows: （40）， （41）， in: Indicates the first Input power of the electric boiler in each time slot; This refers to the maximum input power of the electric boiler. The electro-thermal conversion efficiency of electric boilers; The first The output heat energy of a single thermal time slot electric boiler; The hot water tank model is modeled as follows: （42）， （43）， （44）， （45）， （46）， in: For the first Storage capacity of each hot water tank in a hot time slot; This is the maximum storage capacity of the hot water tank. and These are the input efficiency and output efficiency of the hot water tank, respectively. For the heat retention rate of the hot water tank, and The first The input thermal power of the hot water tank in the first thermal time slot and the first Each hot water tank in a hot time slot outputs heat power. and These are the maximum input heat power and the maximum output heat power of the hot water tank, respectively. For the first One hot-slot binary variable; The energy balance model is as follows: （47）， （48）， （49）， （50）， in: and The first Electrical time slots have priority Electrical loads and those with priority The amount of electricity load reduction, and The first The heat load of the first thermal time slot and the first Heat load reduction per thermal time slot; The uncertainty scenario set generated includes uncertainties related to photovoltaic power generation, electrical load, heat load, and the duration of extreme weather events. The specific modeling method is as follows: （51）， （52）， （53）， （54）， （55）， （56）， （57）， （58）， （59）， （60）， （61）， （62）， （63）， （64）， in: , and The first Predicted value of maximum output power of photovoltaic power in each time slot, the first The predicted value of the heat load in the first thermal time slot and the first Predicted values ​​of electrical load per electrical time slot; , and The scenarios are extreme events following a uniform distribution. Influencing factors of photovoltaic, heat load and electrical load; For the scene Next Photovoltaic power generation per electrical time slot; For the scene Next Heat load demand per thermal slot; For the scene Next The priority of each time slot is The demand for electrical load; For the first A fixed proportion of the demand for priority electrical loads in the total demand for electrical loads; The symbol for the probability distribution; , and For the scene Next Prediction error of maximum photovoltaic output power in individual time slots, and scenarios Next Prediction error of heat load under each thermal time slot and scenarios Next Prediction error of electrical load under individual electrical time slots; and These are the minimum standard deviation of the maximum output power of photovoltaic power and the maximum standard deviation of the maximum output power of photovoltaic power, respectively. and These are the minimum standard deviation and the maximum standard deviation of the electrical load, respectively. and These are the minimum standard deviation and the maximum standard deviation of the heat load, respectively. and Scenes The start and end times of the following extreme events and The start and end times of the extreme event windows are known. and All are random discrete jitter quantities; To generate each scene The probability of; The operating cost model is modeled as follows: （65）， in: , , , as well as These are the photovoltaic cost reduction coefficient, the electrolyzer depreciation cost coefficient, the fuel cell depreciation cost coefficient, the electricity load reduction cost, and the heat load reduction cost coefficient, respectively. For the price of hydrogen; and These are the rental costs for a single electric vehicle and the rental costs for a single hydrogen fuel cell vehicle, respectively. This represents an index of large-scale extreme weather event scenarios, where: and Separate the scene set and the total number of scenes; , , , , and Scenes The following correspondence , , , , and The value; The risk perception model is modeled as follows: （66）， （67）， （68）， in: To calculate the highest Average operating costs; For when Confidence level is Risk value at time, auxiliary variable Used to measure each scenario Costs exceeding Part of it; Secondly, the decision variables include the number of trolleys rented in a single transaction. The number of hydrogen fuel cell vehicles rented at one time , No. Actual output power of photovoltaic power in each time slot , No. The output power of the electric time-slot tram , No. The output power of a hydrogen fuel cell vehicle with a single time slot , No. Input power of a single-slot fuel cell , No. Input power of individual electric boiler time slot , No. The input power of the ground source heat pump in each time slot , No. The priority of each time slot is Electricity load reduction , No. The hot water tank input heat power in each hot time slot , No. Each hot water tank outputs heat power in a specific time slot. , No. Heat load reduction per thermal time slot , No. Input power of each hydrogen time-slot electrolyzer , No. Each hydrogen time slot purchases hydrogen from the hydrogen market in the absence of extreme weather events. ; Finally, the objective function is modeled as follows: （69）， in: For the scene The set of all variables specifically includes the number of trams rented in a single transaction. The number of hydrogen fuel cell vehicles rented at one time , No. Input power of individual time-slot electrolytic cells , No. The output power of a single-time-slot fuel cell , No. The input power of the ground source heat pump in each time slot , No. Input power of individual electric boiler time slot , No. The output power of the electric time-slot tram , No. The output power of a hydrogen fuel cell vehicle in a single time slot , No. Photovoltaic power reduction in individual time slots , No. Electrical time slots have priority Electricity load reduction , No. Total electrical energy storage of all trams in each time slot , No. The output heat energy of the time-slot ground source heat pump , No. The output heat energy of the electric boiler with a time slot , No. The heat energy generated by a time-slot fuel cell , No. The thermal energy output of a hydrogen fuel cell vehicle in a single time slot , No. The hot water tank input heat power in each hot time slot , No. Each hot water tank outputs heat power in a specific time slot. , No. Binary variables related to the operation of the hydrogen time slot and the electrolyzer and fuel cell , No. Heat load reduction per thermal time slot , No. Hot water tank storage capacity per hot time slot , No. Input power of each hydrogen time-slot electrolyzer , No. Binary variables related to the filling and discharging of hot water tanks in each hot time slot , No. Each hydrogen time slot purchases hydrogen from the hydrogen market in the absence of extreme weather events. , No. Hydrogen production per hydrogen time slot electrolyzer , No. Hydrogen consumption per hydrogen time slot fuel cell , No. Storage capacity of each hydrogen time slot hydrogen tank , No. Total hydrogen storage capacity of all hydrogen fuel cell vehicles in each hydrogen time slot ; express The weighting coefficients.

3. The method according to claim 2, characterized in that, Step 2, which uses the farthest point first traversal algorithm to reduce uncertain scenarios and extract representative scenarios and filter scenarios related to the duration of extreme weather, is as follows: (3.1) Input generated large-scale uncertainty scene set and their corresponding probability distribution ,in: , , , and These are the weighting coefficients for photovoltaic power generation, electrical load, and heat load, respectively. Total electrical time slots for the operation of hydrogen-containing multi-energy systems; (3.2) Each extreme weather event scenario The multi-timescale sequence data of photovoltaic power generation, electrical load, and heat load are first mapped to their corresponding feature vectors. Secondly, initialize a representative set of scenarios. That is, to calculate the feature vectors of each scene. The norm of the scene with the largest norm will be indexed. As the first representative scene and Then, for all scenarios and scene Constructing the vector distance matrix Finally, for all scenarios Initialize distance array ,in: Euclidean distance; (3.3) Among the remaining scenes, select the representative scene set that is most relevant to the current scene set. The farthest scene ,in: This is an index for a representative set of scenarios extracted from a set of large-scale uncertain scenarios. The number of representative scenes will be concentrated; subsequently, Add representative scene set Finally, update the distance array. Finally, repeat this process until a representative set of scenes is reached. The number of medium scenes equals the number of target representative scenes. ; (3.4) Based on all scenarios With representative scene set Construct scene assignment vectors Then all scenes Assigned to the representative scene closest to it This leads to clustering to calculate each representative scenario. probability and obtain each scene start time of extreme weather events and stop time ; (3.5) Output the representative scene set extracted Its corresponding probability and representative scenes Corresponding extreme event start and end time parameters and .

4. The method according to claim 3, characterized in that, Step 2 decomposes the risk-aware economic scheduling problem after scenario reduction into two stages: binary variable decision search and inner continuous variable linear programming solution. The system components run the binary variable optimization related outer decision, including scenario optimization. Below are all binary variables related to the operation of the hydrogen time slot, electrolyzer, and fuel cell. and scene All binary variables related to the filling and discharging of hot water tanks in the following hot time slots Based on this, the inner-level linear programming problem related to continuous variable decision-making is established as follows: （70）， （71）， （72）， （73）， （74）， in: Represents the set of real numbers; For the scene Next Each hydrogen time slot is a binary variable related to the operation of the electrolyzer and fuel cell; For the scene Next Binary variables related to the filling and discharging of a hot water tank in a specific thermal time slot; and These represent the start / stop state matrices for the electrolyzer / fuel cell and the start / stop state matrices for the hot water tank charging / discharging, respectively.

5. The method according to claim 1, characterized in that, The specific steps in step 4 for updating the position of each particle and evaluating the fitness value of each particle in each iteration to save the optimal particle index are as follows: (6.1) Calculate the current iteration number Group diversity And set the quantum step size and particle swarm center The specific formula is as follows: （75）， （76）， in: It is a very small positive number. This represents the historical optimal point for each particle i. and These represent the ranges of the quantum potential well contraction-expansion factors; (6.2) Sample quantum update parameters, i.e.: , as well as ; (6.3) Calculate the quantum attractor to generate the reference point to which each particle in the particle swarm is pulled, using the following formula: （77）， in: This is the element-wise multiplication operator; (6.4) Update the position of each particle based on the quantum behavior update mechanism, that is: （78）， (6.5) For each scenario Force the binary code block to be used within the time window of its corresponding extreme weather event. Set to 1; (6.6) Decision matrix for binary variables Perform block repetition and threshold function decoding, and output the decoding results. The input is fed into the inner linear programming problem P2 and solved to obtain the fitness values ​​of each particle. ; (6.7) Update the historical best for each particle, if the fitness value of the current particle at its new position is... Less than the particle's historical best fitness value Then update the optimal position of the particle. And update the optimal fitness value of the particle. ; (6.8) After all particles have completed their individual optimal updates, select the particle with the best fitness as... And update the global optimal fitness. .

6. The method according to claim 5, characterized in that, In step 5, after the iteration is completed, the binary variables of the system components are obtained by decoding the globally optimal particle positions, and the inner-layer linear programming problem is solved to obtain continuous decision-making for the electric-hydrogen mobility resource leasing. The specific steps are as follows: (7.1) After step 5 iteration is completed, the binary variables of the system components are obtained by decoding the global optimal particle position, that is, the binary variable decision matrix of the global optimal particle. After performing block repetition and threshold function decoding, the decoded result is: ; (7.2) Fixed binary variable decision matrix after decoding The input is then used to solve the inner linear programming problem P2 to obtain the continuous decision-making for electric-hydrogen mobility resource leasing, i.e., the number of electric vehicles to be leased. and the number of leased hydrogen fuel cell vehicles .

7. The method according to claim 6, characterized in that, In step 6, the continuous decision-making process for electric-hydrogen mobile resource leasing is rounded and fixed, and the inner-layer linear programming is solved again to output the final economic scheduling scheme. The specific steps are as follows: (8.1) The obtained continuous decision variables for electric-hydrogen mobile resource leasing Rounding to the nearest integer yields the integer form of the decision variables for electric-hydrogen mobile resource leasing. ; (8.2) Fixed binary variable decision matrix after decoding and the rounded-down decision variables for electric-hydrogen mobile resource leasing Solve the inner linear programming problem again; (8.3) Output the economic scheduling scheme, including the binary variable decision matrix. Rounded-down decision variables for electric-hydrogen mobile resource leasing and the optimal target value .

Images