A collaborative optimization scheduling method and device for an integrated energy system and a computer device

Abstract

This application relates to a method, apparatus, and computer device for collaborative optimization scheduling of integrated energy systems. The method includes: upon receiving a collaborative optimization scheduling request for an integrated energy system, acquiring multi-source heterogeneous data; constructing a coupling mechanism model based on the multi-source heterogeneous data; solving the coupling mechanism model using a multi-agent reinforcement learning algorithm to obtain a collaborative optimization scheduling strategy; generating a collaborative optimization scheduling scheme through a three-level scheduling mechanism, and optimizing the collaborative optimization scheduling scheme to obtain a target collaborative optimization scheduling scheme. The scheme of this application overcomes the optimization lag problem by solving the coupling mechanism model using a multi-agent reinforcement learning algorithm, executes the corresponding collaborative optimization scheduling scheme through a three-level scheduling mechanism, and optimizes the collaborative optimization scheduling scheme during execution, thereby improving the collaborative optimization scheduling effect for integrated energy systems.

Description

Technical Field

[0001] This application relates to the field of new energy technology, and in particular to a collaborative optimization scheduling method, apparatus and computer equipment for integrated energy systems. Background Technology

[0002] Driven by the "dual carbon" goals and energy transition, the combined cooling, heating, and power (CCHP) system has become a core solution for improving energy efficiency in industrial parks and settings due to its ability to synergistically supply multiple energy sources, including electricity, heat, and cooling. This highly integrated, multi-energy complementary system can simultaneously provide ten forms of energy or services, achieving efficient, clean, low-carbon, and intelligent energy utilization. It represents a significant expansion in scale, dimensions, and function of traditional combined cooling, heating, and power (CCHP) technology, signifying the future direction of smart energy systems. However, its complex multi-energy coupling relationships necessitate consideration of electricity cost control, carbon emission reduction, and carbon tax compliance. Furthermore, it faces challenges such as curtailment of renewable energy output due to fluctuations, and a high proportion of plant power consumption. Achieving coordinated scheduling of electricity, carbon, and tax to balance economic efficiency and low carbon emissions remains a critical challenge for the industry.

[0003] However, current optimization methods for coordinated scheduling of integrated energy systems have poor adaptability, rely on model-driven algorithms such as mixed-integer programming, require precise system models and parameters, and are prone to optimization lag when facing uncertainties. This affects the optimization effect of the coordinated scheduling process. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for the coordinated optimization scheduling of integrated energy systems, which can improve the effect of coordinated optimization scheduling of integrated energy systems, in order to address the above-mentioned technical problems.

[0005] Firstly, this application provides a collaborative optimization scheduling method for integrated energy systems, including:

[0006] Upon receiving a collaborative optimization scheduling request for the integrated energy system, acquire the multi-source heterogeneous data of the integrated energy system;

[0007] A coupling mechanism model of the integrated energy system is constructed based on the multi-source heterogeneous data.

[0008] The coupling mechanism model is solved by a multi-agent reinforcement learning algorithm to obtain the collaborative optimization scheduling strategy of the integrated energy system.

[0009] A three-level scheduling mechanism is used to generate a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy. During the execution of the collaborative optimization scheduling scheme, the scheme is optimized to obtain the collaborative optimization scheduling result.

[0010] In one embodiment, acquiring the multi-source heterogeneous data of the integrated energy system includes:

[0011] Obtain energy data, equipment data, and system-related data corresponding to the integrated energy system to obtain multi-source data;

[0012] The multi-source data is subjected to anomaly cleaning processing to obtain multi-source completed data;

[0013] The multi-source complete data is subjected to data normalization processing to obtain multi-source heterogeneous data.

[0014] In one embodiment, constructing the coupling mechanism model of the integrated energy system based on the multi-source heterogeneous data includes:

[0015] Identify the energy conversion devices in the integrated energy system and construct the corresponding energy conversion sub-models for the energy conversion devices;

[0016] The formula for determining the carbon emissions of the integrated energy system is based on the multi-source heterogeneous data.

[0017] Construct a carbon tax cost sub-model for the integrated energy system based on the carbon emission formula;

[0018] Based on the energy conversion sub-model and the carbon tax cost sub-model, a coupling mechanism model of the integrated energy system is constructed.

[0019] In one embodiment, the step of solving the coupling mechanism model using a multi-agent reinforcement learning algorithm to obtain the collaborative optimization scheduling strategy for the integrated energy system includes:

[0020] Based on the carbon tax cost sub-model, an objective function is constructed with the goal of minimizing the total daily operating cost of the integrated energy system, and a solution constraint is constructed based on the energy conversion sub-model.

[0021] Construct a solution state space and a solution action space based on the multi-source heterogeneous data;

[0022] Based on the objective function, the constraints, the state space, and the action space, the coupled mechanism model is solved by multi-agents using a multi-agent reinforcement learning algorithm to obtain the collaborative optimization scheduling strategy of the integrated energy system.

[0023] In one embodiment, the method further includes:

[0024] Construct a multi-agent system corresponding to the integrated energy system, wherein the multi-agent system includes an energy supply agent, a load demand agent, a carbon management agent, and an energy storage scheduling agent;

[0025] Obtain the historical operating data corresponding to the integrated energy system;

[0026] Intelligent agent training data is constructed based on the historical operational data;

[0027] The multi-agent model is obtained by training the multi-agent model using the training data of the aforementioned agents, through a parameter sharing mechanism and a cyclic time network.

[0028] In one embodiment, the step of generating a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy through a three-level scheduling mechanism, and optimizing the collaborative optimization scheduling scheme during its execution to obtain a collaborative optimization scheduling result includes:

[0029] During the day-ahead phase of the three-level scheduling mechanism, a collaborative optimization scheduling scheme corresponding to the aforementioned collaborative optimization scheduling strategy is generated;

[0030] During the intraday phase of the three-level dispatch mechanism, the corresponding real-time system data of the integrated energy system is obtained, and the equipment output strategy and energy storage strategy of the collaborative optimization dispatch scheme are adjusted according to the real-time system data to obtain the intraday collaborative optimization dispatch scheme.

[0031] During the real-time phase of the three-level scheduling mechanism, the system carbon emission data corresponding to the intraday collaborative optimization scheduling scheme of the integrated energy system is obtained. If the system carbon emission data exceeds the warning threshold, non-core load reduction processing is performed on the intraday collaborative optimization scheduling scheme, and the collaborative optimization scheduling result corresponding to the intraday collaborative optimization scheduling scheme is obtained.

[0032] Secondly, this application also provides a collaborative optimization scheduling device for integrated energy systems, comprising:

[0033] The data acquisition module is used to acquire multi-source heterogeneous data of the integrated energy system when a collaborative optimization scheduling request for the integrated energy system is received.

[0034] The mechanism model construction module is used to construct a coupling mechanism model of the integrated energy system based on the multi-source heterogeneous data.

[0035] The mechanism model solving module is used to solve the coupled mechanism model through a multi-agent reinforcement learning algorithm to obtain the collaborative optimization scheduling strategy of the integrated energy system.

[0036] The collaborative optimization scheduling module is used to generate a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy through a three-level scheduling mechanism, and to optimize the collaborative optimization scheduling scheme during the execution of the collaborative optimization scheduling scheme to obtain the collaborative optimization scheduling result.

[0037] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0038] Upon receiving a collaborative optimization scheduling request for the integrated energy system, acquire the multi-source heterogeneous data of the integrated energy system;

[0039] A coupling mechanism model of the integrated energy system is constructed based on the multi-source heterogeneous data.

[0040] The coupling mechanism model is solved by a multi-agent reinforcement learning algorithm to obtain the collaborative optimization scheduling strategy of the integrated energy system.

[0041] A three-level scheduling mechanism is used to generate a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy. During the execution of the collaborative optimization scheduling scheme, the scheme is optimized to obtain the collaborative optimization scheduling result.

[0042] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0043] Upon receiving a collaborative optimization scheduling request for the integrated energy system, acquire the multi-source heterogeneous data of the integrated energy system;

[0044] A coupling mechanism model of the integrated energy system is constructed based on the multi-source heterogeneous data.

[0045] The coupling mechanism model is solved by a multi-agent reinforcement learning algorithm to obtain the collaborative optimization scheduling strategy of the integrated energy system.

[0046] A three-level scheduling mechanism is used to generate a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy. During the execution of the collaborative optimization scheduling scheme, the scheme is optimized to obtain the collaborative optimization scheduling result.

[0047] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0048] Upon receiving a collaborative optimization scheduling request for the integrated energy system, acquire the multi-source heterogeneous data of the integrated energy system;

[0049] A coupling mechanism model of the integrated energy system is constructed based on the multi-source heterogeneous data.

[0050] The coupling mechanism model is solved by a multi-agent reinforcement learning algorithm to obtain the collaborative optimization scheduling strategy of the integrated energy system.

[0051] A three-level scheduling mechanism is used to generate a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy. During the execution of the collaborative optimization scheduling scheme, the scheme is optimized to obtain the collaborative optimization scheduling result.

[0052] The aforementioned collaborative optimization scheduling method, apparatus, computer equipment, computer-readable storage medium, and computer program product for integrated energy systems, upon receiving a collaborative optimization scheduling request for an integrated energy system, acquire multi-source heterogeneous data of the integrated energy system; construct a coupling mechanism model of the integrated energy system based on the multi-source heterogeneous data; solve the coupling mechanism model using a multi-agent reinforcement learning algorithm to obtain a collaborative optimization scheduling strategy for the integrated energy system; generate a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy through a three-level scheduling mechanism, and optimize the collaborative optimization scheduling scheme to obtain the target collaborative optimization scheduling scheme. The scheme of this application, by acquiring multi-source heterogeneous data of the integrated energy system, constructing a coupling mechanism model, and solving the coupling mechanism model using a multi-agent reinforcement learning algorithm, overcomes the optimization lag problem, obtains a collaborative optimization scheduling strategy suitable for integrated energy systems, and finally executes the corresponding collaborative optimization scheduling scheme through a three-level scheduling mechanism, optimizing the collaborative optimization scheduling scheme during execution to improve the collaborative optimization scheduling effect for integrated energy systems. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 This is an application environment diagram for a collaborative optimization scheduling method for an integrated energy system in one embodiment.

[0055] Figure 2 This is a flowchart illustrating a collaborative optimization scheduling method for an integrated energy system in one embodiment;

[0056] Figure 3 This is a flowchart illustrating the electricity-carbon-tax synergistic optimization steps based on a combined energy system of ten power generation and supply in another embodiment.

[0057] Figure 4 This is a schematic diagram of a combined cooling, heating, and power (CCHP) system for plant use in one embodiment.

[0058] Figure 5 This is a structural block diagram of a collaborative optimization scheduling device for an integrated energy system in one embodiment;

[0059] Figure 6 An internal structural diagram of a computer device in one embodiment.

[0060] Figure 7 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0062] The collaborative optimization scheduling method for integrated energy systems provided in this application can be applied to, for example... Figure 1In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed on the cloud or other network servers. When a user needs to perform collaborative optimization scheduling analysis on a specified integrated energy system, they can initiate a collaborative optimization scheduling request to server 104 through terminal 102. Upon receiving the collaborative optimization scheduling request for the integrated energy system, server 104 acquires multi-source heterogeneous data of the integrated energy system; constructs a coupling mechanism model of the integrated energy system based on the multi-source heterogeneous data; solves the coupling mechanism model using a multi-agent reinforcement learning algorithm to obtain the collaborative optimization scheduling strategy of the integrated energy system; generates a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy through a three-level scheduling mechanism; and optimizes the collaborative optimization scheduling scheme during its execution to obtain the collaborative optimization scheduling result. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart vehicle devices, projection devices, etc. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted displays. Head-mounted displays can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. Server 104 can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server providing cloud computing services.

[0063] In one exemplary embodiment, such as Figure 2 As shown, a collaborative optimization scheduling method for integrated energy systems is provided, which can be applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps 201 to 207. Wherein:

[0064] Step 201: Upon receiving a collaborative optimization scheduling request for the integrated energy system, acquire multi-source heterogeneous data of the integrated energy system.

[0065] Integrated energy systems refer to energy systems capable of supplying multiple energy types, such as the combined cooling, heating, and power (CCHP) system. This system can collaboratively supply ten core energy types, including electricity, residential heating, industrial high-pressure steam, industrial low-pressure steam, chilled water (for cooling), demineralized water, compressed air, oxygen, calcium chloride solution (waste heat recovery heat transfer medium), and hydrogen. This covers the energy needs of all production and living scenarios within the industrial park, forming a multi-energy complementary and tiered energy supply system. The energy supply end of the integrated energy system integrates the upstream power grid, photovoltaic equipment, and natural gas supply network. The energy conversion end is equipped with a thermal power plant (including heating and power supply modules) and an Organic Rankine Cycle (ORC) unit. The ORC recovers waste heat from the thermal power plant to achieve thermoelectric decoupling. The energy storage end includes electrical and thermal energy storage. Terminal loads are classified into stationary, transferable, and substitutable loads according to their regulation characteristics, forming a broad-based basis for electricity and heat demand response. Collaborative optimization scheduling refers to the unified planning and dynamic control of various energy production, conversion, storage and consumption processes through multi-objective, multi-timescale and multi-entity coordination, in order to maximize the safety, economy, low carbon and intelligence of system operation. In terms of specific implementation, real-time scheduling instructions for each device can be generated through collaborative optimization scheduling, thereby controlling the operation of the integrated energy system.

[0066] For example, when a user on terminal 102 wishes to perform collaborative optimization scheduling on a specified integrated energy system, they can submit a corresponding collaborative optimization scheduling request to server 104 through terminal 102, specifying the integrated energy system to be processed in the request. Upon receiving the collaborative optimization scheduling request for the integrated energy system, server 104 first obtains the multi-source heterogeneous data of the integrated energy system as the basic data used in the processing. In a specific embodiment, this application is applied to the collaborative optimization scheduling of the park's integrated energy system. The system automatically generates collaborative optimization scheduling requests periodically. During the operation of the park's integrated energy system, multiple types of sensing devices are deployed according to a distributed resource monitoring scheme to achieve full-dimensional data coverage: power sensors are installed at key nodes of core equipment such as thermal power plants (including power generation and heating modules) to collect real-time power output, heat output, and fuel input data; carbon concentration monitors are set up at each carbon emission source to simultaneously acquire carbon emission and carbon quota information; time-of-use electricity prices and power purchase / sale data are collected through smart meters and electricity price collection terminals; combined with environmental data collection requirements, meteorological equipment records parameters such as temperature and illumination, and load monitoring terminals collect electricity and heat load consumption data, ensuring that the data covers energy flow, carbon flow, economic parameters, and environmental factors. These data are then preprocessed to obtain multi-source heterogeneous data.

[0067] Step 203: Construct a coupling mechanism model of the integrated energy system based on multi-source heterogeneous data.

[0068] The coupling mechanism model specifically includes an electricity-carbon-tax coupling mechanism model. It can construct specialized sub-models for core energy conversion equipment such as thermal power plants and energy storage devices within an integrated energy system, based on the energy conversion modeling approach. Then, in the carbon emission accounting and carbon tax cost calculation stages, combined with the design logic of the tiered carbon trading mechanism, a carbon-tax mechanism sub-model is constructed, which is then combined into an overall coupling mechanism model to achieve collaborative optimization.

[0069] For example, after obtaining multi-source heterogeneous data, server 104 can use this data to construct a coupling mechanism model of the integrated energy system, and then use this coupling mechanism model to construct a collaborative optimization scheduling system corresponding to the electricity-carbon-tax coupled trading market. In a specific embodiment, the construction process of the coupling mechanism model can be based on the coupling characteristics of energy in the integrated energy system, introducing a tiered carbon trading cost function to define the boundary constraints, and then reconstructing the electricity-carbon-tax optimization model as the coupling mechanism model.

[0070] Step 205: Solve the coupling mechanism model using a multi-agent reinforcement learning algorithm to obtain the collaborative optimization scheduling strategy for the integrated energy system.

[0071] Among them, multi-agent reinforcement learning algorithms refer to reinforcement learning methods based on multiple agents. By pre-constructing multiple agents and then training them in real-world application scenarios, they can solve the coupling mechanism model of integrated energy systems. In an integrated energy system, the multiple agents can be divided into four main types: energy supply, load demand, carbon management, and energy storage scheduling. The energy supply agent is responsible for regulating the output of thermal power plants; the load demand agent manages demand response strategies; the carbon management agent optimizes carbon tax costs; and the energy storage scheduling agent controls the charging and discharging power of energy storage devices, achieving multi-agent collaborative optimization.

[0072] For example, for the process of achieving collaborative optimization scheduling through a coupling mechanism model, the coupling mechanism model can be used to constrain the decision space of multi-agent reinforcement learning, thereby dynamically updating network parameters to form a "perception-diagnosis-decision" closed loop, achieving multi-objective optimization, and obtaining a collaborative optimization scheduling strategy for the integrated energy system. In specific applications, the reinforcement learning modeling method can be referenced, defining the agent's observation state as real-time electricity price, total system carbon emissions, renewable energy output, load demand, and energy storage status, comprehensively capturing key information about system operation; the action state as equipment output adjustment, demand response load adjustment, and energy storage power, covering core control variables. The reward function is centered on the "reduction in system operating costs," comprehensively considering energy purchase, carbon tax, and curtailment costs, deducting electricity sales revenue, and forming a feedback signal to guide agent learning. Based on the historical operation data of the park, the model is trained, a parameter sharing mechanism is introduced to improve the learning efficiency of multi-agents, and a recurrent neural network (RNN) is used to enhance the feature extraction capability of time-series data until the model converges, generating real-time scheduling instructions for each device, and obtaining a collaborative optimization scheduling strategy for the integrated energy system.

[0073] Step 207: Generate a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy through a three-level scheduling mechanism, and optimize the collaborative optimization scheduling scheme during the execution of the collaborative optimization scheduling scheme to obtain the collaborative optimization scheduling result.

[0074] The three-tiered dispatch mechanism specifically refers to the "day-intraday-real-time" dispatch system, a hierarchical, rolling, and closed-loop optimization control architecture adopted for modern energy systems with high proportions of renewable energy integration, multi-energy coupling, and strong uncertainties. It divides the dispatch process into three time scales, refining and dynamically correcting each level to address prediction errors and operational disturbances. In the day-intraday phase, it determines executable collaborative optimization dispatch schemes based on established collaborative optimization dispatch strategies. In the intraday phase, it optimizes and adjusts the collaborative optimization dispatch schemes according to actual conditions. In the real-time phase, it monitors real-time data and responds rapidly to complete the collaborative optimization dispatch of the integrated energy system.

[0075] For example, regarding the execution process of the collaborative optimization scheduling strategy, this application can specifically implement the execution of the collaborative optimization scheduling strategy through a three-level scheduling mechanism. The three-level scheduling mechanism achieves layered decoupling and dynamic closed-loop control across three time dimensions: "day-ahead," "intra-day," and "real-time." Specifically, the three-level scheduling mechanism can ensure the stable implementation of the optimization scheme in multiple scenarios through demand-side adjustment and energy storage strategy correction. Specifically, the day-ahead plan generates an initial scheduling scheme based on the next day's forecast data; intra-day adjustments update real-time data every 15 minutes to correct equipment output and energy storage strategies; and real-time response triggers demand response for emergency situations such as exceeding carbon emission warnings, ensuring system operational stability.

[0076] The aforementioned collaborative optimization scheduling method for integrated energy systems involves acquiring multi-source heterogeneous data of the integrated energy system upon receiving a collaborative optimization scheduling request; constructing a coupling mechanism model of the integrated energy system based on the multi-source heterogeneous data; solving the coupling mechanism model using a multi-agent reinforcement learning algorithm to obtain a collaborative optimization scheduling strategy for the integrated energy system; generating a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy through a three-level scheduling mechanism; and optimizing the collaborative optimization scheduling scheme during its execution to obtain the collaborative optimization scheduling result. The scheme in this application overcomes the optimization lag problem by acquiring multi-source heterogeneous data of the integrated energy system, constructing a coupling mechanism model, and solving the model using a multi-agent reinforcement learning algorithm, thereby obtaining a collaborative optimization scheduling strategy suitable for integrated energy systems. Finally, it executes the corresponding collaborative optimization scheduling scheme through a three-level scheduling mechanism, and optimizes the scheme during execution, thereby improving the collaborative optimization scheduling effect for integrated energy systems.

[0077] Traditional combined heat and power (CHP) systems have significant limitations in coordinated dispatching of electricity, carbon, and taxes. The system equipment is tightly coupled, and dispatching is affected by multiple factors, including energy price fluctuations, changes in carbon policies, and the interplay of interests among multiple stakeholders, making coordinated optimization difficult. At the electricity-carbon coordination level, existing dispatching often separates energy trading from carbon control, or adopts a single carbon tax / carbon trading mechanism, making it difficult to balance economic costs and carbon reduction targets. Furthermore, it fails to fully consider the direct trading needs between multiple microgrids, relying on intermediary platforms leading to low trading efficiency. Regarding tax-electricity linkage, dynamic electricity pricing does not deeply integrate carbon tax costs, failing to accurately guide user demand response and easily resulting in excessive peak-valley load differences. In terms of dispatching optimization, it largely relies on model-driven methods, which are weak in handling uncertainties such as renewable energy output and load demand, lack multi-stakeholder privacy protection mechanisms, and optimization strategies lag behind real-time operating conditions, making it difficult to achieve overall system optimization.

[0078] Addressing the limitations of traditional combined heat and power (CHP) systems, such as insufficient coordination of electricity, carbon, and tax, fragmented demand response, and poor adaptability of optimization methods, this application aims to solve the imbalance between economic efficiency and low-carbon performance caused by the complex multi-energy coupling, singular carbon tax control, and lagging optimization in traditional systems. It constructs a collaborative framework integrating generalized electricity and heat demand response, tiered carbon trading, and reinforcement learning algorithms to achieve dynamic optimization of system scheduling, reduce plant power consumption and carbon emissions, and improve the economic and low-carbon operation of the integrated energy system. This application first constructs an electricity-carbon-tax coupled trading market, designs a multi-microgrid direct trading mechanism and a tiered carbon tax-dynamic electricity price linkage model to achieve efficient allocation of energy and carbon resources. Then, based on multi-agent deep reinforcement learning, it introduces parameter sharing and recurrent neural network structures to improve the adaptability of scheduling strategies under uncertain scenarios while protecting the privacy of stakeholders. Finally, through a three-level scheduling mechanism, based on game equilibrium analysis and online evaluation, it dynamically corrects the scheduling strategy to achieve synergistic optimization of economic costs, carbon emissions, and tax control, providing scientific support for the low-carbon economic operation of CHP systems.

[0079] In an exemplary embodiment, obtaining multi-source heterogeneous data of an integrated energy system includes: obtaining energy data, equipment data, and system-related data corresponding to the integrated energy system to obtain multi-source data; performing abnormal data cleaning processing on the multi-source data to obtain multi-source complete data; and performing data normalization processing on the multi-source complete data to obtain multi-source heterogeneous data.

[0080] Multi-source heterogeneous data specifically includes three types: energy data, equipment data, and system-related data. Examples include electricity purchase and sale data for each microgrid, gas turbine output, carbon emissions, carbon quotas, real-time electricity prices, and ambient temperature. For the specific multi-source heterogeneous data acquisition process, power sensors can be installed at key nodes of core equipment in thermal power plants (including power generation and heating modules) to collect real-time electrical output, thermal output, and fuel input data. Carbon concentration monitors can be installed at each carbon emission source to simultaneously acquire carbon emission and carbon quota information. Time-of-use electricity prices and electricity purchase and sale power data can be collected through smart meters and electricity price collection terminals. In conjunction with environmental data collection requirements, meteorological equipment can be used to record parameters such as temperature and light intensity, and load monitoring terminals can collect electricity and heat load consumption data to ensure that the data covers energy flow, carbon flow, economic parameters, and environmental factors.

[0081] For example, by pre-integrating corresponding data acquisition devices into the corresponding equipment of the integrated energy system, energy data, equipment data, and system-related data of the integrated energy system can be acquired first, obtaining multi-source data as the basis for analysis. Then, anomaly cleaning processing is performed on the multi-source data to obtain multi-source completed data. After data acquisition, processing is carried out according to data preprocessing specifications: first, anomaly data is identified, and data exceeding the rated operating range of the equipment or deviating from historical fluctuation patterns are marked as anomalies. The cause is traced through equipment operation logs. Linear interpolation is used to complete the data in case of sensor failure, and weighted average correction is used for occasional interference. For data with a low missing rate, analogy with data from similar equipment operating during the same period is used for completion. After the anomaly cleaning processing is completed, data normalization processing can be performed on the multi-source completed data to obtain multi-source heterogeneous data. Specifically, data standardization and fusion can be completed through Min-Max normalization processing and data consistency verification. The normalization formula is as follows:

[0082]

[0083] In the formula: Let represent the eigenvector, specifically the i-th eigenvector. This is the normalized value of the i-th vector in vector x. By employing a normalization method, raw data with different dimensions are mapped to a unified interval, eliminating magnitude differences and forming a standardized data set, providing suitable input for subsequent model calculations. In this embodiment, through multi-source data acquisition, cleaning, and normalization processing, the various multi-source data corresponding to the integrated energy system can be effectively extracted, thereby effectively ensuring the accuracy of subsequent collaborative optimization scheduling analysis.

[0084] In an exemplary embodiment, step 203 includes: identifying energy conversion devices in the integrated energy system and constructing an energy conversion sub-model corresponding to the energy conversion devices; determining the carbon emission formula of the integrated energy system based on multi-source heterogeneous data; constructing a carbon tax cost sub-model of the integrated energy system based on the carbon emission formula; and constructing a coupling mechanism model of the integrated energy system based on the energy conversion sub-model and the carbon tax cost sub-model.

[0085] For example, the coupling mechanism model specifically includes two parts: an energy conversion sub-model and a carbon tax cost sub-model. First, it is necessary to identify the energy conversion devices in the integrated energy system and construct corresponding energy conversion sub-models for these devices. For core energy conversion devices such as thermal power plants and energy storage equipment, specific sub-models are constructed based on the energy conversion modeling approach: Thermal power plants adhere to the principle of energy conservation, clearly defining the total energy conversion efficiency, waste heat recovery efficiency range, and power generation efficiency to ensure that the dynamic changes in electrical and thermal output conform to the equipment characteristics; for electrical and thermal energy storage equipment, a dynamic change model of the state of charge (SOC) and thermal storage state is constructed by combining charge / discharge and charge / discharge efficiency to reflect the laws of energy storage and release. In a specific embodiment, a thermal power plant outputs both electrical and thermal energy simultaneously by burning coal or natural gas; the thermoelectric output is correlated and must satisfy energy conservation; the heating module of the thermal power plant outputs thermal energy through fuel combustion, and the conversion efficiency is affected by the equipment's waste heat recovery operating conditions. A conversion model needs to be constructed for the power generation and heating characteristics of the thermal power plant, as shown in the following formula:

[0086]

[0087]

[0088] In the formula, in the formula, This represents the fuel input power (kW) of the thermal power plant at time t, where coal is converted to equivalent power based on its calorific value. The total energy conversion efficiency of a thermal power plant is defined as 0.8 to 0.9 for coal-fired power plants and 0.85 to 0.95 for gas-fired power plants. Let t be the electrical output (kW) of the thermal power plant at time t. Let be the thermal output (kW) of the thermal power plant at time t. The waste heat recovery efficiency of thermal power plants is taken as 0.6~0.75. The power generation efficiency of thermal power plants is 0.35 to 0.45 for coal-fired power plants and 0.5 to 0.6 for gas-fired combined cycle power plants, reflecting the core performance of power plant energy conversion.

[0089] The carbon emission processing for integrated energy systems includes carbon emission accounting and carbon tax cost calculation, which incorporates the design logic of a tiered carbon trading mechanism. In practical application, for the integrated energy system of a park, carbon emission factors can be matched according to equipment type. The upstream power grid purchases electricity based on the level of traditional coal-fired units in the region, the thermal power plant sets its parameters according to fuel combustion characteristics, and the renewable energy factor is set to 0. The carbon emissions of a single device are calculated using equipment operating parameters, factors, and operating time, and the total system emissions are then summed. Carbon tax costs are divided into tiered intervals based on carbon quotas: emission reduction, normal, and excess emission intervals. In the emission reduction interval, revenue is calculated based on the base carbon price plus a reward coefficient; in the normal interval, costs are calculated based on the base carbon price; and in the excess emission interval, costs are calculated based on the base carbon price plus a penalty coefficient. Simultaneously, the concept of "virtual carbon storage" is introduced to break the real-time carbon emission balance constraint, allowing carbon emission responsibilities to flexibly transfer among users in the park, achieving deep coupling between electricity flow and carbon costs. To address this process, a mechanism model can be established based on the system's physical laws and energy coupling characteristics, encompassing multi-energy flow, equipment energy conversion, carbon emission calculation, and carbon tax cost accounting. The carbon emission calculation references the formula relating equipment energy consumption and carbon emission factors.

[0090]

[0091] In the formula, Let be the power of the k-th device at time t. For equipment carbon emission factor, For time intervals.

[0092] Then, based on this carbon emission formula, a carbon tax cost sub-model for the integrated energy system is constructed. The carbon tax cost is calculated using a tiered formula:

[0093]

[0094] In the formula, For carbon tax costs, As for the emission reduction incentive coefficient, Here, d represents the penalty coefficient for exceeding emission limits, and d represents the length of the carbon emission range. For carbon emissions, This model deeply couples electrical energy flow, carbon emission accounting, and carbon tax costs to construct the physical constraint boundary of the data-driven model. This provides mechanistic constraint support for subsequent collaborative optimization, ensuring that the optimization strategy complies with the system's energy conservation and carbon control rules. This process requires a tiered carbon trading mechanism. The construction of this mechanism necessitates clarifying the following key rules: The calculation of free carbon emission allowances must cover different carbon emission sources. For upstream power purchases, the calculation should combine the purchased electricity volume with the free carbon emission coefficient per unit of electricity from coal-fired units. For thermal power plants, the calculation should combine heat output, electricity output, and the free carbon emission coefficient per unit of heat, ensuring that allowance allocation matches the characteristics of carbon emission sources. Actual carbon emission accounting should be based on upstream power purchases and actual carbon emissions from thermal power plants, clarifying the accounting logic for each emission source. Carbon trading cost accounting should refine the tiered rules according to the difference range: when actual emissions are lower than the allowance, revenue is calculated according to the tier corresponding to the difference range; when actual emissions are higher than the allowance, costs are calculated in reverse according to the tier corresponding to the difference range. The range length and price growth rate should be dynamically calibrated based on carbon trading market policies to ensure the market adaptability of cost accounting. In this embodiment, by constructing an energy conversion sub-model and a carbon tax cost sub-model respectively, the electricity-carbon-tax collaborative optimization process of the integrated energy system's collaborative optimization scheduling is achieved, ensuring the optimization effect.

[0095] In an exemplary embodiment, the cooperative optimization scheduling strategy of the integrated energy system is obtained by solving the coupling mechanism model using a multi-agent reinforcement learning algorithm. This includes: constructing an objective function based on the carbon tax cost sub-model, with the goal of minimizing the total daily operating cost of the integrated energy system; constructing constraint conditions based on the energy conversion sub-model; constructing a solution state space and a solution action space based on multi-source heterogeneous data; and solving the coupling mechanism model using a multi-agent model based on the objective function, constraint conditions, state space, and action space to obtain the cooperative optimization scheduling strategy of the integrated energy system.

[0096] For example, this application models the solution process of the coupling mechanism model as a Markov decision process, using the real-time system state as the state space, equipment and load adjustment quantities as the action space, and the negative value of the objective function as the reward function. A multi-agent reinforcement learning framework with parameter sharing and recurrent neural networks is introduced to output the optimal scheduling strategy. Therefore, when solving the coupling mechanism model, the constraint boundary can be defined first by introducing a tiered carbon trading cost function based on the system's energy coupling characteristics, and then an electricity-carbon-tax optimization model can be constructed. The mechanism model constrains the multi-agent reinforcement learning decision space for solution processing. Specifically, the coupling mechanism model for achieving collaborative optimization scheduling takes minimizing the total daily operating cost of the system as its core objective, covering the costs and benefits of all aspects including energy procurement, carbon trading, curtailment of solar power, and demand response. It uses a multi-energy equipment conversion model and a carbon emission accounting model as constraints, and employs a multi-agent reinforcement learning method to solve the problem. A "perception-diagnosis-decision" closed loop is formed by dynamically updating network parameters to adapt to source-load uncertainties. The specific objective function formula is as follows:

[0097]

[0098]

[0099] In the formula, The total daily operating cost of the system; t is the scheduling time, ranging from 1 to 24; The cost of electricity and heat supply from a thermal power plant at time t; Let t be the carbon trading cost at time t; Let t be the cost of discarding light. The cost of demand response compensation at time t; Let t be the revenue from electricity sales. The equivalent electrical power calculated from the coal consumption at time t; This is the unit cost coefficient for coal; Let t be the total output of the coal-fired power plant. This represents the unit operation and maintenance cost coefficient for coal-fired power plants. The equivalent electrical power is calculated based on the gas intake volume at time t. This is the unit cost coefficient for gas.

[0100] As for the constraints, they must cover the supply and demand balance of electrical and thermal power, as well as the operational boundaries of core equipment such as thermal power plants and energy storage, to ensure the feasibility of the dispatch strategy at the physical level. Power balance constraints ensure real-time matching of energy supply and demand, while equipment operation constraints prevent equipment from exceeding its limits, extend equipment lifespan, and ensure system safety. The specific constraint formulas are as follows:

[0101]

[0102]

[0103] In the formula, Let t be the power purchased by the upstream supplier. Let t be the electrical output of the thermal power plant at time t; Let t be the total renewable energy consumption capacity at time t; Let t be the initial electrical load power of the system at time t; , These represent the charging and discharging power of the stored energy at time t, respectively. Let be the power output of the system sold to the grid at time t; Let t be the total adjustment of the demand response. The thermal output of the thermal power plant at time t; Let be the heat release power of the thermal energy storage at time t; Let t be the heat load at time t; Let t be the charging power of thermal energy storage at time t. Then, based on multi-source heterogeneous data, a solution state space and a solution action space are constructed; the real-time state of the integrated energy system is used as the state space, and the equipment and load adjustments within the integrated energy system are used as the action space. Then, based on solving the objective function, solving the constraints, solving the state space, and solving the action space, a multi-agent reinforcement learning algorithm is used to solve the coupling mechanism model, resulting in a collaborative optimization scheduling strategy for the integrated energy system. Referring to reinforcement learning modeling methods, the agent's observation state is defined as real-time electricity price, total system carbon emissions, renewable energy output, load demand, and energy storage status, comprehensively capturing key information about system operation; the action state is defined as equipment output adjustment, demand response load adjustment, and energy storage power, covering core control variables. The reward function is centered on the "reduction in system operating costs," comprehensively considering energy purchase, carbon tax, and curtailment costs, deducting electricity sales revenue, and forming a feedback signal to guide the agent's learning. Based on historical operational data of the park, a training model is developed, and a parameter sharing mechanism is introduced to improve the learning efficiency of multi-agent systems. A recurrent neural network (RNN) is employed to enhance the feature extraction capability of time-series data until the model converges, generating real-time scheduling instructions for each device, thereby obtaining a collaborative optimization scheduling strategy for the integrated energy system. In this embodiment, a Markov decision process is used to solve the coupling mechanism model, effectively ensuring the accuracy and efficiency of the generated collaborative optimization scheduling strategy.

[0104] Furthermore, the method also includes: constructing a multi-agent system corresponding to the integrated energy system, comprising an energy supply agent, a load demand agent, a carbon management agent, and an energy storage scheduling agent; acquiring historical operating data corresponding to the integrated energy system; constructing agent training data based on the historical operating data; and training the multi-agent system using the agent training data through a parameter sharing mechanism and a cyclic time network to obtain the model-solving multi-agent system. For example, in this application's scheme, the solving multi-agent system corresponding to the integrated energy system includes four types of agents: energy supply agent, load demand agent, carbon management agent, and energy storage scheduling agent. The energy supply agent is responsible for regulating the output of thermal power plants, the load demand agent manages demand response strategies, the carbon management agent optimizes carbon tax costs, and the energy storage scheduling agent controls the charging and discharging, and heat dissipation power of energy storage equipment, achieving multi-agent collaborative optimization. The training process of the agents specifically refers to training the machine learning models contained within the agents. In this process, historical operating data of the integrated energy system can be collected first as agent training data. The model solves for the multi-agent problem by training multiple agents using training data, a parameter-sharing mechanism, and a recurrent time network. For the network structure design, both the actor and critic networks employ a fully connected "input layer-hidden layer-output layer" structure. A recurrent neural network component is introduced in the hidden layer to extract temporal dynamic features, and the underlying feature extraction layer is shared, reducing the number of network parameters and improving the algorithm's generalization ability. During training optimization, an experience replay pool is used to store "state-action-reward-next state" samples, and random sampling avoids sample correlation. A generalized advantage estimation method is introduced to reduce reward variance, and a gradient pruning mechanism is set to limit the gradient range and avoid training oscillations. In this embodiment, parameter sharing and a recurrent neural network structure are introduced during the training process to solve for the multi-agent problem, improving the adaptability of the scheduling strategy under uncertain scenarios while protecting the privacy of the agents.

[0105] In an exemplary embodiment, step 207 includes: generating a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy during the day-ahead phase of the three-level scheduling mechanism; acquiring the corresponding real-time system data of the integrated energy system during the intraday phase of the three-level scheduling mechanism, adjusting the equipment output strategy and energy storage strategy of the collaborative optimization scheduling scheme according to the real-time system data, and obtaining the intraday collaborative optimization call scheme; acquiring the system carbon emission data corresponding to the execution of the intraday collaborative optimization call scheme during the real-time phase of the three-level scheduling mechanism, and performing non-core load reduction processing on the intraday collaborative optimization call scheme when the system carbon emission data exceeds the warning threshold, and obtaining the collaborative optimization scheduling result corresponding to the intraday collaborative optimization call scheme.

[0106] For example, to address the impact of source-load uncertainty on scheduling strategies, this application establishes a three-tiered scheduling mechanism of "day-ahead, intraday, and real-time," ensuring the stable implementation of optimized solutions across multiple scenarios through demand-side adjustments and energy storage strategy corrections. Specifically, the day-ahead plan generates an initial scheduling scheme based on collaborative optimization and next-day forecast data; intraday adjustments update real-time data every 15 minutes to correct equipment output and energy storage strategies; and real-time responses trigger demand response for emergency situations involving carbon emission exceeding warning levels, ensuring system operational stability.

[0107] When the source-load deviation (the difference between the actual and predicted values) exceeds 10%, in order to maintain power balance, it is necessary to adjust the output of thermal power plants and the charging and discharging power of energy storage to smooth out fluctuations. The formula is as follows:

[0108]

[0109]

[0110] In the formula, , These are the power output of thermal power plants before and after the correction, respectively. The source load deviation at time t; This is a correction factor for the output of thermal power plants, with a value ranging from 0.4 to 0.5, to prevent sudden changes in the output of thermal power plants from affecting the lifespan of equipment. , These represent the energy storage charging and discharging power before and after the correction, respectively. The source load deviation at time t; It is the power correction factor for electric energy storage, with a value ranging from 0.3 to 0.4, which utilizes the flexibility of energy storage to quickly smooth out source load fluctuations.

[0111] When carbon emissions exceed the warning threshold, an emergency demand response must be triggered to reduce load and offset the excess emissions. The formula is as follows:

[0112]

[0113] In the formula, For emergency demand response load reduction; To maximize the load that the system can reduce, and to avoid excessive reductions that could impact users' electricity consumption; Let t be the total carbon emissions of the system at time t; Let be the system's free carbon allowance at time t; Carbon emission factors for electricity purchased by higher authorities; The scheduling interval is set to ensure that the load reduction amount matches the carbon emission reduction demand.

[0114] In addition, for the collaborative optimization scheduling process, the optimization effect can be verified regularly, and key indicators such as the overall system cost, carbon emissions, and renewable energy consumption rate before and after optimization can be compared. If the effect does not meet expectations, parameters such as the tiered carbon tax range, reward function weight, and equipment carbon emission factor can be adjusted. Through the closed-loop iteration of "data collection-model calculation-strategy execution-evaluation and correction", the scheduling strategy can be continuously adapted to changes in park load, policy adjustments, and dynamic changes in equipment operating conditions, thereby improving the system's economic efficiency and low carbon emissions.

[0115] In one specific embodiment, this application is applied to the coordinated optimization scheduling of the combined heat and power (CHP) system within the park, focusing on electricity, carbon, and tax. The overall coordinated optimization flowchart can be referenced in this case. Figure 3 As shown, the diagram of the ten-generation power supply system can be referenced. Figure 4 As shown, the overall collaborative optimization process can be divided into four main steps: multi-source data acquisition and standardization processing, construction of the electricity-carbon-tax coupling mechanism model, solution of the electricity-carbon-tax collaborative optimization model, and implementation and dynamic correction of optimization strategies.

[0116] (1) Multi-source data acquisition and preprocessing

[0117] During the operation of the park's integrated energy system, multiple types of sensing devices are deployed according to the distributed resource monitoring scheme to achieve full-dimensional data coverage: power sensors are installed at key nodes of core equipment such as thermal power plants (including power generation and heating modules) to collect real-time power output, heat output, and fuel input data; carbon concentration monitors are set up at each carbon emission source to simultaneously acquire carbon emission and carbon quota information; time-of-use electricity prices and power purchase and sale data are collected through smart meters and electricity price collection terminals; and in conjunction with environmental data collection requirements, meteorological equipment is used to record parameters such as temperature and light intensity, and electricity and heat load consumption data are collected through load monitoring terminals to ensure that the data covers energy flow, carbon flow, economic parameters, and environmental factors.

[0118] After data collection, data preprocessing is carried out in accordance with the data preprocessing specifications: First, abnormal data is identified, and data that exceeds the rated operating range of the equipment or deviates from the historical fluctuation pattern are marked as abnormal. The cause is traced through the equipment operation log. When there is a sensor failure, linear interpolation is used to complete the data, and random interference is corrected by weighted average of data from adjacent time periods. For data with a low missing rate, the missing data is completed by analogy with the operating data of similar equipment during the same period. Finally, a normalization method is used to map the original data of different dimensions to a unified interval, eliminate the difference in magnitude, and form a standardized data set to provide suitable input for subsequent model calculations.

[0119] (2) Construction of the coupling mechanism model of electricity-carbon-tax

[0120] For core energy conversion equipment such as thermal power plants and energy storage devices in the park, specialized sub-models are constructed based on the energy conversion modeling approach: thermal power plants follow the principle of energy conservation, clarifying the range of total energy conversion efficiency, waste heat recovery efficiency, and power generation efficiency to ensure that the dynamic changes in electrical and thermal output conform to the characteristics of the equipment; for electrical and thermal energy storage devices, dynamic change models of state of charge (SOC) and thermal storage state are constructed by combining charging and discharging efficiency and charging and discharging efficiency to reflect the laws of energy storage and release.

[0121] The carbon emission accounting and carbon tax cost calculation process incorporates the design logic of a tiered carbon trading mechanism: carbon emission factors are matched according to equipment type, the upstream power grid purchases electricity with reference to the level of traditional coal-fired units in the region, thermal power plants are set according to fuel combustion characteristics, and the renewable energy factor is set to 0. The carbon emissions of a single device are calculated through equipment operating parameters, factors, and operating time, and the total system emissions are summed. The carbon tax cost adopts a tiered interval division, based on carbon quotas, into emission reduction, normal, and over-emission intervals. The emission reduction interval calculates revenue based on the base carbon price plus a reward coefficient, the normal interval calculates costs based on the base carbon price, and the over-emission interval calculates costs based on the base carbon price plus a penalty coefficient. At the same time, the concept of "virtual carbon storage" is introduced to break the real-time balance constraint of carbon emissions, allowing carbon emission responsibilities to flexibly transfer among users in the park, realizing a deep coupling between electricity flow and carbon costs.

[0122] (3) Solving the electricity-carbon-tax collaborative optimization model

[0123] An improved multi-agent near-end strategy optimization algorithm is adopted to divide the park's integrated energy system into four major agents: energy supply, load demand, carbon management, and energy storage scheduling. The energy supply agent is responsible for regulating the output of thermal power plants, the load demand agent manages the demand response strategy, the carbon management agent optimizes carbon tax costs, and the energy storage scheduling agent controls the charging and discharging and heat dissipation power of energy storage equipment, thereby achieving multi-agent collaborative optimization.

[0124] Referring to reinforcement learning modeling methods, the agent's observation state is defined as real-time electricity price, total system carbon emissions, renewable energy output, load demand, and energy storage status, comprehensively capturing key information about system operation; the action state is defined as equipment output adjustment, demand response load adjustment, and energy storage power, covering core control variables. The reward function is centered on the "reduction in system operating costs," comprehensively considering energy purchase, carbon tax, and curtailment costs, deducting electricity sales revenue, to form a feedback signal guiding the agent's learning. The model is trained based on historical operational data of the park, and a parameter sharing mechanism is introduced to improve the learning efficiency of multiple agents. A recurrent neural network (RNN) is used to enhance the feature extraction capability of time-series data until the model converges, generating real-time scheduling instructions for each device.

[0125] (4) Optimize the implementation and dynamic adjustment of strategies

[0126] Based on the design concept of a three-level dispatch mechanism, a "day-ahead-intraday-real-time" dispatch system is established: In the day-ahead phase, an initial dispatch plan is generated based on the next day's load, renewable energy, and electricity price forecast data, clarifying the fuel procurement plan, output, and energy storage plan for thermal power plants; In the intraday phase, real-time data is updated regularly, source-load deviation is calculated, and when the deviation exceeds the threshold, the output of thermal power plants and energy storage capacity are adjusted to smooth out fluctuations; In the real-time phase, carbon emissions are monitored, and when they exceed the warning threshold, an emergency demand response is triggered, prioritizing the reduction of non-core loads to ensure that carbon emissions return to a safe range.

[0127] Regularly conduct optimization effect verification, comparing key indicators such as overall system cost, carbon emissions, and renewable energy consumption rate before and after optimization; if the effect does not meet expectations, adjust parameters such as tiered carbon tax range, reward function weight, and equipment carbon emission factor. Through closed-loop iteration of "data collection-model calculation-strategy execution-evaluation and correction", the scheduling strategy can continuously adapt to changes in park load, policy adjustments, and dynamic changes in equipment operating conditions, thereby improving the system's operational economy and low carbon emissions.

[0128] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0129] Based on the same inventive concept, this application also provides a collaborative optimization scheduling device for integrated energy systems for implementing the aforementioned collaborative optimization scheduling method for integrated energy systems. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the collaborative optimization scheduling device for integrated energy systems provided below can be found in the limitations of the collaborative optimization scheduling method for integrated energy systems described above, and will not be repeated here.

[0130] In one exemplary embodiment, such as Figure 5 As shown, a collaborative optimization scheduling device for integrated energy systems is provided, comprising:

[0131] The data acquisition module 502 is used to acquire multi-source heterogeneous data of the integrated energy system when a collaborative optimization scheduling request for the integrated energy system is received.

[0132] Mechanism model construction module 504 is used to construct a coupled mechanism model of an integrated energy system based on multi-source heterogeneous data.

[0133] The mechanism model solving module 506 is used to solve the coupled mechanism model through a multi-agent reinforcement learning algorithm to obtain the collaborative optimization scheduling strategy of the integrated energy system.

[0134] The collaborative optimization scheduling module 508 is used to generate a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy through a three-level scheduling mechanism, and to optimize the collaborative optimization scheduling scheme during the execution of the collaborative optimization scheduling scheme to obtain the collaborative optimization scheduling result.

[0135] In one embodiment, the data acquisition module 502 is specifically used to: acquire energy data, equipment data and system-related data corresponding to the integrated energy system to obtain multi-source data; perform abnormal data cleaning processing on the multi-source data to obtain multi-source complete data; and perform data normalization processing on the multi-source complete data to obtain multi-source heterogeneous data.

[0136] In one embodiment, the mechanism model construction module 504 is specifically used to: identify energy conversion devices in the integrated energy system and construct energy conversion sub-models corresponding to the energy conversion devices; determine the carbon emission formula of the integrated energy system based on multi-source heterogeneous data; construct a carbon tax cost sub-model of the integrated energy system according to the carbon emission formula; and construct a coupling mechanism model of the integrated energy system based on the energy conversion sub-model and the carbon tax cost sub-model.

[0137] In one embodiment, the mechanism model solving module 506 is specifically used to: construct a solution objective function based on the carbon tax cost sub-model, with the goal of minimizing the total daily operating cost of the integrated energy system; construct solution constraints based on the energy conversion sub-model; construct a solution state space and a solution action space based on multi-source heterogeneous data; and, based on the solution objective function, solution constraints, solution state space, and solution action space, solve the coupled mechanism model using a model solving multi-agent based on a multi-agent reinforcement learning algorithm to obtain a collaborative optimization scheduling strategy for the integrated energy system.

[0138] In one embodiment, the system further includes an agent training module, used to: construct multiple agents corresponding to the integrated energy system, the multiple agents including an energy supply agent, a load demand agent, a carbon management agent, and an energy storage scheduling agent; acquire historical operating data corresponding to the integrated energy system; construct agent training data based on the historical operating data; and train the multiple agents through the agent training data, using a parameter sharing mechanism and a cyclic time network, to obtain a model that solves for the multiple agents.

[0139] In one embodiment, the collaborative optimization scheduling module 508 is specifically used for: generating a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy during the day-ahead phase of the three-level scheduling mechanism; acquiring the corresponding real-time system data of the integrated energy system during the intraday phase of the three-level scheduling mechanism, adjusting the equipment output strategy and energy storage strategy of the collaborative optimization scheduling scheme according to the real-time system data, and obtaining the intraday collaborative optimization call scheme; acquiring the system carbon emission data corresponding to the execution of the intraday collaborative optimization call scheme during the real-time phase of the three-level scheduling mechanism, and performing non-core load reduction processing on the intraday collaborative optimization call scheme when the system carbon emission data exceeds the warning threshold, and obtaining the collaborative optimization scheduling result corresponding to the intraday collaborative optimization call scheme.

[0140] The modules in the aforementioned collaborative optimization scheduling device for integrated energy systems can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0141] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 6 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores data related to collaborative optimization scheduling. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network. When the computer program is executed by the processor, it implements a collaborative optimization scheduling method for an integrated energy system.

[0142] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 7As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a collaborative optimization scheduling method for an integrated energy system. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0143] Those skilled in the art will understand that Figure 6 and Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0144] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0145] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0146] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and executes the computer instructions, causing the computer device to perform the steps in the above method embodiments.

[0147] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0148] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0149] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0150] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A collaborative optimization scheduling method for integrated energy systems, characterized in that, The method includes: Upon receiving a collaborative optimization scheduling request for the integrated energy system, acquire the multi-source heterogeneous data of the integrated energy system; A coupling mechanism model of the integrated energy system is constructed based on the multi-source heterogeneous data. The coupling mechanism model is solved by a multi-agent reinforcement learning algorithm to obtain the collaborative optimization scheduling strategy of the integrated energy system; A three-level scheduling mechanism is used to generate a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy. During the execution of the collaborative optimization scheduling scheme, the scheme is optimized to obtain the collaborative optimization scheduling result.

2. The method according to claim 1, characterized in that, The acquisition of multi-source heterogeneous data from the integrated energy system includes: Obtain energy data, equipment data, and system-related data corresponding to the integrated energy system to obtain multi-source data; The multi-source data is subjected to anomaly cleaning processing to obtain multi-source completed data; The multi-source complete data is subjected to data normalization processing to obtain multi-source heterogeneous data.

3. The method according to claim 1, characterized in that, The construction of the coupling mechanism model of the integrated energy system based on the multi-source heterogeneous data includes: Identify the energy conversion devices in the integrated energy system and construct the corresponding energy conversion sub-models for the energy conversion devices; The formula for determining the carbon emissions of the integrated energy system is based on the multi-source heterogeneous data. Construct a carbon tax cost sub-model for the integrated energy system based on the carbon emission formula; Based on the energy conversion sub-model and the carbon tax cost sub-model, a coupling mechanism model of the integrated energy system is constructed.

4. The method according to claim 3, characterized in that, The method of solving the coupling mechanism model using a multi-agent reinforcement learning algorithm to obtain the collaborative optimization scheduling strategy for the integrated energy system includes: Based on the carbon tax cost sub-model, an objective function is constructed with the goal of minimizing the total daily operating cost of the integrated energy system, and a solution constraint is constructed based on the energy conversion sub-model. Construct a solution state space and a solution action space based on the multi-source heterogeneous data; Based on the objective function, the constraints, the state space, and the action space, the coupled mechanism model is solved by multi-agents using a multi-agent reinforcement learning algorithm to obtain the collaborative optimization scheduling strategy of the integrated energy system.

5. The method according to claim 4, characterized in that, The method further includes: Construct a multi-agent system corresponding to the integrated energy system, wherein the multi-agent system includes an energy supply agent, a load demand agent, a carbon management agent, and an energy storage scheduling agent; Obtain the historical operating data corresponding to the integrated energy system; Intelligent agent training data is constructed based on the historical operational data; The multi-agent model is obtained by training the multi-agent model using the training data of the aforementioned agents, through a parameter sharing mechanism and a cyclic time network.

6. The method according to any one of claims 1 to 5, characterized in that, The process of generating a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy through a three-level scheduling mechanism, and optimizing the collaborative optimization scheduling scheme during its execution to obtain the collaborative optimization scheduling result includes: During the day-ahead phase of the three-level scheduling mechanism, a collaborative optimization scheduling scheme corresponding to the aforementioned collaborative optimization scheduling strategy is generated; During the intraday phase of the three-level scheduling mechanism, the corresponding real-time system data of the integrated energy system is obtained, and the equipment output strategy and energy storage strategy of the collaborative optimization scheduling scheme are adjusted according to the real-time system data to obtain the intraday collaborative optimization call scheme. During the real-time phase of the three-level scheduling mechanism, the system carbon emission data corresponding to the intraday collaborative optimization scheduling scheme of the integrated energy system is obtained. If the system carbon emission data exceeds the warning threshold, non-core load reduction processing is performed on the intraday collaborative optimization scheduling scheme, and the collaborative optimization scheduling result corresponding to the intraday collaborative optimization scheduling scheme is obtained.

7. A collaborative optimization scheduling device for integrated energy systems, characterized in that, The device includes: The data acquisition module is used to acquire multi-source heterogeneous data of the integrated energy system when a collaborative optimization scheduling request for the integrated energy system is received. The mechanism model construction module is used to construct a coupling mechanism model of the integrated energy system based on the multi-source heterogeneous data. The mechanism model solving module is used to solve the coupled mechanism model through a multi-agent reinforcement learning algorithm to obtain the collaborative optimization scheduling strategy of the integrated energy system. The collaborative optimization scheduling module is used to generate a collaborative optimization scheduling scheme corresponding to the collaborative optimization scheduling strategy through a three-level scheduling mechanism, and to optimize the collaborative optimization scheduling scheme during the execution of the collaborative optimization scheduling scheme to obtain the collaborative optimization scheduling result.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images