Two-stage energy management method for integrated energy system based on improved grey wolf algorithm

Abstract

The application is suitable for the technical field of integrated energy system optimization scheduling, and provides a two-stage energy management method for integrated energy system based on improved grey wolf algorithm, which firstly constructs an integrated energy system framework; secondly, the traditional grey wolf algorithm is improved by introducing Hammersley low difference sequence initialization strategy, golden sine strategy and dynamic reverse learning strategy; then, in the upper layer, the improved grey wolf algorithm is used to optimize the capacity configuration of each device with the minimum life cycle economic cost as the target; finally, in the lower layer, the CPLEX solver is used to carry out the intraday time-sharing energy scheduling optimization by taking the capacity optimized in the upper layer as the constraint and considering the flexible load and carbon trading. Through algorithm improvement and two-stage optimization framework, the application effectively solves the economic-low carbon collaborative optimization problem in the planning and operation of integrated energy system, and improves the overall energy efficiency and economy of the system.

Description

Technical Field

[0001] This invention belongs to the field of integrated energy system optimization and scheduling technology, and particularly relates to a two-stage energy management method for integrated energy systems based on an improved gray wolf algorithm. Background Technology

[0002] With the depletion of fossil fuels and the upgrading of environmental protection requirements, integrated energy systems (IES) have become the core development direction of energy supply systems due to their advantages of multi-energy flow coupling and complementarity and high energy utilization efficiency.

[0003] However, existing integrated energy system energy management technologies still have some problems: First, there is a disconnect between equipment capacity optimization and energy dispatching. Traditional solutions often conduct capacity configuration or output dispatching separately, and the capacity determination is not fully linked to subsequent operating conditions, resulting in a mismatch between equipment capacity and actual load demand and the characteristics of renewable energy output. This leads to insufficient dispatching flexibility and an inability to achieve optimal life-cycle cost. Second, traditional optimization algorithms have poor adaptability. Intelligent optimization algorithms such as the Grey Wolf algorithm and Particle Swarm Optimization algorithm generally suffer from poor initial population dispersion, weak global search capabilities, and a tendency to get trapped in local optima in the later stages when optimizing equipment capacity. This results in slow convergence speed and insufficient accuracy of the optimal solution, making it difficult to adapt to the multi-constraint and high-dimensional optimization needs of integrated energy systems. Third, the dispatching objectives are not comprehensive enough. Some studies only consider economic costs and do not incorporate flexible load response and carbon trading mechanisms into a unified dispatching framework. This fails to take into account both system economy and low carbon emissions, and does not meet the energy development needs under the "dual carbon" objective.

[0004] In existing technologies, a two-layer optimization model is constructed for integrated cooling, heating, and power (CCHP) energy systems, but the performance of the optimization algorithm has not been improved, resulting in limited solution accuracy and convergence speed. Other studies have incorporated carbon trading and flexible loads into the scheduling objectives, but have not established a comprehensive collaborative mechanism for "capacity optimization-energy scheduling," failing to achieve deep coupling between equipment configuration and operational output. Furthermore, the traditional Grey Wolf algorithm uses a random sequence for initialization, leading to population aggregation and decreased search capability in later iterations. When applied to capacity optimization of integrated energy systems, it often results in excessively long iteration cycles and significant deviations from the optimal capacity, making it difficult to meet the actual operational needs of engineering projects.

[0005] Therefore, there is an urgent need for an integrated energy system energy management solution that takes into account capacity and scheduling coordination, algorithm optimization performance, economy and low carbon goals, in order to solve the shortcomings of existing technologies such as low optimization accuracy, high cost, insufficient utilization of new energy sources and poor algorithm adaptability. Summary of the Invention

[0006] The purpose of this invention is to provide a two-stage energy management method for integrated energy systems based on an improved gray wolf algorithm, aiming to solve the problems mentioned in the background art.

[0007] The present invention is implemented as follows: a two-stage energy management method for integrated energy systems based on an improved gray wolf algorithm, comprising the following steps:

[0008] Step 1: Construct a comprehensive energy system framework, which integrates energy production equipment, energy conversion equipment, energy storage equipment, and flexible loads, and interconnects with the external power grid and gas grid;

[0009] Step 2: Construct an improved gray wolf algorithm. The improved gray wolf algorithm improves the traditional gray wolf algorithm through the following strategies: Hammersley low-difference sequence initialization strategy based on generating the initial population based on Hammersley sequences, golden sine strategy based on the golden ratio and sine function to optimize individual position updates, and dynamic reverse learning strategy to generate and select reverse learning individuals in the later stage of iteration.

[0010] Step 3: Upper-level equipment capacity optimization. With the goal of minimizing the economic cost of the entire system life cycle, the improved Grey Wolf algorithm is used to solve for the optimal capacity of the energy production equipment, energy conversion equipment and energy storage equipment under preset constraints.

[0011] Step 4: Lower-level energy dispatch optimization. With optimal capacity as the operational constraint, the CPLEX solver is used to minimize the total dispatch cost, including flexible load compensation cost and carbon trading cost. Under preset constraints, the time-sharing optimal output power of each device, flexible load dispatch scheme and carbon trading strategy are solved within the dispatch period.

[0012] Another objective of this invention is to provide a two-stage energy management system for integrated energy systems based on an improved gray wolf algorithm. The two-stage energy management method for integrated energy systems described above includes:

[0013] The system framework module is used to integrate energy production equipment, energy conversion equipment, energy storage equipment and flexible loads, establish connections with the external power grid and gas grid, and realize the collection and transmission of system operation-related data;

[0014] The algorithm improvement module is used to construct an improved Grey Wolf algorithm based on the Hammersley low-difference sequence initialization strategy, the golden sine strategy, and the dynamic reverse learning strategy.

[0015] The capacity optimization module is used to solve for the optimal capacity of each device by calling the improved Grey Wolf algorithm with the goal of minimizing the economic cost of the entire system life cycle.

[0016] The energy scheduling module is used to call the CPLEX solver with optimal capacity as the operating constraint, aiming to minimize the total scheduling cost including flexible load compensation cost and carbon trading cost, and solve for the time-sharing optimal output power and flexible load scheduling scheme of each device within the scheduling period.

[0017] Further technical solutions also include: a result output module, used to generate and output equipment capacity configuration reports, energy scheduling schemes, and cost analysis reports.

[0018] The two-stage energy management method for integrated energy systems based on the improved gray wolf algorithm provided in this invention has the following beneficial effects:

[0019] (1) Improved performance of the optimization algorithm: By integrating the Hammersley low-difference sequence initialization strategy, the golden sine strategy and the dynamic reverse learning strategy, the Grey Wolf algorithm is improved, which effectively enhances the global exploration capability, local development accuracy and convergence speed of the algorithm when solving complex system optimization problems, and provides a reliable tool for efficient optimization of the system.

[0020] (2) Good synergy between economy and low carbon: The two-stage optimization framework of "planning-operation" is adopted. The full life cycle cost is coordinated in the upper-level capacity planning, and the flexible load and carbon trading mechanism are introduced in the lower-level scheduling and operation. This realizes the synergistic optimization of long-term investment economy and short-term operation low carbon, and reduces the overall operating cost of the system.

[0021] (3) The model is highly practical: the optimization model fully considers the actual constraints such as multi-energy balance, equipment operation, power grid interaction and flexible load, and the output results are consistent with the actual engineering. The system has a clear modular design and can provide direct and reliable decision support for the planning, design and scheduling of integrated energy systems. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the integrated energy system structure;

[0023] Figure 2 A flowchart of a two-stage energy management method for an integrated energy system based on an improved gray wolf algorithm, provided in an embodiment of the present invention;

[0024] Figure 3 To improve the flowchart of the Grey Wolf algorithm;

[0025] Figure 4 Initialize the policy comparison graph (where a is the rand random policy and b is the two subgraphs of the Hammersley policy).

[0026] Figure 5 Flowchart for optimizing equipment capacity;

[0027] Figure 6Energy management flowchart;

[0028] Figure 7 Input a graph for summer data (where a is the electric heating load data, and b is the wind speed and solar radiation data);

[0029] Figure 8 Input a graph for winter data (where a is the electric heating load data, and b is the wind speed and solar radiation data);

[0030] Figure 9 The graph shows the convergence of the algorithm's iterations (where a represents summer and b represents winter).

[0031] Figure 10 This is a power balance diagram for the capacity optimization phase (where a represents summer and b represents winter).

[0032] Figure 11 This is a thermal power balance diagram for the capacity optimization phase (where a represents summer and b represents winter).

[0033] Figure 12 A comparison chart of power output and demand for new energy sources (where a represents summer and b represents winter).

[0034] Figure 13 The summer flexible load dispatch response diagram is shown (where a is the flexible electrical load response diagram and b is the flexible thermal load response diagram).

[0035] Figure 14 The diagram shows the response of flexible load dispatching in winter (where a is the response of flexible electrical load and b is the response of flexible thermal load).

[0036] Figure 15 The diagram shows the power balance during the summer energy dispatch phase (where a is the electrical power balance diagram and b is the thermal power balance diagram).

[0037] Figure 16 This is a power balance diagram for the winter energy dispatch phase (where a is the electrical power balance diagram and b is the thermal power balance diagram). Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0039] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0040] like Figure 1 As shown, a two-stage energy management method for an integrated energy system based on an improved gray wolf algorithm, provided by an embodiment of the present invention, includes the following steps:

[0041] Step 1: Construct a comprehensive energy system framework;

[0042] A comprehensive energy system framework is established, comprising four main categories of equipment: energy production, energy conversion, energy storage, and flexible loads. The system is interconnected with external power distribution networks and natural gas networks via power and gas supply networks, forming a multi-energy collaborative supply system. Energy production equipment includes photovoltaic (PV) power generation equipment and wind power generation equipment (WT); energy conversion equipment includes gas turbines (GT), gas boilers (GB), and waste heat boilers (HRB), with the latter used to recover waste heat generated during gas turbine operation and convert it into usable thermal energy; energy storage equipment includes batteries (BAT) and thermal storage tanks (TST); and flexible loads include load transfer, load reduction, and heat load reduction, which can participate in system energy balance regulation through scheduling strategies.

[0043] Step 2: Construct the Improved Gray Wolf Algorithm (IGWO);

[0044] To address the shortcomings of the traditional Grey Wolf algorithm, three major strategies are employed to improve its adaptability and optimize its performance.

[0045] The Hammersley low-discrepancy sequence initialization strategy generates an initial population based on Hammersley sequences. The population dimension is consistent with the capacity parameter dimension of the device to be optimized (corresponding to six major categories of devices: PV, WT, GT, GB, BAT, and TST). The population size is set to 30~50 (30 is preferred in this embodiment). This strategy can make the initial population evenly distributed in the solution space, avoid repeated sampling and clustered sampling, improve sampling efficiency and population diversity, and lay the foundation for subsequent iterative optimization.

[0046] The Golden Sine Strategy introduces an optimization and iterative update mechanism using the golden ratio and sine function to enhance the algorithm's global search capability. Its mathematical expression is:

[0047] (1)

[0048] In the formula, This represents the number of iterations. For the first The individual position is updated in the next iteration using the golden sine strategy; For the first Initial position of the individual in the next iteration; and A random number within the interval [0, 2π]. Determine the distance to move. Determine the direction of movement; and This is a value derived from the golden ratio. , ; For the first The optimal position of the individual in the next iteration of the population; For the first The next iteration randomly positions individuals in the population. This strategy expands the solution set coverage and prevents the algorithm from converging prematurely.

[0049] The dynamic back-learning strategy generates a back-learning individual for the current best individual in the later stages of each iteration. A greedy strategy is used to select the optimal solution, preventing the algorithm from getting trapped in local optima. The formula for generating the back-learning individual is:

[0050] (2)

[0051] In the formula, To learn the individual's location in reverse; and These are the upper and lower boundaries of the population location (corresponding to the reasonable engineering range of each device's capacity, such as a PV capacity range of 5000~20000kW); calculation and The fitness value is used to select individuals with better fitness to enter the next iteration, thereby improving the local search accuracy of the algorithm.

[0052] Step 3: Optimize the capacity of upper-layer devices;

[0053] With the goal of minimizing the economic cost over the entire system lifecycle, the optimal capacity of each device is determined using the IGWO algorithm, as detailed below:

[0054] Objective function:

[0055] (3)

[0056] In the formula, The total economic cost of the system throughout its entire lifecycle; For equipment investment costs; For equipment operation and maintenance costs; Fees for penalties related to pollutant emissions; For grid interaction costs; Cost of natural gas fuel; This includes the penalty cost for power imbalance. The formulas for calculating each item of cost are as follows:

[0057] Equipment investment cost:

[0058] (4)

[0059] In the formula, This refers to the equipment depreciation factor. , The discount rate is 0.06 to 0.1, and is preferably 0.08 in this embodiment. For the first Service life of this type of equipment; For the first Unit investment cost of this type of equipment; For the first Number of equipment of this type; For the first The optimal capacity for this type of device.

[0060] Equipment maintenance costs:

[0061] (5)

[0062] In the formula, For the first Equipment maintenance cost coefficient; For the first similar devices Always ready to contribute.

[0063] Pollutant emission penalty fees:

[0064] (6)

[0065] In the formula, For the first Penalty unit price for pollutants; , and The first unit power output of the gas turbine, gas boiler, and power distribution network are respectively Emissions of pollutants of this type; and They are respectively Real-time electrical power of gas turbine, thermal power of gas boiler; for Time system and grid transaction volume.

[0066] Grid interaction costs:

[0067] (7)

[0068] In the formula, for The unit price of electricity at any given time for The unit price of electricity at any given moment; for Purchase power at any time for Power sold at any time.

[0069] Natural gas fuel costs:

[0070] (8)

[0071] In the formula, for Real-time natural gas price per unit; and They are respectively The amount of natural gas consumed by gas turbines and gas boilers at all times.

[0072] Power imbalance penalty fee:

[0073] (9)

[0074] In the formula, for Penalty unit price for power imbalance at any moment; for Excess electrical power at all times for Constant power shortage; for Excess heat power at all times for The thermal power is always insufficient.

[0075] Constraints include equality constraints and inequality constraints to ensure that capacity optimization conforms to engineering practice.

[0076] Equality constraints:

[0077] Power balance constraints:

[0078] (10)

[0079] Thermal energy balance constraints:

[0080] (11)

[0081] In the formula, for Photovoltaic power output at all times for Wind power output at all times; for Constantly monitor battery discharge power. for Battery charging power at all times; for Constant electrical load; for The output thermal power of the gas turbine at any given time; for The output thermal power of the gas-fired boiler at all times; for The heat storage tank's heat release power at all times. for The constant charging power of the thermal storage tank; for Constant heat load; for The heat output of the waste heat boiler at all times.

[0082] Inequality constraints: These cover the output limits of various devices, ramping constraints, and grid trading restrictions, such as the upper limit of the electric power of gas turbines. Lower limit Battery charging and discharging power constraints , Grid trading power limit Lower limit wait.

[0083] Optimized output: The optimal capacity of PV, WT, GT, GB, BAT, and TST is obtained through iterative solution using the IGWO algorithm, serving as the operating limit of the equipment in the lower-level energy scheduling.

[0084] Step 4: Optimize lower-level energy scheduling;

[0085] Using the optimal capacity output from the upper layer as the operational constraint, the time-sharing output of each device is optimized based on the CPLEX solver, taking into account both economy and low carbon emissions, as detailed below:

[0086] Objective function:

[0087] (12)

[0088] In the formula, This represents the total cost of the energy dispatch phase. To compensate for costs related to flexible loads; This represents the carbon trading cost; the remaining components are consistent with the upper-level objective function.

[0089] The formulas for calculating the cost of each newly added item are as follows:

[0090] Flexible load compensation cost:

[0091] (13)

[0092] In the formula, The unit price for compensation of transferred electrical load; To reduce the unit price for electricity / heat load compensation; , and They are respectively Constantly shift electrical load, reduce electrical load, and reduce heat load.

[0093] Carbon trading costs:

[0094] (14)

[0095] In the formula, The price is the market price for carbon trading. For carbon emission quotas; This represents the total CO2 emissions of the system. , For the first Carbon emission coefficient of Taiwan equipment in energy transportation process For the first Carbon emission coefficient of equipment application process.

[0096] Constraints: Add flexible load and carbon emission constraints to the lower-level scheduling; other constraints are the same as those in the upper level.

[0097] Flexible load constraints:

[0098] (15)

[0099] In the formula, and These are the electrical load and thermal load after flexible loads participate in the scheduling process.

[0100] Carbon emission constraints:

[0101] (16)

[0102] In the formula, The floating threshold for carbon emissions (in this embodiment, we take...) 10% of the total.

[0103] Optimized output: The CPLEX solver is used to solve the problem and output the optimal 24-hour time-sharing power output of each device, the flexible load scheduling scheme, and the carbon trading strategy.

[0104] Another embodiment of the present invention provides a two-stage energy management system for integrated energy systems based on an improved gray wolf algorithm. Based on the above-mentioned two-stage energy management method for integrated energy systems, it includes:

[0105] The system framework module integrates energy production equipment, energy conversion equipment, energy storage equipment, and flexible loads, establishing interconnection channels with the power distribution network and natural gas network. It possesses data acquisition, transmission, and preprocessing functions. Acquired data includes: equipment parameters (investment cost, operation and maintenance cost, service life, emission coefficients, etc.), typical summer / winter daily / heat load data, solar irradiance and wind speed data, time-of-use electricity prices, natural gas prices, and carbon trading market prices. Data transmission accuracy is no less than 0.001 units to ensure the accuracy of optimization calculations.

[0106] The algorithm improvement module improves the traditional Grey Wolf algorithm based on the Hammersley low-difference sequence initialization strategy, the golden sine strategy, and the dynamic reverse learning strategy. It provides configurable interfaces for parameters such as population size, maximum number of iterations, and discount rate, and outputs the IGWO algorithm adapted to the optimization needs of integrated energy systems.

[0107] The capacity optimization module aims to minimize the economic cost of the system's entire lifecycle. It receives the IGWO algorithm output by the algorithm improvement module and the data transmitted by the system framework module. Based on the preset objective function and constraints, it solves the optimal capacity of each device through iterative calculation.

[0108] The energy scheduling module uses the optimal capacity as the operating limit, calls the CPLEX solver, and combines real-time system data to optimize flexible load scheduling and carbon trading strategies. Under the condition of meeting various constraints, it outputs the time-sharing optimal output power of each device to form an energy scheduling scheme.

[0109] The output module receives output data from the capacity optimization module and the energy scheduling module, and generates equipment capacity configuration reports, energy scheduling schemes, cost analysis reports, and algorithm performance analysis reports. The reports include key indicators such as summer / winter typical daily cost comparisons, renewable energy utilization rates, and carbon emissions, providing decision-making basis for engineering applications.

[0110] A case study was conducted using typical daily load data from summer / winter in northern my country. System equipment parameters, pollutant emission coefficients, time-of-use electricity prices, and natural gas prices are shown in the tables below. All parameters were determined based on actual engineering surveys. Equipment performance parameters are shown in Table 1, pollutant emission coefficients and penalty unit prices are shown in Table 2, and time-of-use electricity prices, natural gas prices, and carbon trading prices are shown in Table 3. In addition, equipment depreciation factors were set. The power generation efficiency of the gas turbine is 35%, the heating efficiency of the gas boiler is 80%, the charging and discharging efficiency of the battery is 90%, the charging and discharging efficiency of the heat storage tank is 90%, and the upper and lower limits of the SOC of the battery are set to 10% and 90% of the rated capacity, respectively.

[0111] Table 1 Equipment Performance Parameters

[0112]

[0113] Table 2 Pollutant Emission Coefficients and Penalty Unit Prices

[0114]

[0115] Table 3. Time-of-use electricity prices, natural gas prices, and carbon trading prices.

[0116]

[0117] The specific implementation steps are as follows:

[0118] Step 1: Data Acquisition and Preprocessing;

[0119] The system framework modules collect data on sunlight, wind speed, and electricity / heat load in northern my country during summer (typical days in July) and winter (typical days in January). Abnormal data is removed and interpolated to ensure data integrity and accuracy before being entered into the system database.

[0120] Step 2: Algorithm initialization configuration;

[0121] pass Figure 3 The algorithm improvement module shown generates the IGWO algorithm, with the population size set to 30, the maximum number of iterations to 100, and the discount rate set to... An initial population is generated based on Hammersley sequences, with population boundaries corresponding to reasonable ranges for each device's capacity engineering. Random points generated by the rand random generation strategy and the Hammersley low-difference sequence strategy are compared. Figure 4 As shown, compared to the rand random generation strategy, the Hammersley low-discrepancy sequence strategy has better initial population dispersion and universality, and the population individuals are more evenly distributed in the solution vector space, which is beneficial to the later solution process of the algorithm.

[0122] Step 3: Calculate the equipment depreciation factor;

[0123] According to the formula Based on the service life of each piece of equipment in Table 1, the depreciation factors for each piece of equipment are calculated as follows: PV is 0.1019, WT is 0.0937, GT is 0.0888, GB is 0.1019, BAT is 0.1231, and TST is 0.1019.

[0124] Step 4: Solve for upper-level capacity optimization;

[0125] The capacity optimization module calls the IGWO algorithm, substitutes the objective function and constraints, and follows... Figure 5 The equipment capacity optimization flowchart uses logical iterative calculations to obtain the optimal capacity of each device in summer and winter: Summer: PV 12038 kW, WT 14978 kW, GT 2949 kW, GB 4402 kW, BAT 10000 kW, TST 10000 kW; Winter: PV 20000 kW, WT 14240 kW, GT 12657 kW, GB 7000 kW, BAT 1000 kW, TST 9045 kW.

[0126] Step 5: Solve the lower-level energy scheduling problem;

[0127] The energy scheduling module, constrained by optimal capacity, calls the CPLEX solver and combines... Figure 6 The energy dispatch flowchart logic optimizes flexible load dispatch and carbon trading strategies, and outputs the time-sharing output of each device: in summer, PV is fully powered from 11 to 3 o'clock and sells electricity to the grid, with a peak power of 7490 kW; in winter, wind power is the main source of power supply, with power sold from 11 to 3 o'clock ranging from 1613 to 6198 kW.

[0128] Step 6: Result verification and analysis;

[0129] An analysis report is generated through the results output module. Comparing the traditional IES strategy with the IES-car strategy of this invention, the summer scheduling cost decreased from RMB 40,366.91 to RMB 25,392.81, and the winter scheduling cost decreased from RMB 30,583.34 to RMB 20,360.23. Comparing the carbon-free trading mechanism with the carbon trading mechanism of this invention, the summer carbon emissions decreased from 86,085.54 kg to 60,280.13 kg, and the winter carbon emissions decreased from 60,272.24 kg to 43,941.68 kg, verifying the effectiveness and advantages of this invention.

[0130] Implementation results: Based on Figure 1 and Figure 2 In addition to the equipment parameters and price data in Tables 1-3, this embodiment selects actual electric heating load data, light intensity, and temperature data from northern my country during summer (typical days in July) and winter (typical days in January) for verification (e.g. Figure 7 and Figure 8 As shown in the figure, all test conditions are consistent with actual engineering scenarios, and the verification results are as follows:

[0131] Improved algorithm performance verification: by Figure 9 It can be seen that the improved gray wolf algorithm (IGWO) significantly improves both convergence speed and solution accuracy compared to the traditional gray wolf algorithm (GWO). For example... Figure 9 As shown in (a): Under summer conditions, the initial iteration cost of the GWO algorithm is 1.87 × 10⁻⁶. 8 The fitness value gradually stabilized after about 40-50 iterations, with the optimal fitness value corresponding to a system cost of 1.5 × 10⁻⁶. 7 The initial iteration cost of the IGWO algorithm is 2.29 × 10⁻⁶. 7 Furthermore, it rapidly decreases and approaches the global optimum within the first 3-5 iterations, corresponding to a cost of 1.49 × 10⁻⁶. 7 Yuan; as Figure 9 As shown in (b): Under winter conditions, the GWO algorithm tends to the optimal solution when the number of convergence iterations is 70, while the IGWO algorithm tends to the optimal solution on the 23rd iteration, with a convergence speed improvement of about 67.14%. This effectively solves the problem of poor initial population dispersion and easy trapping in local optima in traditional algorithms.

[0132] Power balance results during capacity optimization phase: From Figure 10 It can be seen that in summer, wind power accounts for a relatively small proportion, and new energy power generation mainly relies on photovoltaic power generation. As sunlight intensifies, the output power of photovoltaic power increases significantly between 11:00 and 14:00, reaching a peak of 14010 kW at 13:00, which is sufficient to support the electricity load demand during this period. As a result, the system is in a state of surplus power between 10:00 and 14:00, and the power sold to the main grid is as high as 7000 kW or more. When photovoltaic power generation is insufficient, the system needs to purchase electricity from the grid to meet the load demand, mainly concentrated between 1:00 and 7:00 and after 16:00. In winter, the output of wind power generation is stable at 2825-8022 kW, undertaking the core power supply task. The system sells electricity to the grid mainly at 5:00 and 12:00-14:00, and the amount of electricity purchased decreases significantly.

[0133] The results of the thermal power balance during the capacity optimization phase: Figure 11 It can be seen that in summer, the heat load is mainly met by the heat generated by the gas turbine, with the excess heat energy mainly concentrated between 5 PM and 9 PM and stored in the heat storage tank. The heat charge between 5 PM and 7 PM exceeds 1000 kW. During periods of sufficient sunlight, the heat load between 10 AM and 3 PM is met by the heat energy output from the gas boiler. In winter, the gas turbine mainly operates between 7 AM and 10 AM, 5 PM and 8 PM, and 11 PM and 12 AM. The excess heat generated by the gas turbine between 5 PM and 7 PM is stored in the heat storage equipment, and the remaining heat energy demand is mainly met by the gas boiler.

[0134] New energy output results: by Figure 12 It can be seen that in summer, the output power of photovoltaic power generation is concentrated between 7:00 and 19:00, while the output power of wind power generation ranges from 1001 to 8100 kW. Between 10:00 and 14:00, the total output power of wind power generation and photovoltaic power generation exceeds the load demand. In winter, wind power generation undertakes the main power supply task. The time when wind power generation can meet the load demand is between 1:00 and 5:00. The output power of photovoltaic power generation is distributed between 9:00 and 17:00, and the output power reaches a peak of 7479 kW at 13:00.

[0135] Flexible load dispatching effect: After flexible loads participate in dispatching, a significant "peak shaving and valley filling" effect is achieved, such as... Figure 13 and 14 As shown, the summer electricity load reduction range is 820~1300 kW, with the highest transfer amount at 9 o'clock being 1950 kW, and the maximum heat load reduction at 5 o'clock being 810 kW; the winter electricity load reduction range is 630~1100 kW, with the highest transfer amount at 13 o'clock being 880 kW, and the maximum heat load reduction at 8 o'clock being 1500 kW. The reduction amount is positively correlated with the real-time load. After adjustment, the peak-valley load difference of the system is reduced, the time-sharing output coordination of core equipment is significantly improved, and the source-load interaction efficiency is optimized.

[0136] Power balance results during the summer energy dispatch phase: as follows Figure 15 As shown, the system's electrical and thermal loads are mainly met by the combined efforts of gas turbines and new energy equipment. Regarding power dispatch: from 8 AM to 4 PM, the gas turbines and new energy output alone can cover all electricity demand. From 11 AM to 3 PM, the system sells surplus electricity to the main grid, reaching a maximum of 7490 kW. At other times, the system is mostly in a power purchase state, with the purchased power fluctuating between 149 and 4533 kW. Simultaneously, batteries are charged in the early morning and discharged during the evening peak to assist in peak shaving and valley filling. Regarding thermal dispatch: the gas turbines undertake the vast majority of basic heating tasks, while the gas boilers are only briefly activated in the early morning and morning to supplement heating, with the maximum heat release time at 5 AM at 1578 kW. The thermal storage tanks only operate in a charging state at 5 PM and 6 PM, with the charging power reaching 1123 kW at 6 PM.

[0137] Power balance results during the winter energy dispatch phase: as follows Figure 16 As shown, during a typical winter day, the system's electrical and thermal loads are mainly met by wind power generation and gas turbines working together. In terms of power dispatch: wind power is the main source of supply, with output power often exceeding 4000kW. Photovoltaic power generation has a higher output power at midday. The remaining load demand is mainly met by gas turbines, with gas turbine output power exceeding 2000kW most of the time. The system sells a large amount of electricity to the grid between 11:00 and 15:00, with the power sold fluctuating between 1613 and 6198 kW. Batteries are flexibly charged and discharged according to the surplus and deficit of power source and load to assist in regulation. In terms of thermal dispatch: gas turbines have the highest output thermal power, reaching 4000kW most of the time. The output of gas boilers is mainly concentrated between 1:00 and 8:00, with an output power reaching 5950 kW at 5:00. The thermal storage tank operates in a charging state at 12:00, 14:00, and 18:00-19:00, with a maximum charging power not exceeding 2000kW.

[0138] Comparison of Flexible Load Strategies: Table 4 shows the impact of the IES strategy (the system's energy dispatch method without considering flexible scheduling of thermal and electrical loads) and the IES-car strategy proposed in this invention (the system's energy dispatch method considering flexible scheduling of thermal and electrical loads) on various economic costs in the lower-level energy dispatch operation. As shown in Table 4, compared to the IES strategy, the IES-car strategy saves 37.10% and 33.43% in energy dispatch costs in summer and winter, respectively. In summer and winter, compared to the IES strategy, the IES-car strategy reduces operation and maintenance costs by 1.38% and 1.80%, electricity purchase costs by 66.14% and 91.89%, electricity sales revenue by 16% and 35.73%, fuel costs by 11.95% and 14.39%, respectively, and the compensation costs generated by flexible loads are RMB 3958.55 and RMB 3591.46, respectively. Therefore, it can be concluded that after considering the participation of flexible loads in energy dispatch, the total energy dispatch cost of the integrated energy system is significantly reduced. Of the various cost components, all costs except for the compensation costs added due to dispatch have decreased, and the system's electricity sales revenue has been significantly improved.

[0139] Table 4 Comparison of Integrated Energy System Operation Strategies

[0140]

[0141] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A two-stage energy management method for integrated energy systems based on an improved gray wolf algorithm, characterized in that, Includes the following steps: Step 1: Construct a comprehensive energy system framework, which integrates energy production equipment, energy conversion equipment, energy storage equipment, and flexible loads, and interconnects with the external power grid and gas grid; Step 2: Construct an improved gray wolf algorithm. The improved gray wolf algorithm improves the traditional gray wolf algorithm through the following strategies: Hammersley low-difference sequence initialization strategy based on generating the initial population based on Hammersley sequences, golden sine strategy based on the golden ratio and sine function to optimize individual position updates, and dynamic reverse learning strategy to generate and select reverse learning individuals in the later stage of iteration. Step 3: Upper-level equipment capacity optimization. With the goal of minimizing the economic cost of the entire system life cycle, the improved Grey Wolf algorithm is used to solve for the optimal capacity of the energy production equipment, energy conversion equipment and energy storage equipment under preset constraints. Step 4: Lower-level energy dispatch optimization. With optimal capacity as the operational constraint, the CPLEX solver is used to minimize the total dispatch cost, including flexible load compensation cost and carbon trading cost. Under preset constraints, the time-sharing optimal output power of each device, flexible load dispatch scheme and carbon trading strategy are solved within the dispatch period.

2. The two-stage energy management method for integrated energy systems based on the improved gray wolf algorithm according to claim 1, characterized in that, In step 1, the energy production equipment includes photovoltaic power generation equipment and wind power generation equipment; the energy conversion equipment includes gas turbines, gas boilers and waste heat boilers, with the waste heat boilers used to recover the waste heat generated during the operation of the gas turbines and convert it into usable thermal energy; the energy storage equipment includes batteries and thermal storage tanks; and the flexible loads include transferring electrical loads, reducing electrical loads and reducing thermal loads, participating in the system energy balance regulation through scheduling strategies.

3. The two-stage energy management method for integrated energy systems based on the improved gray wolf algorithm according to claim 1, characterized in that, In step 2, the Hammersley low-discrepancy sequence initialization strategy generates an initial population with the same dimension as the capacity parameter of the device to be optimized, and the population size is set to 30 to 50.

4. The two-stage energy management method for integrated energy systems based on the improved gray wolf algorithm according to claim 3, characterized in that, In step 2, the golden sine strategy introduces an optimization and iterative update mechanism using the golden section coefficient and sine function, the mathematical expression of which is: （1） In the formula, This represents the number of iterations. For the first The individual position is updated in the next iteration using the golden sine strategy; For the first Initial position of the individual in the next iteration; and A random number within the interval [0, 2π]. Determine the distance to move. Determine the direction of movement; and This is a value derived from the golden ratio. , ; For the first The optimal position of the population in the next iteration; For the first The random position of an individual in the population during the next iteration.

5. The two-stage energy management method for integrated energy systems based on the improved gray wolf algorithm according to claim 4, characterized in that, In step 2, the dynamic back-learning strategy generates a back-learning individual for the current best individual in the later stages of each iteration, and selects the optimal solution using a greedy strategy; the formula for generating the back-learning individual is: （2） In the formula, To learn the individual's location in reverse; and These represent the upper and lower boundaries of the population location; calculate and The fitness value is used to select individuals with better fitness to proceed to the next iteration.

6. The two-stage energy management method for integrated energy systems based on the improved gray wolf algorithm according to claim 5, characterized in that, The specific steps of step 3 are as follows: With the goal of minimizing the economic cost over the entire system lifecycle, an improved Grey Wolf algorithm is used to solve for the optimal capacity of each device, as detailed below: Objective function: （3） In the formula, The total economic cost of the system throughout its entire lifecycle; For equipment investment costs; For equipment operation and maintenance costs; Fees for penalties related to pollutant emissions; For grid interaction costs; Cost of natural gas fuel; The cost is a penalty for power imbalance; the formulas for calculating the cost of each item are as follows: Equipment investment cost: （4） In the formula, This is the equipment depreciation factor. ; The discount rate; For the first Service life of this type of equipment; For the first Unit investment cost of this type of equipment; For the first Number of devices of this type; For the first Optimal capacity for this type of equipment; Equipment maintenance costs: （5） In the formula, For the first Equipment maintenance cost coefficient; For the first similar devices Always put in the effort; Pollutant emission penalty fees: （6） In the formula, For the first Penalty unit price for pollutants; , and The first unit power output of the gas turbine, gas boiler, and power distribution network are respectively Emissions of pollutants of this type; and They are respectively Real-time electrical power of gas turbine, thermal power of gas boiler; for Time system and grid transaction volume; Grid interaction costs: （7） In the formula, for The unit price of electricity at any given time for The unit price of electricity at any given moment; for Purchase power at any time for Real-time electricity sales capacity; Natural gas fuel costs: （8） In the formula, for Real-time natural gas price per unit; and They are respectively The natural gas consumption of gas turbines and gas boilers at all times; Power imbalance penalty fee: （9） In the formula, for Penalty unit price for power imbalance at any moment; for Excess electrical power at all times for Constant power shortage; for Excess heat power at all times for Constantly insufficient heat power; Constraints include equality constraints and inequality constraints; Equality constraints: Power balance constraints: （10） Thermal energy balance constraints: （11） In the formula, for Photovoltaic power output at all times for Wind power output at all times; for Constantly monitor battery discharge power. for Battery charging power at all times; for Constant electrical load; for The output thermal power of the gas turbine at any given time; for The output thermal power of the gas-fired boiler at all times; for The heat storage tank's heat release power at all times. for The constant charging power of the thermal storage tank; for Constant heat load; for The heat output power of the waste heat boiler at any given time; Inequality constraints include the output limits of each device, ramping constraints, and grid trading restrictions; Optimized output: By iteratively solving the improved Grey Wolf algorithm, the optimal capacity of PV, WT, GT, GB, BAT, and TST is output as the operating limit of the equipment in the lower-level energy scheduling.

7. The two-stage energy management method for integrated energy systems based on the improved gray wolf algorithm according to claim 6, characterized in that, The specific steps of step 4 are as follows: Using the optimal capacity output from step 3 as the operational constraint, the time-sharing output of each device is optimized based on the CPLEX solver, as follows: Objective function: （12） In the formula, This represents the total cost of the energy dispatch phase. To compensate for costs related to flexible loads; For carbon trading costs; The formulas for calculating the cost of each newly added item are as follows: Flexible load compensation cost: （13） In the formula, The unit price for compensation of transferred electrical load; To reduce the unit price for electricity / heat load compensation; , and They are respectively Constantly shift electrical load, reduce electrical load, and reduce heat load; Carbon trading costs: （14） In the formula, The price is the market price for carbon trading. For carbon emission quotas; This represents the total CO2 emissions of the system. ; For the first Carbon emission coefficient of Taiwan equipment in energy transportation process For the first Carbon emission coefficient of equipment application process; Constraints: Add flexible load and carbon emission constraints to the lower-level scheduling; Flexible load constraints: （15） In the formula, and These are the electrical load and thermal load after flexible loads participate in dispatching; Carbon emission constraints: （16） In the formula, The threshold for floating carbon emissions; Optimized output: The CPLEX solver is used to solve the problem and output the optimal 24-hour time-sharing power output of each device, the flexible load scheduling scheme, and the carbon trading strategy.

8. A two-stage energy management system for integrated energy systems based on an improved gray wolf algorithm, wherein the two-stage energy management method for integrated energy systems based on an improved gray wolf algorithm as described in any one of claims 1-7 is characterized in that, include: The system framework module is used to integrate energy production equipment, energy conversion equipment, energy storage equipment and flexible loads, establish connections with external power grids and gas grids, and realize the collection and transmission of system operation-related data; The algorithm improvement module is used to build an improved Grey Wolf algorithm based on the Hammersley low-difference sequence initialization strategy, the golden sine strategy, and the dynamic reverse learning strategy. The capacity optimization module is used to solve for the optimal capacity of each device by calling the improved Grey Wolf algorithm with the goal of minimizing the economic cost of the entire system life cycle. The energy scheduling module is used to call the CPLEX solver with optimal capacity as the operating constraint, aiming to minimize the total scheduling cost including flexible load compensation cost and carbon trading cost, and solve for the time-sharing optimal output power and flexible load scheduling scheme of each device within the scheduling period.

9. The two-stage energy management system for integrated energy systems based on the improved gray wolf algorithm according to claim 8, characterized in that, Also includes: The results output module is used to generate and output equipment capacity configuration reports, energy scheduling schemes, and cost analysis reports.

Images