Rolling optimization method and related device for integrated energy system based on cloud-edge collaboration

Abstract

The application discloses a kind of based on cloud edge coordination's integrated energy system rolling optimization method and related device, belong to integrated energy system optimization technical field;Through acquisition system operation data and electric power market price, carbon emission intensity factor;Adopt recursive least square method on-line identification key node Thevenin equivalent impedance, combine equipment physical limit calculation time-varying available adjustment domain TVAR (t);Carbon electricity joint cost function of fusion time-sharing electricity price and real-time carbon intensity is constructed;With cost minimization as goal, with TVAR (t) as hard constraint, based on model predictive control framework rolling solving scheduling instruction;Through cloud-edge-end three-level architecture execution instruction, communication interruption when edge layer starts autonomous mode and guarantees key load power supply;Based on execution feedback correction model parameter realizes closed loop iteration.The application solves scheduling instruction and can reach 99.6% feasibility, carbon emission reduces 22.3%, when communication interruption, key load power failure time is less than 30 seconds, with high feasibility, low carbon and strong toughness, and engineering deployment cost is low.

Description

Technical Field

[0001] This invention belongs to the field of integrated energy system optimization technology, and relates to a rolling optimization method and related device for integrated energy systems based on cloud-edge collaboration. Background Technology

[0002] Integrated energy systems, by integrating multiple energy forms such as electricity, gas, and heat, achieve cascaded utilization and synergistic optimization of energy, and are the core carrier for the construction of new power systems. Integrated energy systems, represented by urban microgrids, are widely used in industrial parks, commercial complexes, and public buildings, and their operation optimization technology has become crucial for improving energy efficiency and reducing carbon emissions. However, existing integrated energy system optimization methods mostly adopt a two-stage paradigm, which is difficult to adapt to the operational requirements of dual uncertainties in source and load and complex communication environments, specifically exhibiting the following three bottlenecks: First, static modeling of regulation capacity leads to insufficient feasibility of optimization results. Traditional methods typically simplify equipment such as energy storage and gas turbines as adjustable resources with fixed capacity, pre-setting reserve capacity boundaries and ignoring the real-time impact of equipment dynamic constraints on regulation capacity. For example, the available capacity of energy storage changes in real time with the state of charge (SOC), the output of gas turbines is limited by the ramp rate, and the reactive power output of inverters is subject to voltage coupling constraints. Static modeling cannot capture these dynamic characteristics, resulting in optimized dispatch instructions often exceeding the actual regulation range of the equipment. Actual measurement data shows that the feasibility rate of dispatch instructions generated by traditional methods is low, easily leading to safety problems such as equipment overload and voltage exceeding limits, and even causing protection actions to cut off power supply. Second, carbon cost accounting is lagging and unable to respond to dynamic market signals. Existing optimization models mostly use daily averages or fixed coefficients to calculate carbon emission costs, which cannot match the frequently updated price signals of the electricity market. With the advancement of the electricity spot market, time-of-use pricing has been shortened to the 15-minute level, and the carbon emission intensity factor of provincial power grids also fluctuates in real time with the power supply structure. Static carbon cost models fail to reflect the optimization logic of reducing electricity consumption during high-carbon periods and increasing consumption during low-carbon periods. This leads to the possibility that while the system reduces electricity costs, it may indirectly increase carbon emissions, making it difficult to achieve synergistic optimization of economic efficiency and low carbon emissions. Third, the execution layer lacks closed-loop feedback and fault tolerance mechanisms. In traditional centralized optimization architectures, instructions generated in the cloud are directly sent to terminal devices without establishing a dynamic correction mechanism for execution deviations. When communication links are interrupted due to faults, the edge side lacks autonomous decision-making capabilities and must wait for manual intervention, resulting in power outages of critical loads lasting for several minutes. This problem is particularly prominent in complex urban electromagnetic environments or extreme weather conditions, failing to meet the needs of scenarios with extremely high power supply reliability requirements, such as hospitals and data centers. Furthermore, existing methods do not consider the dynamic redistribution of adjustment tasks when equipment fails, resulting in insufficient system resilience.

[0003] In recent years, cloud-edge collaborative architecture and model predictive control (MPC) have provided new solutions to the aforementioned problems. Cloud-edge collaboration combines global optimization with rapid local response by decentralizing computing power, while MPC can handle dynamic optimization problems with multiple constraints and variables through rolling time-domain optimization. However, existing research has not yet solved key issues such as online identification of dynamic adjustment capabilities, real-time coupling of carbon electricity market signals, and deep integration of edge autonomy fault tolerance mechanisms, making it difficult to translate theoretical results into engineering applications. Therefore, there is an urgent need for a rolling optimization method that integrates dynamic boundary awareness, carbon electricity market-driven approaches, and cloud-edge collaborative fault tolerance to improve the feasibility, economy, and resilience of integrated energy systems. Summary of the Invention

[0004] The purpose of this invention is to provide a rolling optimization method and related device for integrated energy systems based on cloud-edge collaboration, so as to solve the technical problems in the prior art where static modeling of regulation capacity leads to infeasibility of instructions and carbon cost lag makes it difficult to respond to high-frequency market signals.

[0005] To achieve the above objectives, the present invention employs the following technical solution: In a first aspect, the present invention provides a rolling optimization method for a comprehensive energy system based on cloud-edge collaboration, comprising the following steps: Acquire real-time operational data of the integrated energy system and external market data, preprocess the acquired data, and generate standardized data sequences; Collect bus voltage and current data at the grid connection point, identify the Thevenin equivalent impedance parameters of key nodes, and calculate the time-varying available regulation domain TVAR(t) in combination with the physical limit parameters of the equipment. Based on the standardized data sequence, a carbon-electric joint cost function is constructed. With the goal of minimizing the carbon-electric joint cost, and with constraints such as TVAR(t), power balance, equipment operation limitations, and network security, a rolling optimization model is constructed using a model predictive control framework. The model is solved by a mixed integer linear programming solver to generate a sequence of scheduling instructions for each device. The scheduling instruction sequence is sent from the cloud to the edge layer, where the edge layer performs local fast correction and fault tolerance processing. When communication is interrupted, the edge autonomous mode is activated, and the terminal layer executes specific device control. The terminal layer feeds back the actual execution results of specific device control to the edge layer. The edge layer calculates the execution deviation and uploads it to the cloud for model parameter updates, thereby achieving rolling optimization.

[0006] Furthermore, the steps of acquiring real-time operational data of the integrated energy system and external market data, preprocessing the acquired data, and generating standardized data sequences specifically include: Acquire real-time operating data of the integrated energy system, including voltage amplitude and phase angle of each node, current of each branch, output of each distributed power source, SOC status of energy storage, operating status of gas turbine, and load power. Simultaneously, external market data is obtained through cloud API interfaces, including time-of-use electricity prices in the electricity spot market and provincial carbon emission intensity factors. Outlier detection and filtering preprocessing are performed on the acquired real-time operational data and external market data to generate standardized data sequences.

[0007] Furthermore, the steps of collecting bus voltage and current data at the grid connection point, identifying the Thevenin equivalent impedance parameters of key nodes, and calculating the time-varying available adjustable domain TVAR(t) in conjunction with the equipment physical limit parameters specifically include: The voltage and current data of the bus at the grid connection point are collected. The raw data is digitally filtered to remove high-frequency noise and DC bias. Then, the Thevenin equivalent impedance parameters of the key nodes are identified online using the recursive least squares method. The time-varying available adjustment domain TVAR(t) is calculated by combining the physical limit parameters of the equipment; the available adjustment domain TVAR(t) is a four-dimensional feasible domain, which consists of the active power adjustment range and the reactive power adjustment range; Wherein, the upper limit of active power adjustment in the active power adjustment range = the sum of the maximum active power output of each distributed power source + the maximum discharge power of the energy storage device + the maximum power purchased from the grid - the total active load of the system; The lower limit of active power regulation in the active power regulation range = maximum power sold to the grid - total active load of the system - maximum charging power of the energy storage device; The upper limit of reactive power regulation in the reactive power regulation is the sum of the maximum reactive power output of each distributed power source + the maximum reactive power compensation capacity of the capacitor bank + the maximum inductive reactive power capacity of the static var generator. The lower limit of reactive power regulation in the reactive power regulation is equal to the sum of the maximum reactive power absorption capacity of each distributed power source and the maximum capacitive reactive power capacity of the static var generator.

[0008] Furthermore, the step of identifying the Thevenin equivalent impedance parameters of key nodes online using the recursive least squares method specifically includes: Initialize the parameter vector and covariance matrix. The parameter vector contains the resistive and reactance components of the Thevenin equivalent impedance, and the covariance matrix is ​​initialized to 10. 4 ~10 6 A positive real number multiplied by the identity matrix; Calculate the gain matrix at each sampling time: K ( k )= P (k- 1) φ ( k )[ λ + φ ( k ) P ( k- 1) φ ( k )] - ¹ Where K(k) is the gain matrix at time k, P(k-1) is the covariance matrix at the previous time, φ(k) is the regression vector at time k composed of the real and imaginary parts of the current measurement value, and λ is the forgetting factor of 0.95~0.995. Update parameter vector: θ ( k )= θ ( k - 1 )+ K ( k )[ u ( k )- φ ( k ) θ ( k -1)] Where θ(k) is the parameter vector at time k, and u(k) is the voltage measurement value at time k; Update the covariance matrix: P ( k )=[ I - K ( k ) φ ( k )]P( k -1) / λ Where I is the identity matrix; Set a convergence criterion: when the norm of the parameter change within a consecutive preset sampling period is less than 10. -4 ~10 -6 When the identification results converge, it is determined that the identification results have converged.

[0009] Furthermore, the expression for the carbon-electricity joint cost function is as follows:

[0010] in, The carbon price conversion factor is set by the user or automatically synchronized by the market. Let t be the power purchased by the integrated energy system from the public grid at time t. A positive value indicates that the power is purchased from the grid, and a negative value indicates that the power is sold to the grid. The time-of-use electricity price in the spot market at that moment; This represents the carbon emission intensity factor of the provincial power grid at that moment.

[0011] Furthermore, the parameters of the model prediction control framework are as follows: the prediction time domain is 4 to 24 hours, corresponding to 16 to 96 time steps, with each time step being 15 minutes; the control time domain is 1 to 4 steps; the rolling cycle is 5 to 15 minutes; and the single optimization solution time does not exceed 1 / 10 of the rolling cycle.

[0012] Furthermore, the local optimization method of the edge autonomous mode is as follows: based on the locally identified TVAR(t), with the goal of minimizing the load shedding amount or maximizing the renewable energy consumption, a heuristic algorithm or simplified linear programming method is used to perform local optimization at the second to minute level; after communication is restored, the edge layer automatically switches back to the cloud collaborative mode and uploads the operating data during the autonomous period for state synchronization and model correction.

[0013] Secondly, the present invention provides a cloud-edge collaborative integrated energy system rolling optimization system, comprising: The data acquisition module is used to acquire real-time operational data of the integrated energy system and external market data, preprocess the acquired data, and generate standardized data sequences. The online identification module is used to collect bus voltage and current data at the grid connection point, identify the Thevenin equivalent impedance parameters of key nodes, and calculate the time-varying available regulation domain TVAR(t) in combination with the physical limit parameters of the equipment. The rolling optimization solution module is used to construct a carbon-electric joint cost function based on the standardized data sequence. With the goal of minimizing the carbon-electric joint cost, and with constraints such as TVAR(t), power balance, equipment operation limitations, and network security, a rolling optimization model is constructed using a model predictive control framework. The model is solved by a mixed integer linear programming solver to generate a sequence of scheduling instructions for each device. The cloud-edge-device collaborative execution module is used to send the sequence of scheduling instructions from the cloud to the edge layer, where the edge layer performs local rapid correction and fault tolerance processing. When communication is interrupted, the edge autonomous mode is activated, and the terminal layer executes specific device control. The feedback update module is used to feed back the actual execution results of specific device control to the edge layer through the terminal layer. The edge layer calculates the execution deviation and uploads it to the cloud for model parameter updates, thereby achieving rolling optimization.

[0014] Thirdly, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the cloud-edge collaborative integrated energy system rolling optimization method described above.

[0015] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the cloud-edge collaborative integrated energy system rolling optimization method described above.

[0016] Compared with the prior art, the present invention has the following beneficial effects: This invention discloses a rolling optimization method and related device for integrated energy systems based on cloud-edge collaboration. Through an online dynamic adjustment capability identification mechanism, it identifies the Thevenin equivalent impedance of key nodes in real time using recursive least squares method. Combining this with equipment physical limits, it constructs a four-dimensional time-varying available adjustment domain (TVAR(t)) as a hard constraint for optimization, increasing the feasibility rate of scheduling commands from 78% to 99.6%, thus solving the command over-limit problem caused by traditional static modeling. Through a rolling optimization framework driven by carbon-electricity market signals, it couples 15-minute time-of-use electricity prices and provincial carbon emission intensity factors in real time, constructing a carbon-electricity joint cost function, achieving carbon-electricity joint cost under the same operating cost. Emissions are reduced by 22.3%, balancing economic efficiency and low carbon emissions. Relying on a cloud-edge-device three-level collaborative execution architecture, the edge layer has autonomous fault tolerance capabilities in the event of communication interruptions. In 5G interruption scenarios, the power outage time of critical loads is reduced from 5 minutes to less than 30 seconds, meeting the power supply reliability requirements of less than 100ms for primary loads such as hospital operating rooms and ICUs. At the same time, this invention has strong engineering adaptability, requiring only the addition of a TVAR identification module and edge agent to the existing EMS system for deployment, without the need to replace hardware. It combines technological advancement with engineering implementation value, providing a complete solution for the efficient, low-carbon, and resilient operation of integrated energy systems in complex environments. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the system of the present invention; Figure 3 This is a schematic diagram of the system architecture of an embodiment of the present invention. Detailed Implementation

[0019] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present application.

[0020] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0021] It should be understood that in the embodiments of this application, "at least one" means one or more, and "more than one" means two or more. "And / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. The character " / " generally indicates that the related objects before and after it are in an "or" relationship. "Contains A, B and / or C" means containing any one, two, or three of A, B, and C.

[0022] It should be understood that in the embodiments of this application, "B corresponding to A", "B corresponding to A", "A corresponds to B", or "B corresponds to A" means that B is associated with A, and B can be determined based on A. Determining B based on A does not mean that B is determined solely based on A; B can also be determined based on A and / or other information.

[0023] The cloud-edge collaborative integrated energy system rolling optimization method provided by this invention can be executed by an electronic device, such as a terminal or server. The terminal can be a smartphone, tablet, laptop, or other similar device. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. It is understood that this invention does not limit the specific entity executing this cloud-edge collaborative integrated energy system rolling optimization method.

[0024] The technical solution of this application will be described in detail below through specific embodiments. It should be noted that the following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments described below are used to explain the technical solution of this application and are not intended to limit actual use.

[0025] See Figure 1 This invention discloses a rolling optimization method for a comprehensive energy system based on cloud-edge collaboration, comprising the following steps: S1: Acquire real-time operating data of the integrated energy system and external market data, preprocess the acquired data, and generate standardized data sequences; S101, real-time operation data of the integrated energy system is acquired through the edge layer data acquisition unit. The real-time operation data includes the voltage amplitude and phase angle of each node, the current of each branch, the output of each distributed power source (photovoltaic, wind turbine, etc.), the SOC status of energy storage, the operating status of gas turbine, and the load power. S102, simultaneously obtain external market data through the cloud API interface, the external market data including time-of-use electricity prices in the electricity spot market. and provincial carbon emission intensity factor (Unit: kgCO2 / kWh); S103 performs outlier detection and filtering preprocessing on the acquired real-time operating data and external market data to generate a standardized data sequence.

[0026] S2, collect bus voltage and current data at the grid connection point, identify the Thevenin equivalent impedance parameters of key nodes, and calculate the time-varying available regulation domain TVAR(t) in combination with the physical limit parameters of the equipment. S201 collects bus voltage and current data at the grid connection point. The collected raw data is digitally filtered to remove high-frequency noise and DC bias. Then, the Thevenin equivalent impedance parameters of key nodes are identified online using the recursive least squares method. The specific implementation steps of the recursive least squares method are as follows: Initialize the parameter vector and covariance matrix. The parameter vector contains the resistive and reactance components of the Thevenin equivalent impedance, and the covariance matrix is ​​initialized to 10. 4 ~10 6 A positive real number multiplied by the identity matrix; At each sampling time, the gain matrix is ​​calculated. The gain matrix is ​​equal to the covariance matrix of the previous time step multiplied by the regression vector, and then divided by (the forgetting factor plus the transpose of the regression vector multiplied by the covariance matrix multiplied by the regression vector). K ( k )= P ( k- 1) φ ( k )[ λ + φ ( k ) P ( k- 1) φ ( k )] - ¹ Where K(k) is the gain matrix at time k, P(k-1) is the covariance matrix at the previous time, φ(k) is the regression vector at time k composed of the real and imaginary parts of the current measurement value, and λ is the forgetting factor of 0.95~0.995. The updated parameter vector is equal to the previous parameter vector plus the gain matrix multiplied by (voltage measurement minus the transpose of the regression vector multiplied by the previous parameter vector): θ ( k )= θ ( k - 1 )+ K ( k )[ u ( k )- φ ( k ) θ ( k -1)] Where θ(k) is the parameter vector at time k, and u(k) is the voltage measurement value at time k; The updated covariance matrix is ​​equal to (identity matrix minus gain matrix multiplied by the transpose of the regression vector) multiplied by the previous covariance matrix and then divided by the forgetting factor. P ( k )=[ I - K ( k ) φ ( k )]P( k -1) / λ Where I is the identity matrix; The convergence criterion of the RLS algorithm is as follows: when the norm of the parameter change within N consecutive sampling periods is less than a preset threshold, the identification result is considered convergent, where N is usually between 10 and 20, and the threshold is usually 10. -4 ~10 -6 The update frequency of the identification algorithm is consistent with the data sampling frequency, typically from 100 milliseconds to 1 second, to ensure that TVAR can reflect changes in the system's operating status in real time. The physical meaning of Thevenin equivalent impedance is to characterize the equivalent electrical strength of the system at that node. The smaller the impedance, the larger the short-circuit capacity and the stronger the voltage support capability of the system. By identifying the Thevenin equivalent impedance online, the real-time operating status of the system can be accurately grasped, avoiding misjudgment of regulation capability caused by changes in system topology or fluctuations in the output of distributed power sources. The TVAR serves as a hard constraint for rolling optimization, ensuring that scheduling instructions are always within the feasible domain.

[0027] S202, calculate the time-varying adjustable region (TVAR(t)) by combining the physical limit parameters of the equipment (inverter reactive power limit, MGT ramp rate, energy storage charging and discharging power limit, etc.); the adjustable region TVAR(t) is a four-dimensional feasible region, which consists of the active power adjustment range and the reactive power adjustment range; wherein the lower limit of active power adjustment is the lower limit of the system active power adjustment at this time (unit: kW), the upper limit of active power adjustment is the upper limit of the system active power adjustment at this time (unit: kW), the lower limit of reactive power adjustment is the lower limit of the system reactive power adjustment at this time (unit: kvar), and Q_max(t) is the upper limit of the system reactive power adjustment at time t (unit: kvar). The calculation method for the active power regulation range is as follows: the upper limit of the system active power regulation is equal to the sum of the maximum active power output of each distributed power source, plus the maximum discharge power of the energy storage device, plus the maximum power purchased from the grid, minus the total active load of the system; the lower limit of the system active power regulation is equal to the maximum power sold to the grid, minus the total active load of the system, minus the maximum charging power of the energy storage device; where each distributed power source includes adjustable power sources such as photovoltaics, wind turbines, and gas turbines; The calculation method for the reactive power regulation range is as follows: the upper limit of reactive power regulation of the system is equal to the sum of the maximum reactive power output of each distributed power source, plus the maximum reactive power compensation capacity of the capacitor bank, plus the maximum inductive reactive power capacity of the static var generator; the lower limit of reactive power regulation of the system is equal to the negative sum of the maximum reactive power absorption capacity of each distributed power source, minus the maximum capacitive reactive power capacity of the static var generator; through the four-dimensional feasible domain of active and reactive power, the real-time regulation capability boundary of the system is fully characterized.

[0028] S3, Construct a carbon-electricity joint cost function based on the standardized data sequence; The expression for the carbon-electricity joint cost function is as follows:

[0029] in, The carbon price conversion factor is set by the user or automatically synchronized by the market. Let t be the power (in kW) that the integrated energy system purchases from the public grid at time t. A positive value indicates that the power is purchased from the grid, and a negative value indicates that the power is sold to the grid. The time-of-use electricity price in the spot market at that moment (unit: yuan / kWh); This represents the carbon emission intensity factor of the provincial power grid at that moment (unit: kgCO2 / kWh).

[0030] The rolling optimization model adopts a model predictive control (MPC) framework, and its complete objective function is: to minimize the total operating cost within the prediction time domain, where the total operating cost equals the sum of the electricity purchase cost, carbon emission cost, energy storage loss cost, and distributed power generation cost at each time step; the number of prediction time steps is typically 16 to 96 steps, corresponding to 4 to 24 hours, with each time step lasting 15 minutes; the unit loss cost of energy storage charging and discharging includes battery depreciation cost and charging and discharging efficiency loss; the unit power generation cost of distributed power includes fuel cost and operation and maintenance cost; The constraints of the rolling optimization model include: (1) Power balance constraint: The purchased power plus the active power output of each distributed power source plus the active power of energy storage equals the total active load of the system, where energy storage discharge is positive and charging is negative; (2) TVAR constraint: The purchased power and reactive power must be within the time-varying available adjustment domain to ensure that the dispatching instructions are within the feasible domain; (3) Equipment operation constraints: including upper and lower limits of distributed power output, upper and lower limits of energy storage SOC (state of charge), upper and lower limits of energy storage charging and discharging power, energy storage ramp rate constraints, etc.; (4) Network constraints: including upper and lower limits of node voltage, upper and lower limits of branch current, etc. The rolling strategy of the MPC framework is as follows: at each control time, based on the current system state measurement value and the prediction information of the number of future prediction time-domain steps, the above optimization problem is solved to obtain the optimal control sequence in the future prediction time domain; then only the instruction of the first control step is executed, and at the next control time, the system state is remeasured and the prediction information is updated, and the optimization problem is solved again, thus rolling iteratively; the control time domain is usually 1 to 4 steps, that is, each optimization only executes the instructions of the first few steps; the rolling cycle is usually 5 to 15 minutes to balance optimization accuracy and computational efficiency; The optimization algorithm employs mixed-integer linear programming (MILP), and a piecewise linearization method is used to handle nonlinear constraints (such as the PQ coupling characteristics of the inverter). Solvers can be Gurobi, CPLEX, or open-source algorithms such as GLPK and CBC. To ensure real-time performance, the time for a single optimization solution should not exceed 1 / 10 of the rolling cycle; that is, for a 15-minute rolling cycle, the solution time should be controlled within 1.5 minutes. For large-scale systems, distributed optimization methods or parallel computing techniques can be used to improve solution efficiency. S4 aims to minimize the cost of carbon-electricity co-production, and uses TVAR(t), power balance, equipment operation limitations and network security as constraints. It adopts a model predictive control framework to build a rolling optimization model, and solves it through a mixed integer linear programming solver to generate a sequence of scheduling instructions for each device. This invention employs a model predictive control (MPC) framework to construct a rolling optimization model; the optimization time window is set to the next 4 hours, and it is updated every 15 minutes; the optimization problem is solved using a mixed integer linear programming (MILP) solver to generate a sequence of scheduling instructions for each device, including energy storage charging and discharging power, gas turbine output, and reactive power compensation device switching, etc. S5 sends a sequence of scheduling instructions to the edge layer via the cloud. The edge layer performs local fast correction and fault tolerance processing. When communication is interrupted, the edge autonomous mode is activated, and local optimization is performed based on local TVAR. The terminal layer executes specific device control, including real-time adjustment of inverters, energy storage converters, circuit breakers, etc. This invention adopts a three-level collaborative execution architecture of "cloud-edge-device", as shown in Table 1 below: Table 1

[0031] The cloud-edge-device three-level collaborative architecture uses the MQTT protocol combined with the OPC UA standard for communication. The cloud and edge layers communicate via 5G or fiber optic Ethernet with a communication cycle of 1 to 5 seconds. The edge layer and terminal devices communicate via industrial Ethernet or wireless communication methods such as LoRa and ZigBee with a communication cycle of 100ms to 1 second. The data format uses JSON or Protobuf for serialization to ensure the efficiency and compatibility of data transmission. The triggering conditions for edge autonomous mode include: (1) Communication interruption detection: When the edge layer does not receive instructions from the cloud for N consecutive communication cycles, the edge autonomous mode is triggered, where N is usually 3 to 5 cycles; (2) Instruction anomaly detection: When the cloud instruction exceeds the TVAR feasible domain or has obvious errors, the edge layer refuses to execute and starts local optimization; (3) Emergency event triggering: When an emergency such as voltage over-limit or frequency abnormality is detected, the edge layer immediately starts local protection control without waiting for instructions from the cloud. The local optimization method in the edge autonomy mode is as follows: based on the local real-time identified TVAR, with the goal of minimizing load shedding or maximizing renewable energy consumption, a heuristic algorithm or a simplified linear programming method is used for local optimization; the optimization time scale is from seconds to minutes to ensure that the system can still operate safely and stably during communication interruption; when communication is restored, the edge layer automatically switches back to the cloud collaborative mode and uploads the operation data during the autonomy period to the cloud for state synchronization and model correction. The data security and privacy protection mechanism adopts a layered encryption strategy: the transport layer uses the TLS 1.3 protocol for encryption, and the application layer uses the AES-256 symmetric encryption algorithm to encrypt sensitive data; identity authentication adopts a two-way authentication mechanism based on digital certificates to ensure the trustworthiness of the identities of both communicating parties; at the same time, data anonymization technology is used to anonymize user data uploaded to the cloud to protect user privacy. It also includes fault tolerance mechanisms and fault recovery strategies: (1) Device-level fault tolerance: When a terminal device fails, the edge layer automatically removes it from the optimization model and reassigns adjustment tasks; (2) Edge layer fault tolerance: The main and backup dual-machine hot standby architecture is adopted. When the main edge node fails, the backup node takes over control within 100ms; (3) Cloud fault tolerance: Distributed cluster deployment is adopted to support load balancing and automatic fault migration. After the fault is recovered, the system automatically performs state synchronization and parameter calibration to ensure a smooth transition to normal operation mode. S6 feeds back the actual execution results of specific device control to the edge layer through the terminal layer. The edge layer calculates the execution deviation and uploads it to the cloud. The cloud updates the system state estimate based on the feedback data and corrects the initial conditions for the next cycle of optimization. At the same time, it updates the parameters of the TVAR recognition model to achieve closed-loop iteration of rolling optimization.

[0032] See Figure 2 and Figure 3 This invention discloses a cloud-edge collaborative integrated energy system rolling optimization system, including a data acquisition module, an online identification module, a rolling optimization solution module, a cloud-edge-device collaborative execution module, and a feedback update module. The system comprises the following modules: a data acquisition module for acquiring real-time operational data of the integrated energy system and external market data, preprocessing the acquired data to generate a standardized data sequence; an online identification module for collecting bus voltage and current data at the grid connection point, identifying the Thevenin equivalent impedance parameters of key nodes, and calculating the time-varying available regulation domain (TVAR(t)) in conjunction with equipment physical limit parameters; a rolling optimization solution module for constructing a carbon-electricity joint cost function based on the standardized data sequence, aiming to minimize the carbon-electricity joint cost, and using TVAR(t), power balance, equipment operation limitations, and network security as constraints, constructing a rolling optimization model using a model predictive control framework, solving it through a mixed-integer linear programming solver, and generating a scheduling instruction sequence for each device; a cloud-edge-device collaborative execution module for distributing the scheduling instruction sequence from the cloud to the edge layer, where the edge layer performs local rapid correction and fault tolerance processing, and initiates an edge autonomous mode when communication is interrupted, with the terminal layer executing specific device control; and a feedback update module for feeding back the actual execution results of specific device control from the terminal layer to the edge layer, where the edge layer calculates the execution deviation and uploads it to the cloud for model parameter updates, thus achieving rolling optimization.

[0033] Example: This embodiment uses a provincial hospital microgrid as an example (different from a conventional industrial park). Special requirements: the operating room and ICU are primary loads, with a power outage tolerance of <100ms; configuration: 1.5MW rooftop photovoltaic + 800kW diesel generator (cold standby) + 1.2MWh cascaded battery + combined cooling, heating and power (CCHP) unit; specific implementation steps are as follows: 1. Market Data Acquisition and Preprocessing: Every 15 minutes, the central optimizer obtains real-time market data from the provincial power trading center via a dedicated API interface. This data includes the daily 96-point time-of-use electricity price curve, the next day's predicted electricity price, and the provincial power grid's real-time carbon emission intensity factor. The data acquisition process employs two-way authentication and encrypted transmission mechanisms to ensure data security and reliability. After acquiring the data, the central optimizer performs integrity checks and outlier detection on the raw data. Missing data is filled using interpolation, and outlier data is replaced with historical averages, ultimately generating a standardized market data sequence that provides a reliable data foundation for subsequent optimization calculations. Simultaneously, the central optimizer also obtains meteorological forecast data such as solar irradiance, temperature, and wind speed for the next 24 hours from the meteorological department's API for photovoltaic and wind turbine output forecasting. 2. Online Identification of Dynamic Adjustment Capability: The edge agent collects grid connection point bus voltage and current data every 5 seconds through a high-precision synchronous phasor measurement unit, with a sampling frequency of 100Hz and a data accuracy of 0.2%. The collected raw data is first digitally filtered to remove high-frequency noise and DC bias. Then, the Thevenin equivalent impedance parameters of key nodes, including resistive and reactive components, are identified online using the recursive least squares method. The forgetting factor of the RLS algorithm is set to 0.98 to balance the weights of historical and new data. The covariance matrix is ​​initialized to 10 to the power of 5 multiplied by the identity matrix to ensure rapid convergence of the algorithm. When the parameter change is less than a preset threshold within 15 consecutive sampling periods, the identification result is considered converged. Combining the physical limit parameters of the equipment, including inverter reactive power limit, micro gas turbine ramp rate, energy storage charging and discharging power limit, and capacitor bank switching capacity, the time-varying available adjustment domain (TVAR) is calculated as a hard constraint for subsequent optimization. The TVAR calculation result is updated every 5 seconds to ensure that the changes in the system operating status can be reflected in real time. 3. Communication Interruption Detection and Edge Autonomy Trigger: When a power grid failure causes a 5G communication interruption, the edge agent monitors the communication link status in real time through a heartbeat detection mechanism, sending a heartbeat packet to the cloud every second. If no confirmation response is received from the cloud three times consecutively, the edge agent determines that the communication is interrupted within 10 seconds and immediately triggers the edge autonomy mode. After the edge autonomy mode is activated, the edge agent first sends a hot start command to the diesel generator and simultaneously disconnects the grid connection point circuit breaker, preparing to enter the islanded operation mode. After receiving the start command, the diesel generator completes the start-up and voltage build-up process within 15 seconds, reaching the rated voltage and frequency output. During this transition period, the energy storage battery provides short-term power support to ensure uninterrupted power supply to critical loads. 4. Seamless Switching and Load Tiering Assurance: The end-side controller employs a fast solid-state switching switch, completing a seamless switch from grid power to diesel generator power within 200 milliseconds. During the switching process, the end-side controller performs tiered control based on load priority, first ensuring uninterrupted power supply to primary loads such as operating rooms and ICUs, then ensuring secondary loads such as medical equipment and lighting, and finally considering tertiary loads such as air conditioners and water heaters. For non-critical loads, appropriate disconnection can be performed according to the system's power supply capacity to ensure stable system operation. After the switch is completed, the end-side controller monitors the operating status of the diesel generator and the voltage and current of each bus in real time to ensure that the power supply quality meets the requirements of medical equipment; voltage deviation is controlled within ±5%, and frequency deviation is controlled within ±0.2Hz. 5. Rolling Optimization Modeling and Solution: Under normal operating conditions, the central optimizer uses the real-time identified TVAR as the hard constraint of the feasible region, and minimizes the cost of carbon-electricity co-operation as the objective function. A rolling optimization model is constructed using a model predictive control framework. The optimization time window is set to the next 4 hours, with a total of 16 time steps, each step lasting 15 minutes. The objective function includes four parts: electricity purchase cost, carbon emission cost, energy storage loss cost, and distributed power generation cost. The constraints include four categories: power balance constraints, TVAR constraints, equipment operation constraints, and network security constraints. Among them, equipment operation constraints are further subdivided into... The optimization model includes upper and lower limits for distributed power output, upper and lower limits for energy storage state of charge, upper and lower limits for energy storage charging and discharging power, and limits for energy storage ramp-up rate. A mixed-integer linear programming method is used to solve the optimization model, and a piecewise linearization method is used to approximate nonlinear constraints such as inverter PQ coupling characteristics. A high-performance commercial solver is selected, with the single optimization solution time controlled within 30 seconds to ensure real-time performance. After the solution is completed, a sequence of scheduling instructions for each device is generated, including energy storage charging and discharging power curves, combined cooling, heating and power (CCHP) unit output plans, reactive power compensation device switching strategies, and interruptible load scheduling schemes. 6. Cloud-Edge Collaborative Command Issuance and Verification: The cloud sends optimized scheduling commands to the edge agent via a dedicated 5G network, with a communication cycle of 5 seconds. MQTT protocol is used for message transmission to ensure low latency and high reliability. Upon receiving the command from the cloud, the edge agent first verifies its validity, including checking if the command is within the current TVAR feasible domain, meets device physical constraints, and has no obvious logical errors. If the command passes verification, the edge agent forwards it to each terminal device for execution. If the command fails verification, the edge agent refuses to execute it and returns the reason for the failure to the cloud, while simultaneously activating a local backup optimization strategy. Terminal devices include energy storage converters, photovoltaic inverters, diesel generator controllers, reactive power compensation device controllers, load controllers, etc. These devices are connected to the edge agent via industrial Ethernet or RS485 bus, achieving millisecond-level command execution accuracy. 7. Feedback Correction and Closed-Loop Iteration: During the execution of scheduling instructions, terminal devices collect real-time operating status data, including output power, voltage, current, temperature, and operating status, and feed this data back to the cloud via an edge agent once per second. Upon receiving the feedback data, the cloud calculates the deviation between the actual execution result and the optimization target, including power deviation, cost deviation, and voltage deviation. Based on the magnitude of the deviation, the cloud corrects the initial conditions for the next optimization cycle, including updating system state estimates, correcting prediction model parameters, and adjusting constraint boundaries. Simultaneously, the cloud uses the feedback data to update and calibrate the TVAR identification model online, improving identification accuracy. Through this rolling optimization-execution-feedback-correction closed-loop mechanism, the system can continuously adapt to changes in operating conditions and continuously optimize operating performance. Furthermore, the cloud performs big data analysis on historical operating data to uncover patterns in device characteristics, optimize the prediction model, and further improve the overall system performance. 8. Implementation Effect Verification Plan: To verify the actual effect of the present invention, a three-month field test was conducted on the provincial hospital's microgrid system. During the test, the system ran continuously for 92 days, collecting over 1.3 million operational data points. The test content included five aspects: feasibility verification of dispatch instructions, carbon emission reduction effect assessment, communication interruption reliability test, photovoltaic absorption rate test, and voltage qualification rate test. The test method adopted a comparative experiment method, that is, under the same load and weather conditions, the traditional static optimization scheme and the dynamic rolling optimization scheme of the present invention were run respectively, and the performance indicators were compared. During the test, a high-precision power quality analyzer was used to synchronously measure electrical quantities such as voltage, current, and power, with a measurement accuracy of 0.1 level. At the same time, a dedicated data acquisition system recorded all dispatch instructions and equipment operating status, with a sampling frequency of 1Hz, to ensure the integrity and accuracy of the test data. 9. Economic Analysis Results: The hospital's microgrid system consumes approximately 8 million kWh of electricity annually, with primary load accounting for about 35%, secondary load about 45%, and tertiary load about 20%. After adopting the solution of this invention, annual electricity costs will be reduced by approximately RMB 864,000, specifically as follows: RMB 420,000 is saved through peak-valley price arbitrage by charging during off-peak hours and discharging during peak hours, fully utilizing the peak-shaving and valley-filling capabilities of energy storage; RMB 264,000 is saved in carbon costs by optimizing operation to reduce carbon emissions and lower carbon trading costs; RMB 180,000 is saved by reducing equipment wear and tear by optimizing charging and discharging strategies to extend battery life and reduce equipment maintenance costs; the initial system investment is approximately RMB 3.2 million, including edge proxy servers, 5G communication modules, software system upgrades, and sensor installations; the static investment payback period is approximately 3.7 years, far lower than the industry average, demonstrating good economic benefits and investment value; furthermore, the system can generate additional revenue by participating in demand response and ancillary service markets, further shortening the investment payback period. 10. Carbon Emission Reduction Benefit Analysis: The hospital's microgrid system has a baseline annual carbon emission of approximately 6,400 tons of carbon dioxide, mainly from grid power purchases and diesel generator power generation. After adopting the proposed solution, the annual carbon emission reduction is approximately 1,427 tons of carbon dioxide, representing a reduction of 22.3%. The main pathways to carbon emission reduction include: increasing photovoltaic absorption rates to reduce thermal power purchases, optimizing energy storage operation strategies to reduce peak-hour high-carbon power procurement, and improving energy efficiency to reduce total energy consumption. From an environmental perspective, the annual emission reduction is equivalent to the carbon sequestration capacity of planting approximately 79,000 trees, or reducing the annual emissions of approximately 620 gasoline-powered cars, or saving approximately 583 tons of standard coal. From an economic perspective, the carbon emission reduction can generate additional revenue through the carbon trading market. Based on the current carbon price of approximately 50 yuan per ton, the annual carbon trading revenue is approximately 71,000 yuan. As carbon prices gradually rise, the economic benefits of carbon emission reduction will further increase. 11. A detailed comparison with traditional solutions reveals the following advantages of this invention: (1) Feasibility rate of scheduling instructions: The traditional static optimization scheme is 78%, mainly because it is impossible to accurately grasp the real-time operating status of the system, resulting in some scheduling instructions exceeding the actual adjustment capacity of the equipment; the dynamic TVAR scheme of this invention is 99.6%, which is 21.6 percentage points higher, and basically achieves 100% feasibility of scheduling instructions; (2) Carbon emission reduction effect: The traditional optimization scheme that only considers electricity price has an annual carbon emission of 6,400 tons, while the carbon-electricity combined optimization scheme of this invention has an annual carbon emission of 4,973 tons, with an emission reduction ratio of 22.3%; (3) Communication interruption reliability: In the traditional centralized solution, the power loss time of critical loads during communication interruption is 5 minutes, which requires manual activation of backup power; the edge autonomous solution of this invention is within 30 seconds, which is more than 90% shorter and meets the power supply reliability requirements of the hospital's first-level load. (4) Photovoltaic absorption rate: The photovoltaic absorption rate of the traditional scheme is 82%, which has a certain curtailment phenomenon; the scheme of this invention is 95.7%, which is 13.7 percentage points higher and significantly improves the utilization rate of renewable energy. (5) Voltage qualification rate: The voltage qualification rate of the traditional solution is 96.5%, which poses a risk of voltage exceeding the limit; the solution of this invention is 99.9%, which is 3.4 percentage points higher, and the power supply quality is significantly improved. 12. Typical Scenario Test Results: To verify the system's performance under different operating conditions, four typical scenarios were selected for specific testing: (1) Sunny noon scenario: The photovoltaic output reaches a peak of 1.5MW, exceeding the system load, and the power backflow phenomenon occurs; the solution of this invention adjusts the reactive power compensation capacity in real time through TVAR, controls the grid connection point voltage within 1.02 per unit value, avoids overvoltage protection action, and ensures full photovoltaic consumption; (2) Evening peak load scenario: The load reaches a peak of 2.8MW, and the electricity price is at its peak. The solution of this invention prioritizes the use of energy storage discharge and combined cooling, heating and power units to reduce the power grid purchase of electricity by 23%, effectively reducing peak load costs. (3) Power grid fault island scenario: When the power grid suddenly fails, the edge agent detects the fault within 10 seconds and starts the diesel generator. The seamless switch is completed within 200 milliseconds. The power supply to the operating room, ICU and other primary loads is not interrupted, which fully meets the high reliability requirements of medical loads. (4) Communication interruption scenario: Simulate 5G communication interruption for 30 minutes, the edge autonomous mode operates normally, and the power supply reliability of critical loads reaches 99.999%; after communication is restored, the system automatically synchronizes data and smoothly switches back to cloud collaborative mode, without any impact or disturbance in the whole process.

[0034] In one embodiment of the present invention, a computer device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions from the computer storage medium to achieve a corresponding method flow or corresponding function. The processor described in this embodiment of the present invention can be used in the operation of a cloud-edge collaborative integrated energy system rolling optimization method.

[0035] This invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the rolling optimization method for a cloud-edge collaborative integrated energy system in the above embodiments.

[0036] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0037] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0038] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1The function specified in one or more boxes.

[0039] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A rolling optimization method for a comprehensive energy system based on cloud-edge collaboration, characterized in that, Includes the following steps: Acquire real-time operational data of the integrated energy system and external market data, preprocess the acquired data, and generate standardized data sequences; Collect bus voltage and current data at the grid connection point, identify the Thevenin equivalent impedance parameters of key nodes, and calculate the time-varying available regulation domain TVAR(t) in combination with the physical limit parameters of the equipment. Based on the standardized data sequence, a carbon-electric joint cost function is constructed. With the goal of minimizing the carbon-electric joint cost, and with constraints such as TVAR(t), power balance, equipment operation limitations, and network security, a rolling optimization model is constructed using a model predictive control framework. The model is solved by a mixed integer linear programming solver to generate a sequence of scheduling instructions for each device. The scheduling instruction sequence is sent from the cloud to the edge layer, where the edge layer performs local fast correction and fault tolerance processing. When communication is interrupted, the edge autonomous mode is activated, and the terminal layer executes specific device control. The terminal layer feeds back the actual execution results of specific device control to the edge layer. The edge layer calculates the execution deviation and uploads it to the cloud for model parameter updates, thereby achieving rolling optimization.

2. The rolling optimization method for a comprehensive energy system based on cloud-edge collaboration according to claim 1, characterized in that, The steps of acquiring real-time operational data and external market data of the integrated energy system, preprocessing the acquired data, and generating standardized data sequences specifically include: Acquire real-time operating data of the integrated energy system, including voltage amplitude and phase angle of each node, current of each branch, output of each distributed power source, SOC status of energy storage, operating status of gas turbine, and load power. Simultaneously, external market data is obtained through cloud API interfaces, including time-of-use electricity prices in the electricity spot market and provincial carbon emission intensity factors. Outlier detection and filtering preprocessing are performed on the acquired real-time operational data and external market data to generate standardized data sequences.

3. The rolling optimization method for a comprehensive energy system based on cloud-edge collaboration according to claim 1, characterized in that, The steps of collecting bus voltage and current data at the grid connection point, identifying the Thevenin equivalent impedance parameters of key nodes, and calculating the time-varying available adjustable domain TVAR(t) in conjunction with the equipment's physical limit parameters specifically include: The voltage and current data of the bus at the grid connection point are collected. The raw data is digitally filtered to remove high-frequency noise and DC bias. Then, the Thevenin equivalent impedance parameters of the key nodes are identified online using the recursive least squares method. The time-varying available adjustment domain TVAR(t) is calculated by combining the physical limit parameters of the equipment; the available adjustment domain TVAR(t) is a four-dimensional feasible domain, which consists of the active power adjustment range and the reactive power adjustment range; Wherein, the upper limit of active power adjustment in the active power adjustment range = the sum of the maximum active power output of each distributed power source + the maximum discharge power of the energy storage device + the maximum power purchased from the grid - the total active load of the system; The lower limit of active power regulation in the active power regulation range = maximum power sold to the grid - total active load of the system - maximum charging power of the energy storage device; The upper limit of reactive power regulation in the reactive power regulation is the sum of the maximum reactive power output of each distributed power source + the maximum reactive power compensation capacity of the capacitor bank + the maximum inductive reactive power capacity of the static var generator. The lower limit of reactive power regulation in the reactive power regulation is equal to the sum of the maximum reactive power absorption capacity of each distributed power source and the maximum capacitive reactive power capacity of the static var generator.

4. The rolling optimization method for a comprehensive energy system based on cloud-edge collaboration according to claim 3, characterized in that, The steps for online identification of the Thevenin equivalent impedance parameters of key nodes using the recursive least squares method specifically include: Initialize the parameter vector and covariance matrix. The parameter vector contains the resistive and reactance components of the Thevenin equivalent impedance, and the covariance matrix is ​​initialized to 10. 4 ~10 6 A positive real number multiplied by the identity matrix; Calculate the gain matrix at each sampling time: K ( k )= P ( k- 1) φ ( k )[ λ + φ ( k ) P ( k- 1) φ ( k )] - ¹ Where K(k) is the gain matrix at time k, P(k-1) is the covariance matrix at the previous time, φ(k) is the regression vector at time k composed of the real and imaginary parts of the current measurement value, and λ is the forgetting factor of 0.95~0.

995. Update parameter vector: θ ( k )= θ ( k - 1 )+ K ( k )[ u ( k )- φ ( k ) θ ( k -1)] Where θ(k) is the parameter vector at time k, and u(k) is the voltage measurement value at time k; Update the covariance matrix: P ( k )=[ I - K ( k ) φ ( k )]P( k -1) / λ Where I is the identity matrix; Set a convergence criterion: when the norm of the parameter change within a consecutive preset sampling period is less than 10. -4 ~10 -6 When the identification results converge, it is determined that the identification results have converged.

5. The rolling optimization method for a comprehensive energy system based on cloud-edge collaboration according to claim 1, characterized in that, The expression for the carbon-electricity joint cost function is as follows: in, The carbon price conversion factor is set by the user or automatically synchronized by the market. Let t be the power purchased by the integrated energy system from the public grid at time t. A positive value indicates that the power is purchased from the grid, and a negative value indicates that the power is sold to the grid. The time-of-use electricity price in the spot market at that moment; This represents the carbon emission intensity factor of the provincial power grid at that moment.

6. The rolling optimization method for a comprehensive energy system based on cloud-edge collaboration according to claim 1, characterized in that, The parameters of the model prediction control framework are as follows: the prediction time domain is 4 to 24 hours, corresponding to 16 to 96 time steps, with each time step being 15 minutes; the control time domain is 1 to 4 steps; the rolling cycle is 5 to 15 minutes; and the single optimization solution time does not exceed 1 / 10 of the rolling cycle.

7. The rolling optimization method for a comprehensive energy system based on cloud-edge collaboration according to claim 1, characterized in that, The local optimization method of the edge autonomy mode is as follows: based on the locally identified TVAR(t), with the goal of minimizing the load shedding amount or maximizing the renewable energy consumption, a heuristic algorithm or a simplified linear programming method is used to perform local optimization at the second to minute level. After communication is restored, the edge layer automatically switches back to cloud collaboration mode and uploads the operational data during the autonomous period for status synchronization and model correction.

8. A rolling optimization system for integrated energy systems based on cloud-edge collaboration, characterized in that, include: The data acquisition module is used to acquire real-time operational data of the integrated energy system and external market data, preprocess the acquired data, and generate standardized data sequences. The online identification module is used to collect bus voltage and current data at the grid connection point, identify the Thevenin equivalent impedance parameters of key nodes, and calculate the time-varying available regulation domain TVAR(t) in combination with the physical limit parameters of the equipment. The rolling optimization solution module is used to construct a carbon-electric joint cost function based on the standardized data sequence. With the goal of minimizing the carbon-electric joint cost, and with constraints such as TVAR(t), power balance, equipment operation limitations, and network security, a rolling optimization model is constructed using a model predictive control framework. The model is solved by a mixed integer linear programming solver to generate a sequence of scheduling instructions for each device. The cloud-edge-device collaborative execution module is used to send the sequence of scheduling instructions from the cloud to the edge layer, where the edge layer performs local rapid correction and fault tolerance processing. When communication is interrupted, the edge autonomous mode is activated, and the terminal layer executes specific device control. The feedback update module is used to feed back the actual execution results of specific device control to the edge layer through the terminal layer. The edge layer calculates the execution deviation and uploads it to the cloud for model parameter updates, thereby achieving rolling optimization.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the rolling optimization method for a cloud-edge collaborative integrated energy system as described in any one of claims 1-7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the rolling optimization method for a cloud-edge collaborative integrated energy system as described in any one of claims 1-7.

Images