A building group integrated energy system dynamic collaborative optimization control method and system

Abstract

This invention proposes a dynamic collaborative optimization control method and system for integrated energy systems in building complexes. The method includes: constructing a system operation optimization model based on mixed-integer linear programming theory and multi-source heterogeneous data of the building energy system; dynamically generating high and low electricity price thresholds to trigger demand response actions based on real-time electricity price data in the predicted time domain and a preset flexibility intensity adjustment factor; solving the system operation optimization model in parallel under a baseline operating scenario and a flexibility response scenario; calculating the flexibility-specific marginal cost based on the solution results; if the flexibility-specific marginal cost is lower than the clearing price or willingness-to-pay threshold, determining to execute the optimal control command sequence corresponding to the flexibility response scenario; otherwise, executing the control command sequence corresponding to the baseline operating scenario, thereby improving the reliability of dynamic collaborative optimization control of integrated energy systems in building complexes.

Description

Technical Field

[0001] This invention relates to the field of interdisciplinary technology of smart grid and building energy management, and in particular to a dynamic collaborative optimization control method and system for integrated energy systems of building complexes. Background Technology

[0002] As global efforts to address climate change deepen and the energy structure undergoes profound transformation, the proportion of intermittent renewable energy sources, such as wind and solar power, in the power system is growing at an unprecedented rate. The output of these energy sources is characterized by significant randomness, volatility, and intermittency, posing serious challenges to the real-time supply-demand balance, frequency stability, and power quality of the power system. To address this challenge, the need for power systems to provide flexible regulation capabilities for demand-side resources is becoming increasingly urgent. Buildings, as one of the world's largest energy consumption terminals, account for approximately 40% of total social energy consumption, with heating, ventilation, and air conditioning (HVAC) systems being the primary energy-consuming component. Due to the significant thermal inertia inherent in building envelopes, they can act like a giant, low-cost virtual thermal battery, storing or releasing heat over a certain period. Therefore, buildings are widely recognized as a huge potential resource for providing demand response (DR) services and supporting the stable operation of the power grid.

[0003] However, in transforming the theoretical potential of building complexes into a practically dispatchable commercially flexible resource, existing technologies still face several pressing technical bottlenecks that severely restrict the effective participation of buildings in the energy market and the realization of their economic value.

[0004] First, existing technologies generally lack precise and standardized quantitative assessment methods for the costs of building flexibility invocation. Most current mainstream Model Predictive Control (MPC) methods focus solely on minimizing total system operating costs. While these methods can respond to grid price signals or peak-shaving commands to some extent—for example, using more electricity when prices are low and less when prices are high—they fail to clearly answer a core business question: how much additional cost or benefit does the building user incur to provide this flexibility response, such as reducing 1kW of load for one hour? This additional cost may stem from equipment operating outside its most efficient condition, extra heat loss during preheating or precooling processes, and slight compromises in user comfort. Lacking a clear marginal cost indicator measured in yuan per kilowatt-hour, building users cannot make scientific and accurate bids when participating in the electricity ancillary services market. They often fall into the trap of blindly responding or incurring losses due to information asymmetry, which significantly dampens their enthusiasm for participating in demand response.

[0005] Secondly, the physical models in existing control strategies are significantly decoupled from actual operating characteristics, leading to distorted optimization results. To reduce model complexity and computational burden, traditional control methods often oversimplify building and equipment models. For example, they simplify building thermodynamic processes to static models, ignoring the dynamic thermal resistance and capacity effects of the building envelope as a virtual energy storage device; or, for key energy efficiency equipment such as ground source heat pumps and air source heat pumps, a fixed or univariate energy efficiency ratio (EER) that only varies with outdoor temperature is commonly used. ) is used for calculation. However, in physical reality, heat pumps It is a strongly nonlinear function of its source-side temperature and load-side temperature. When the system actively increases the supply water temperature to utilize low electricity prices for heat storage, the condensing temperature of the heat pump rises accordingly, which will inevitably lead to its... Significant decrease. Optimization models that ignore this key physical characteristic will arrive at overly optimistic estimates of energy savings and cost reductions. When the control strategy is implemented, energy consumption may increase instead of decrease, seriously affecting the robustness and actual effectiveness of the control strategy.

[0006] Furthermore, existing control strategies are poorly adapted to highly dynamic and uncertain market environments. Under real-time electricity pricing mechanisms or day-ahead and real-time market linkage mechanisms, electricity price signals fluctuate dramatically and are often accompanied by high-frequency noise. Many control strategies based on fixed schedules or simple fixed thresholds are prone to causing equipment such as heat pumps to frequently start and stop within a short period of time when faced with such complex electricity price curves. This not only makes it difficult to capture true price arbitrage opportunities but also exacerbates mechanical wear and tear on the equipment, shortens its service life, and increases long-term maintenance costs.

[0007] Therefore, there is an urgent need in this field for a new technical solution that can establish a high-fidelity fully physical quantity coupled model, intelligently adapt to and filter out noise in price signals, and, more importantly, provide a standardized method to clearly quantify the specific marginal cost of flexible call, thereby providing a scientific basis for building participation in the energy market and ultimately achieving synergistic optimization of economic benefits, user comfort and grid friendliness. Summary of the Invention

[0008] To address the problems existing in the prior art, this invention innovatively proposes a dynamic collaborative optimization control method and system for integrated energy systems of building complexes. This effectively solves the problem of low reliability in dynamic collaborative optimization control of integrated energy systems of building complexes caused by existing technologies, and effectively improves the reliability of dynamic collaborative optimization control of integrated energy systems of building complexes.

[0009] The first aspect of this invention provides a dynamic collaborative optimization control method for an integrated energy system of a building complex, comprising: Collect and preprocess multi-source heterogeneous data from building energy systems; Based on mixed-integer linear programming theory and multi-source heterogeneous data of building energy systems, a system operation optimization model is constructed with the objective function of minimizing the total operating cost of the system in the prediction time domain. Based on real-time electricity price data in the prediction time domain and preset flexibility intensity adjustment factors, high and low electricity price thresholds are dynamically generated to trigger demand response actions. Parallel solution of system operation optimization models under baseline and flexible response scenarios; Based on the solution results, the flexibility-specific marginal cost is calculated; the flexibility-specific marginal cost is the ratio of the total operating cost increased or decreased in the flexibility response scenario compared to the baseline operating scenario to the total amount of electrical energy successfully transferred or reduced in that scenario. The system acquires the current clearing price of the power ancillary services market or the user's preset willingness-to-pay threshold for economic benefits in real time, and compares it with the flexibility-specific marginal cost. If the flexibility-specific marginal cost is lower than the clearing price or willingness-to-pay threshold, the system determines the optimal control instruction sequence corresponding to the flexibility response scenario; otherwise, it executes the control instruction sequence corresponding to the baseline operating scenario.

[0010] A second aspect of the present invention provides a dynamic collaborative optimization control system for an integrated energy system of a building complex, comprising: The data acquisition module collects and preprocesses multi-source heterogeneous data from the building energy system; The modeling and prediction module, based on mixed-integer linear programming theory and multi-source heterogeneous data of building energy systems, constructs a system operation optimization model with the objective function of minimizing the total operating cost of the system in the prediction time domain. The dynamic threshold generation module dynamically generates high and low electricity price thresholds to trigger demand response actions based on real-time electricity price data in the prediction time domain and preset flexibility intensity adjustment factors. A two-layer optimization solution module solves the system operation optimization model in parallel under the baseline operation scenario and the flexible response scenario; The evaluation and decision-making module calculates the flexibility-specific marginal cost based on the solution results; the flexibility-specific marginal cost is the ratio of the total operating cost increased or decreased in the flexibility response scenario compared to the baseline operating scenario to the total amount of electrical energy successfully transferred or reduced in that scenario. The execution control module acquires the current clearing price of the power ancillary services market or the user's preset willingness-to-pay threshold for economic benefits in real time, and compares it with the flexibility-specific marginal cost. If the flexibility-specific marginal cost is lower than the clearing price or willingness-to-pay threshold, the optimal control instruction sequence corresponding to the flexibility response scenario is determined; otherwise, the control instruction sequence corresponding to the baseline operating scenario is executed.

[0011] The technical solution adopted in this invention has the following technical effects: 1. The technical solution of this invention is based on mixed-integer linear programming theory and multi-source heterogeneous data of building energy systems. It constructs a system operation optimization model with the objective function of minimizing the total operating cost of the system within the prediction time domain. Based on real-time electricity price data within the prediction time domain and a preset flexibility intensity adjustment factor, it dynamically generates high and low electricity price thresholds to trigger demand response actions. It solves the system operation optimization model in parallel under the baseline operating scenario and the flexibility response scenario. Based on the solution results, it calculates the flexibility-specific marginal cost. It obtains the clearing price of the current electricity ancillary service market or the user's preset economic benefit payment willingness threshold in real time and compares it with the flexibility-specific marginal cost. If the flexibility-specific marginal cost is lower than the clearing price or payment willingness threshold, it determines to execute the optimal control command sequence corresponding to the flexibility response scenario; otherwise, it executes the control command sequence corresponding to the baseline operating scenario. This effectively solves the problem of low reliability of dynamic collaborative optimization control of building complex integrated energy systems caused by existing technologies, and effectively improves the reliability of dynamic collaborative optimization control of building complex integrated energy systems.

[0012] 2. In this invention, the inherent thermal inertia (i.e., heat capacity) of the building envelope is incorporated into the optimization scheduling model as a huge, zero-marginal-investment-cost virtual thermal battery by introducing and accurately solving the 5R1C dynamic thermal network model. Compared to technical solutions that rely solely on explicit energy storage devices such as hot water storage tanks, phase change energy storage, or electrochemical batteries, this invention significantly expands the system's regulation capacity and flexible supply potential, and significantly improves the system's ability to perform peak shaving and valley filling and absorb fluctuating renewable energy without increasing any additional hardware investment.

[0013] 3. This invention fully considers heat pumps in its optimization model. The model utilizes advanced linearization techniques, such as special ordered sets (SOS2), to achieve a high-fidelity approximation of the strong nonlinear characteristics of the system as the temperature dynamically changes with both the load and source sides, within the MILP framework. This means that when the system performs thermal storage (e.g., increasing the supply water temperature), the model can automatically and accurately account for the factors... The additional power consumption cost resulting from the reduction. This high-fidelity modeling avoids the overly optimistic optimization results and the disconnect between theory and practice caused by traditional simplified models, ensuring that the final generated control strategy is physically feasible and economically accurate, thereby greatly improving the robustness of control and the actual energy-saving effect.

[0014] 4. The dynamic threshold generation mechanism based on statistical characteristics (moving average + standard deviation) employed in this invention can intelligently track the overall trend (benchmark drift) of the electricity price curve and adapt to its volatility. By setting a non-responsive zone near the mean, this mechanism can effectively filter out high-frequency, small-amplitude price noise, preventing critical equipment such as heat pump compressors and water pumps from frequently starting and stopping due to negligible electricity price fluctuations. This not only ensures that the system only responds to significant price signals with arbitrage value, but also greatly extends the service life of the equipment and reduces the overall lifecycle maintenance costs.

[0015] 5. The technical solution of this invention proposes a flexibility-specific marginal cost (…). This core metric, flexibility, is presented with a complete calculation paradigm. It transforms the previously vague concept of flexibility into a quantifiable and comparable cost per kilowatt-hour. Users or system operators can then clearly understand the economic cost or benefit of each flexibility deployment. By... By comparing with real-time market prices, this invention constructs a closed-loop, economically based decision-making mechanism that ensures building users only participate in demand response when it is profitable, thereby fundamentally solving the problems of blind response and response losses, and providing a solid scientific pricing basis for buildings to participate in the electricity ancillary services market as a commercial resource.

[0016] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the method in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the physical architecture and signal interaction of the integrated energy system in the method of Embodiment 1 of the present invention; Figure 3 This is a schematic diagram of the equivalent circuit principle of building thermal inertia based on the 5R1C model in the method of Embodiment 1 of the present invention; Figure 4 This is a schematic diagram of the dynamic threshold generation and response signal triggering logic based on real-time electricity price statistical characteristics in the method of Embodiment 1 of the present invention; Figure 5This is a schematic diagram comparing the load curves and indoor temperatures under the baseline operating scenario and the flexibility response scenario in the method of Embodiment 1 of the present invention. Figure 6 This is a schematic diagram of the system in Embodiment 2 of the present invention; Figure 7 This is a schematic diagram of the hardware structure of the computer device used to execute the method in Embodiment 2 of the present invention. Detailed Implementation

[0019] To clearly illustrate the technical features of this solution, the invention will be described in detail below through specific embodiments and in conjunction with the accompanying drawings. The following disclosure provides many different embodiments or examples for implementing different structures of the invention. To simplify the disclosure of the invention, components and arrangements of specific examples are described below. Furthermore, reference numerals and / or letters may be repeated in different examples. This repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed. It should be noted that the components illustrated in the drawings are not necessarily drawn to scale. Descriptions of well-known components, processing techniques, and processes are omitted in this invention to avoid unnecessarily limiting the invention.

[0020] Example 1 This invention aims to solve the aforementioned problems in the prior art. Its main technical objective is to provide a method, system, and related products for flexibility assessment and dynamic collaborative optimization control of integrated energy systems for building complexes, in order to solve one or more of the following technical problems: 1. To address the problem that existing technologies lack precise quantitative assessment of building flexibility call costs, hindering users from making scientific bids and decisions in the electricity market, a method is proposed that can calculate flexibility-specific marginal costs (…). Standardization methods.

[0021] 2. Existing control models oversimplify the physical characteristics of buildings and equipment, particularly neglecting the building's thermal inertia dynamics and heat pumps. To address the problem of significant deviations between optimized results and actual operation caused by nonlinear variations, an optimization control method based on high-fidelity full physical quantity coupling modeling is proposed.

[0022] 3. To address the problem that existing control strategies are poorly adaptable to real-time dynamic electricity prices and are susceptible to price noise interference leading to frequent equipment operation, a dynamic threshold generation mechanism based on statistical characteristics is provided to enhance the robustness of the control strategy and the equipment protection performance.

[0023] To achieve the above objectives, such as Figure 1 As shown, the present invention provides a dynamic collaborative optimization control method for an integrated energy system of a building complex, comprising: Step S1, Multi-source heterogeneous data acquisition and preprocessing steps: Acquire and preprocess multi-source heterogeneous data of the building energy system; Step S2, Integrated Energy System Full Physical Quantity Coupling Modeling Step: Based on mixed integer linear programming theory and multi-source heterogeneous data of building energy systems, construct a system operation optimization model with the objective function of minimizing the total operating cost of the system in the prediction time domain; Step S3, Dynamic Price Threshold Generation Step Based on Statistical Features: Based on real-time electricity price data in the prediction time domain and a preset flexibility intensity adjustment factor, dynamically generate high and low electricity price thresholds to trigger demand response actions. Step S4, Parallel Optimization Solution Step in Two-Layer Scenarios: Parallel solution of the system operation optimization model under the baseline operation scenario and the flexible response scenario; Step S5, Quantification and Evaluation of Flexibility Core Indicators: Based on the solution results, calculate the flexibility-specific marginal cost; the flexibility-specific marginal cost is the ratio of the total operating cost increased or decreased in the flexibility response scenario compared to the baseline operating scenario to the total amount of electrical energy successfully transferred or reduced in that scenario. Step S6, Coordinated Control Strategy Decision and Execution Steps: Real-time acquisition of the current power ancillary service market clearing price or the user's preset economic benefit payment willingness threshold, and comparison with the flexibility-specific marginal cost; if the flexibility-specific marginal cost is lower than the clearing price or payment willingness threshold, then determine the optimal control instruction sequence corresponding to the flexibility response scenario; otherwise, execute the control instruction sequence corresponding to the baseline operation scenario.

[0024] In step S1, the multi-source heterogeneous data of the building energy system includes real-time outdoor meteorological parameters of the area where the building complex is located, future weather forecast data, real-time electricity price data and forecast curves of the electricity market, building internal environmental status parameters, user-defined thermal comfort setting range, and operating status parameters of each underlying device in the integrated energy system.

[0025] Specifically, the integrated energy system for a building complex in this embodiment can be applied to a smart office park, representing the specific implementation process of a flexible assessment and dynamic collaborative optimization control method for an integrated energy system for a building complex. The park is equipped with a ground source heat pump system as the main heating and cooling equipment, a large hot water storage tank as a visible thermal storage unit, and a distributed photovoltaic power generation system installed on the roof. The core control of the system is an optimized control system deployed on an edge computing gateway.

[0026] like Figure 2 As shown, step S1 is the data foundation for the entire optimization control process, which can be achieved through a data acquisition module. A comprehensive data acquisition network is constructed by deploying various sensors and communication interfaces within the park.

[0027] At the physical level, the system connects to the ground source heat pump controller, variable frequency water pump controller, and smart meters via industrial Ethernet (using EtherCAT or Profinet protocols) to collect real-time operating status data of the equipment at a high frequency (once per minute). This data includes, but is not limited to: heat pump compressor frequency, start / stop status, real-time input power, inlet and outlet water temperatures on the source side (buried pipe) and load side (heating / cooling water); the liquid level of the hot water storage tank and the temperature values ​​of multiple temperature measuring points distributed along its height; the total power consumption of the park and the output power of the photovoltaic system.

[0028] Simultaneously, it connects to indoor environmental sensors deployed in various office areas via a wireless sensor network (LoRaWAN or ZigBee) to collect indoor air temperature, relative humidity, CO2 concentration, and the presence of people, using infrared or microwave sensors. Users can set their desired comfortable temperature range via a smart panel on the wall or a mobile app, for example, setting it to 20°C to 24°C in winter.

[0029] At the information level, the system obtains real-time outdoor meteorological parameters, such as dry-bulb temperature, total solar radiation intensity, and wind speed, as well as hourly weather forecast data for the next 24 to 72 hours from local meteorological service providers through Internet API interfaces. On the other hand, it obtains day-ahead market electricity price curves, real-time electricity price prediction sequences, and ancillary service market call signals and clearing price information from the public information platforms of power trading centers or grid operators.

[0030] All collected raw data is sent to the preprocessing unit in the data acquisition module. This unit first performs data cleaning, using the 3σ criterion or the Isolation Forest algorithm to remove obvious outliers caused by sensor malfunctions or communication errors. Then, for missing data due to network latency or packet loss, Lagrange interpolation or time-series-based nearest neighbor imputation is used to complete the data. Finally, all data from different sources and at different time scales are resampled and aligned to form time-series data consistent with the Model Predictive Control (MPC) optimization step size (15 minutes or 1 hour), and stored in a real-time database for use in subsequent steps.

[0031] In step S2, the aim is to construct a mixed-integer linear programming (MILP) model (system operation optimization model) that accurately reflects the dynamic behavior and physical constraints of the system. This can be achieved through... Figure 2 The modeling and prediction module is implemented in the model, and the objective function of the model is to minimize the total operating cost of the system in a future prediction time domain (24 hours).

[0032] objective function Its composition is as follows:

[0033] in, The power purchased from the grid during time period t. This refers to the real-time electricity price at time t. The cost of starting and stopping the equipment during time period t. Let t be the equivalent depreciation cost of the equipment during time period t. The penalty cost for the indoor temperature deviating from the user's most desired value during time period t. To predict the total number of time periods within the time domain.

[0034] The constraints of the system operation optimization model mainly include the following parts: 1. Building thermal inertia model (5R1C model, linear equality constraints): such as Figure 3 As shown, the thermodynamic behavior of a building is abstracted as an equivalent circuit network consisting of five thermal resistances and a single heat capacity. This model can capture the storage and release of heat within the building envelope (walls, roof, etc.). Its discretized state-space equations are represented as a set of linear equality constraints in the MILP model. Specifically, the temperature of the core thermal mass nodes of the building structure is described. The dynamic equation is: ; The equation describing the indoor air node thermal equilibrium is: ; in, For the current time period, For time intervals; For the effective heat capacity of the main building structure, The temperature of the thermal mass node of the building structure at time t; The indoor air temperature at time t; The overall outdoor temperature at time t; The heat transfer coefficient between the thermal mass node and the indoor air; The heat transfer coefficient between the thermal mass node and the outdoor environment; The heat gain projected onto the thermal mass inside the building during time period t; The heat capacity of indoor air; The inner surface temperature during time period t; The heat transfer coefficient between the inner surface and the indoor air; The ventilation heat transfer coefficient; The outdoor dry-bulb temperature at time t; The power supplied to the room by the heating or cooling system during time period t; The internal heat gain generated by indoor personnel and equipment during time period t.

[0035] The equation shows that the rate of temperature change of a building's thermal mass depends on its convective heat transfer with indoor air, its conductive heat transfer with the outdoor environment, and the absorbed solar radiation and internal heat gain. Meanwhile, the indoor air temperature... The heat balance equation is also introduced as a constraint, relating to the heat supplied from the heating / cooling system. The indoor temperature is determined by the following equations: heat loss / gain through windows and ventilation, and heat exchange with the building's interior surfaces. It is no longer an isolated variable, but is related to the thermal state of the building structure. Tight coupling. When allowed When fluctuating within the comfort range, the optimizer can utilize... This heat capacity can be increased when electricity prices are low. and It charges (stores heat) and stops heating when electricity prices are high, allowing the building to slowly discharge heat to maintain room temperature.

[0036] 2. Nonlinear efficiency model and linearization of heat pumps (a set of linear inequalities constraining the transformation of the nonlinear function): Energy efficiency ratio of ground source heat pumps Its performance is key, and it is highly dependent on the temperature of the source side (buried pipe circulating water). and load-side (cold / hot water supply to the building) temperature This invention does not employ a fixed method. Instead of focusing on the value, we first establish a nonlinear function or a two-dimensional lookup table. This function can be obtained from data provided by the manufacturer or through regression analysis of historical data.

[0037] To integrate this nonlinear relationship into the MILP model, this embodiment employs the Special Ordered Set Type 2 (SOS2) method. Specifically, it involves: and The typical operating range is divided into grids to obtain a series of discrete operating points. and its corresponding The values ​​(indices i and j) are used to identify pre-defined, discrete heat pump operating points, where i represents the heat source side temperature (T). source The discrete grid point number, j represents the load-side temperature (T) load The discrete grid points are numbered; a combined index (i,j) collectively identifies a specific, discrete reference operating point on the two-dimensional temperature plane. In the optimization model, a set of continuous weight variables is introduced. and weight variables Apply SOS2 constraints:

[0038]

[0039]

[0040]

[0041] That is, among all weight variables, at most two (or four in the two-dimensional case) adjacent variables can be non-zero. Thus, at any given running point... The values ​​are all determined by the surrounding grid points. Value linear interpolation is used to obtain the result. This high-fidelity modeling method ensures that when the optimizer decides to improve... When performing thermal storage, the model can automatically take into account factors. Decreasing electricity consumption while increasing power consumption allows for more realistic economic decisions.

[0042] 3. Energy Storage and Other Equipment Model (Preferred): For hot water storage tanks, a layered or homogeneous model is used to describe their energy balance, including charging and discharging power, stored energy, energy loss, and constraints such as upper and lower temperature limits. For photovoltaic systems, their power generation... The load term, provided by the prediction module, is a negative term in the energy balance constraint. In addition, a series of operational constraints are included, such as the minimum / maximum power, minimum start-up and shutdown time, and ramp rate of the heat pump and water pump, to ensure the physical feasibility of the control strategy.

[0043] Specifically, step S3 involves generating a dynamic price threshold based on statistical characteristics. like Figure 4 As shown, to enable the control strategy to intelligently adapt to the fluctuations in electricity prices rather than being interfered with by high-frequency noise, a dynamic threshold generator is designed in this step. It operates on a sliding time window that covers both a past period (e.g., 12 hours) and a future forecast period (e.g., 24 hours). For the electricity price series within this window, its moving average is calculated. and standard deviation Then, two electricity price thresholds, high and low, are dynamically generated:

[0044]

[0045] Among them, among them, The high electricity price trigger threshold for time period t. The low electricity price trigger threshold for time period t; It is the moving average of electricity prices within a sliding window centered at time t, covering a period of the past and future. The standard deviation of electricity prices within a sliding window centered at time t, covering the past and a period of time to the future; The preset flexibility intensity adjustment factor; It is a flexibility intensity adjustment factor that can be adjusted by the user according to their risk preference and willingness to respond. This indicates a mild response strategy. This indicates an aggressive strategy. When the predicted real-time electricity price... When the price exceeds the upper red boundary line, the period is marked as a high-price response zone, and the system should tend to reduce load. When the price falls below the lower green boundary line, the period is marked as a low-price energy storage zone, and the system should tend to increase load for energy storage. The area between the two boundary lines is the normal operating zone or dead zone. Within this zone, the system does not perform aggressive flexible scheduling to avoid unnecessary equipment operations.

[0046] The dynamic generation in this invention refers to the fact that the demand response trigger threshold is not a fixed value, but rather it is adjusted in real time and adaptively according to the changing patterns of electricity market prices over time. This process is continuous and automatic, ensuring that the threshold always reflects the relative level of electricity prices in the current and future period, rather than an absolute value.

[0047] This method does not rely on historical, fixed electricity price data. Its core basis is a real-time and predicted electricity price sequence within a "sliding time window". This window is centered on the current time t, covering a portion of historical time (such as the past 12 hours) and the entire prediction time domain (such as the next 24 hours).

[0048] In each rolling optimization step of Model Predictive Control (MPC) (every 15 minutes), the system performs the following steps to recalculate its specific threshold for each time t in the next prediction time domain: (a) Window data update: First, the system updates the electricity price data sequence within the sliding time window. Over time, old historical data is removed from the window, and new real-time data and updated prediction data are moved into the window. (b) Statistical feature calculation: For this updated electricity price data sequence, the system calculates two key statistical indicators: the moving average (…). ): The average of all electricity prices within the calculated window. This value represents the central trend or benchmark level of electricity prices in the current market environment. Standard deviation ( (c) Threshold Dynamic Calculation: Finally, the two dynamically calculated statistical characteristics are combined with a preset adjustment factor. By combining these factors, new high and low electricity price thresholds can be generated.

[0049] This dynamic generation mechanism gives the threshold a strong adaptive capability: (1) adapting to price benchmark drift. During the winter heating season, the overall electricity price level ( (2) Adapting to price volatility. On days when renewable energy output fluctuates drastically, electricity price volatility ( The threshold range of this method (the difference between the high and low thresholds) will increase. The threshold range automatically widens, forming a larger "non-response zone," thereby filtering out severe but potentially unprofitable short-term price noise and making the system response more robust. Conversely, when electricity prices are stable, the threshold range narrows, allowing the system to capture more subtle and effective price signals.

[0050] In general, dynamic generation is a process that iteratively calculates future time domain values ​​within each control period based on a sliding time window, using the moving average and standard deviation of electricity prices. This allows the trigger threshold to track the central trend and fluctuation range of electricity prices in real time, thereby intelligently and adaptively distinguishing between truly worthwhile relative high and low prices, rather than relying on outdated or unsuitable fixed values.

[0051] In step S4, the corresponding step is the parallel optimization solution for the two-layer scene, which can be achieved through... Figure 2 The implementation of the two-layer optimization solution module is the core of achieving quantitative evaluation of flexibility. In each MPC rolling optimization step, it starts every 15 minutes, optimizing two different MILP problems in parallel over the next 24 hours.

[0052] Scenario A: Baseline Operation Scenario. The goal of this scenario is to answer: what is the optimal operating cost of the system, provided that no grid flexibility is provided and only the most basic user comfort needs are met? Therefore, its optimization model is based on the complete model constructed in step S2, with a strict constraint imposed: at all times... Indoor air temperature It must be exactly equal to the user-set center value of the comfort temperature. (21°C). After solving this problem, the total cost of the baseline scenario is obtained. And the power purchase curve of the grid under the benchmark scenario. .

[0053] Scenario B: Flexible Response Scenario. The goal of this scenario is to answer: what is the optimal operating strategy and cost for the system when building flexibility is allowed and electricity price signals are proactively responded to for greater economic benefits? The main difference between its optimization model and the baseline scenario lies in the constraints: firstly, the indoor temperature constraint is relaxed to... , This allows for the utilization of building thermal inertia (i.e., in the 5R1C model). The charging and discharging of heat opens up space. Secondly, a price response constraint linked to the dynamic threshold generated in step S3 is introduced. One implementation is to add a penalty term to the objective function: when... At that time, the power purchased from the power grid Apply a very large penalty coefficient to force the optimizer to find strategies to reduce electricity purchases (such as utilizing energy storage); when In this scenario, a virtual reward (negative cost) can be given to encourage the system to use more electricity during that period. Solving this problem yields the total cost of the flexible response scenario. Power purchase curve of the power grid .

[0054] In the parallel solution of the system operation optimization model under the baseline operating scenario and the flexible response scenario, the constraint of the baseline operating scenario is to strictly maintain the indoor air temperature at a single temperature set by the user; the constraint of the flexible response scenario is to allow the indoor air temperature to fluctuate within the upper and lower limits of the user-set comfort level, and to impose a penalty cost in the objective function or introduce a mandatory power adjustment command in the constraint when the real-time electricity price reaches the high or low electricity price threshold; wherein, the mandatory power adjustment command is specifically as follows: When the predicted real-time electricity price exceeds the high electricity price threshold, at time t, the power purchased from the grid will be forcibly reduced to the maximum power purchase limit allowed during the demand response period. When the predicted real-time electricity price is lower than the low electricity price threshold, during the time period t, the input power of the heat pump is forcibly set to the minimum target power required for the heat pump to operate during the low electricity price period, so as to utilize low-priced electricity for energy storage by forcibly increasing the load.

[0055] Introducing mandatory power adjustment commands into the constraints is an alternative or parallel, more forceful response strategy in this invention, in the flexible response scenario (Scenario B), besides imposing punitive costs in the objective function. Its core idea is that when the electricity price reaches a dynamic threshold, instead of indirectly guiding the optimizer through cost signals, specific equality or inequality constraints are directly added or activated in the mixed-integer linear programming (MILP) model, thereby mathematically forcing the system to make explicit power adjustment actions within a specific time period.

[0056] The specific implementation methods can be divided into the following two cases: 1. Targeting high electricity prices ( Mandatory peak-shaving command: When the predicted real-time electricity price exceeds the high electricity price threshold, the system needs to forcibly reduce the power purchased from the grid during that time period t. This is achieved by introducing the following power cap constraint into the MILP model:

[0057] in: It is the power purchased from the grid at time t, and is a decision variable in the optimization model.

[0058] It is the maximum allowable power purchase capacity during demand response, which is a preset, mandatory parameter value.

[0059] in: zero value ( This means that during this period, the building must operate in a grid-friendly manner, relying entirely on its own photovoltaic power generation, energy storage discharge, or by reducing the load to meet internal needs, and must not purchase electricity from the grid.

[0060] The base load value is the power that is only allowed to meet the uninterrupted base loads (such as emergency lighting, servers, etc.) within the building.

[0061] The contractually agreed value refers to the maximum power value agreed upon in the contract with the grid operator if the building participates in a specific demand response project.

[0062] By adding this hard constraint, the optimization solver is forced to ensure that, while meeting the building's basic requirements, it does so by scheduling energy storage (such as heat release from hot water tanks and building structures) or reducing equipment operating load. The result does not exceed this mandatory upper limit, thus achieving mandatory peak shaving.

[0063] 2. Targeting low electricity price triggers ( Mandatory energy storage command: When the predicted real-time electricity price is lower than the low-price threshold, the system needs to forcibly increase the load during that period t to utilize the low-priced electricity for energy storage. This is achieved by introducing minimum operating power constraints on major adjustable loads (such as heat pumps).

[0064] in: It is the input electrical power of the heat pump at time t, and it is a decision variable in the optimization model.

[0065] This is the minimum target power required for the heat pump to operate during periods of low electricity prices; it is a preset, mandatory parameter value.

[0066] It can be set to: rated power A certain proportion: such as During this period, the forced heat pump operates at a high load rate to maximize the preheating or precooling rate. The power value that can meet the maximum energy storage rate is the power that the heat pump should achieve, calculated from the maximum heat charging rate of the hot water tank or the maximum temperature rise rate of the building structure.

[0067] By adding this hard constraint, the optimization solver is forced to turn on the heat pump during that period and keep its power at a high level, thereby efficiently converting and storing low-cost electricity as thermal energy, achieving instructive valley filling and energy storage.

[0068] In general, mandatory power adjustment commands are implemented by dynamically adding hard constraints (such as...) to the MILP model. or This is achieved through [the following]. This method is more direct and reliable than soft constraints that impose punitive costs, and is especially suitable for business scenarios that require guaranteed precise response volumes (such as participation in capacity markets or contractual demand response).

[0069] like Figure 5 As shown, the optimization results for the two scenarios exhibit significant differences in indoor temperature and grid power curves. In the baseline scenario (black solid line), the temperature is constant, and the power curve mainly follows the fluctuations in heat load. In the flexibility scenario (red solid line), the temperature fluctuates within the comfort range, achieving a peak-shifting and valley-filling effect: a surge in power (preheating and energy storage) in low-electricity-price areas (early morning) and almost zero power (peak shaving) in high-electricity-price areas (daytime peak hours).

[0070] Preferably, to address the issue of grid integration of rooftop photovoltaic power generation in the industrial park, the flexibility response scenario optimization model in step S4 can be further enhanced with additional strategies. Specifically, this includes a strategy to enhance on-site grid integration of photovoltaic power generation. This strategy involves incorporating photovoltaic power generation capacity into the energy balance constraints of step S2. As an endogenous energy source of the system; during the optimization solution of the flexibility response scenario in step S4, when the predicted photovoltaic power generation is greater than the building foundation's electricity load, the electricity price parameter in the objective function for the corresponding time period will be... The dynamic correction is set to a preset virtual negative value or a minimum value far below the actual electricity price, thereby incentivizing the solver in mathematical optimization to force the activation of adjustable loads such as heat pumps. This efficiently converts excess photovoltaic power that might otherwise be abandoned or fed into the grid at low prices into heat energy and stores it in the building envelope or hot water storage tank until the upper limit of equipment operating power, energy storage capacity, or indoor temperature comfort level is reached.

[0071] At the beginning of each optimization step, the predicted photovoltaic power generation is first compared. and the building's foundation non-heating, ventilation and electrical loads (Such as lighting, sockets, elevators, etc.). When it is predicted that during a certain period of time... There is a photovoltaic surplus, that is At that time, without intervention, this surplus electricity The electricity will be fed back to the grid at a lower on-grid price, which is not economically efficient.

[0072] The strategy of this invention is to, when constructing the optimization model for scenario B, consider these periods with photovoltaic surpluses. The electricity price coefficient in its objective function Dynamically modify it to a virtual negative electricity price, for example. ,in It is a small positive number.

[0073] This mathematical treatment creates a strong economic incentive for the optimization solver: during these periods, purchasing electricity not only costs nothing but can actually generate profit (reducing total costs). Therefore, to minimize the objective function, the solver will automatically and maximally increase the electricity load during those periods. In the physical world, this means the optimizer will make the following decisions: Force the ground source heat pump to start and operate at its maximum permissible power. Prioritize supplying the generated heat to the hot water storage tank, heating it to the maximum permissible temperature (e.g., 55°C).

[0074] If the water tank is full, the heat energy is used to preheat the building, raising the indoor temperature to the upper limit of the comfort range (e.g., 24°C), utilizing the building's own heat. For heat storage.

[0075] In this way, the previously wasted solar power is efficiently converted into valuable thermal energy and stored in water tanks and building structures. When the output of photovoltaic power decreases in the evening or at night and the electricity price rebounds, it is released to meet the heating demand, thereby realizing the transfer and value enhancement of energy over time, significantly improving the local consumption rate of photovoltaic power and the overall economic benefits of the system.

[0076] To address the issue of distributed photovoltaic (PV) power generation within building complexes, this invention proactively creates a strong economic incentive signal when PV output is predicted to exceed local load. This signal guides the optimization model to forcibly convert the nearly zero-cost excess electrical energy into heat energy via heat pumps or similar equipment, storing it within the building or in a hot water storage tank. This synergistic solar-thermal energy management approach provides an economical and efficient localized solution to the midday PV power consumption challenge under the duck curve problem.

[0077] In step S5, the core flexibility indicators are quantified and evaluated, which can be achieved through... Figure 2 The evaluation and decision-making module is implemented based on the two sets of paired outputs obtained in step S4. This step calculates two core indicators proposed in this invention.

[0078] 1. Available Electrical Energy Flexibility (AEEF, a preferred calculation index): This index quantifies the total electrical energy that the system can reduce or transfer through flexibility adjustments within a dispatch cycle. Its calculation formula is the integral of the absolute value of the difference between the power curves of the two scenarios: ; in, and These represent the grid power purchases during time period t, respectively, for the flexibility response scenario and the baseline operation scenario.

[0079] 2. Specificity Cost of Flexibility (SC): This is one of the innovations of this invention. It is defined as the additional operating cost incurred by the system to obtain one unit of AEEF. Its calculation formula is as follows:

[0080] The value has clear economic significance: if Negative values ​​(usually) It will be lower than due to buying low and selling high. This indicates that providing flexibility is itself profitable; the negative value represents the net profit gained per kilowatt-hour transferred. A positive value indicates that, in order to provide flexibility, the system incurs additional costs in response to mandatory peak shaving commands from the power grid, when electricity price differences are insufficient to cover preheating heat losses and efficiency reduction costs. The value is the break-even price that provides flexibility per kilowatt-hour.

[0081] In step S6, the corresponding collaborative control strategy decision-making and execution can be achieved through... Figure 2 The execution control module is implemented in the middle.

[0082] In obtaining After that, the final decision-making and execution phase begins. For example... Figure 1 As shown in the diamond-shaped decision box, the execution control module will obtain the clearing price of the power ancillary services market in real time. Or a willingness-to-pay threshold preset by the construction operator according to its own business strategy. .

[0083] The decision-making logic is very clear: judge the inequality. Whether it is valid or not.

[0084] If true, it means the cost of utilizing flexibility is lower than its market value or the user's psychological price point, making it a worthwhile deal. Therefore, the system decides to adopt and execute the optimized result of Scenario B (Flexibility Response Scenario). The control module will convert the optimal control sequence output by Scenario B—that is, the setpoints for heat pump start / stop status, compressor frequency, water pump speed, valve opening, etc., at each time step within the next 24 hours—into specific Modbus or BACnet instructions, which will then be sent to the field PLC or DDC controller for execution. The MPC framework will ensure that this process is rolled over again in the next time step based on the latest system status and forecast information.

[0085] If this condition is not met, it means that the cost of utilizing flexibility is too high, making it not worthwhile. The system then determines the optimization result of execution scenario A (the baseline operating scenario), which is to maintain constant temperature operation in the most energy-efficient way, and refuses to participate in this unprofitable grid interaction.

[0086] Preferably, step S6 also includes a safety mechanism. If communication with the market is interrupted, or if the MILP solver fails to find the optimal solution within a preset time (e.g., 5 minutes), it will automatically switch to a pre-programmed, rule-based local backup control mode (simple temperature difference control or timetable control) to ensure that the basic operation of the HVAC system is not interrupted.

[0087] In summary, this invention provides a dynamic collaborative optimization control method for integrated energy systems in building complexes. Within a rolling optimization model predictive control (MPC) framework, two related yet distinct optimization problems are solved in parallel, with the common objective of minimizing the total system operating cost: one is a baseline operating scenario that strictly constrains indoor temperature to a single setpoint, representing the optimal operating state without providing flexibility; the other is a flexible response scenario that allows indoor temperature fluctuations within a comfort range, designed to proactively respond to electricity price signals triggered by dynamic thresholds. In the collaborative control strategy decision-making and execution steps, the real-time calculated... The value is directly compared with the clearing price obtained from the electricity market or the user's preset willingness to pay. If the use of flexibility is profitable, the control strategy for the flexibility response scenario is executed; otherwise, the strategy for the basic operation scenario is executed, thereby ensuring that every flexibility call is economically reasonable. The final control command is issued to the underlying equipment through the industrial communication interface. In other words, this invention, from high-fidelity modeling and intelligent threshold generation to innovative dual-scenario parallel optimization and marginal cost quantification, and finally to economically based closed-loop decision-making, completely solves the core technical challenges of building cluster flexibility resource assessment and optimization control, providing a practical and cost-effective technical path for buildings to deeply participate in future energy systems.

[0088] This invention's technical solution is based on mixed-integer linear programming theory and multi-source heterogeneous data of building energy systems. It constructs a system operation optimization model with the objective function of minimizing the total system operating cost within the prediction time domain. Based on real-time electricity price data within the prediction time domain and a preset flexibility intensity adjustment factor, it dynamically generates high and low electricity price thresholds to trigger demand response actions. It solves the system operation optimization model in parallel under the baseline operating scenario and the flexibility response scenario. Based on the solution results, it calculates the flexibility-specific marginal cost. It obtains in real-time the current clearing price of the electricity ancillary services market or the user's preset economic benefit payment willingness threshold and compares it with the flexibility-specific marginal cost. If the flexibility-specific marginal cost is lower than the clearing price or payment willingness threshold, it determines to execute the optimal control command sequence corresponding to the flexibility response scenario; otherwise, it executes the control command sequence corresponding to the baseline operating scenario. This effectively solves the problem of low reliability in dynamic collaborative optimization control of integrated energy systems in building complexes caused by existing technologies, and effectively improves the reliability of dynamic collaborative optimization control of integrated energy systems in building complexes.

[0089] This invention introduces and precisely solves the 5R1C dynamic thermal network model, incorporating the inherent thermal inertia (i.e., heat capacity) of the building envelope as a massive, zero-marginal-cost virtual thermal battery into the optimization scheduling model. Compared to solutions relying solely on explicit energy storage devices such as hot water tanks, phase-change energy storage, or electrochemical batteries, this invention significantly expands the system's regulation capacity and flexible supply potential, substantially enhancing the system's ability to peak and valley loads and absorb fluctuating renewable energy sources without any additional hardware investment.

[0090] This invention fully considers heat pumps in its optimization model. The model utilizes advanced linearization techniques, such as special ordered sets (SOS2), to achieve a high-fidelity approximation of the strong nonlinear characteristics of the system as the temperature dynamically changes with both the load and source sides, within the MILP framework. This means that when the system performs thermal storage (e.g., increasing the supply water temperature), the model can automatically and accurately account for the factors... The additional power consumption cost resulting from the reduction. This high-fidelity modeling avoids the overly optimistic optimization results and the disconnect between theory and practice caused by traditional simplified models, ensuring that the final generated control strategy is physically feasible and economically accurate, thereby greatly improving the robustness of control and the actual energy-saving effect.

[0091] The dynamic threshold generation mechanism based on statistical characteristics (moving average + standard deviation) employed in this invention can intelligently track the overall trend (benchmark drift) of the electricity price curve and adapt to its volatility. By setting a non-responsive zone near the mean, this mechanism can effectively filter out high-frequency, small-amplitude price noise, preventing critical equipment such as heat pump compressors and water pumps from frequently starting and stopping due to negligible electricity price fluctuations. This not only ensures that the system only responds to significant price signals with arbitrage value, but also greatly extends the service life of the equipment and reduces the overall lifecycle maintenance costs.

[0092] The technical solution of this invention proposes a flexibility-specific marginal cost ( This core metric, flexibility, is presented with a complete calculation paradigm. It transforms the previously vague concept of flexibility into a quantifiable and comparable cost per kilowatt-hour. Users or system operators can then clearly understand the economic cost or benefit of each flexibility deployment. By... By comparing with real-time market prices, this invention constructs a closed-loop, economically based decision-making mechanism that ensures building users only participate in demand response when it is profitable, thereby fundamentally solving the problems of blind response and response losses, and providing a solid scientific pricing basis for buildings to participate in the electricity ancillary services market as a commercial resource.

[0093] Example 2 like Figure 6 As shown, the system implementing the above method is divided into multiple functional modules at the software level. These modules can run on the same edge computing device or be deployed in a distributed manner in the cloud and at the edge. Specifically, the technical solution of this invention also provides a dynamic collaborative optimization control system for a building complex integrated energy system, including: Data acquisition module ( Figure 6 -Module 1): Responsible for the interface with the physical world, it collects data from sensors and devices through various drivers and communication protocol stacks to collect and preprocess multi-source heterogeneous data of the building energy system; This module is configured to: acquire outdoor meteorological parameters, real-time electricity price data, building internal environmental status parameters and operating status parameters of various devices in the integrated energy system in real time and periodically through standardized communication protocol interfaces. Modeling and Prediction Module ( Figure 6Module 2: Based on mixed-integer linear programming theory and multi-source heterogeneous data of building energy systems, a system operation optimization model is constructed with the objective function of minimizing the total operating cost of the system within the predicted time domain. It embeds a machine learning model (LSTM or Transformer) for predicting photovoltaic output and building load, and stores a 5R1C model RC parameter library and a heat pump COP model library. These parameters can be updated periodically through an online identification algorithm. This module is configured to: based on the fixed-stored 5R1C-based building thermal network model parameter library and heat pump nonlinear efficiency model parameter library, and run machine learning or time series prediction algorithms to obtain a predicted sequence of photovoltaic power generation and building foundation heat load within a future time domain; Dynamic threshold generation module ( Figure 6 -Module 3): Based on real-time electricity price data in the predicted time domain and a preset flexibility intensity adjustment factor, dynamically generate high and low electricity price thresholds to trigger demand response actions; this module is configured to: run a sliding window statistical algorithm based on real-time and predicted electricity price data, and combine it with the flexibility intensity adjustment factor to generate dynamic high and low electricity price trigger thresholds for demand response in real time. Two-level optimization solution module ( Figure 6 -Module 4): Parallel solution of system operation optimization models in baseline and flexible response scenarios; This module is the computational core of the system and is configured to: embed or call a commercial or open-source mixed integer linear programming solver. It is responsible for dynamically constructing MILP problems for scenario A and scenario B based on real-time data, and calling commercial solvers (Gurobi, CPLEX) or open-source solvers (SCIP, CBC) to solve them. It constructs and solves optimization problems in the baseline and flexible response scenarios in parallel, and outputs two sets of corresponding optimal scheduling strategies and predicted costs. Assessment and Decision Module ( Figure 6 -Module 5): Implements the indicator calculation in step S5 and the economic judgment logic in step S6, that is, calculates the flexibility-specific marginal cost based on the solution results; the flexibility-specific marginal cost is the ratio of the total operating cost increased or decreased by the flexibility response scenario compared to the baseline operating scenario to the total amount of electrical energy successfully transferred or reduced in the scenario; this module is configured to calculate the available electrical energy flexibility (AEEF) and flexibility-specific marginal cost (SC) by calling the preset formula based on the output of the two-layer optimization solution module, and finally decide which operating strategy to adopt based on the comparison results with market price signals or user willingness to pay; Execution control module ( Figure 6Module 6: Serving as a bridge between decision-making and physical execution, this module translates abstract control sequences into specific instructions recognizable by the underlying devices. Specifically, it acquires the current clearing price of the power ancillary services market or the user's preset willingness-to-pay threshold for economic benefits in real time, and compares it with the flexibility-specific marginal cost. If the flexibility-specific marginal cost is lower than the clearing price or willingness-to-pay threshold, the optimal control instruction sequence corresponding to the flexibility response scenario is determined; otherwise, the control instruction sequence corresponding to the baseline operating scenario is executed. This module is configured to receive the final strategy instructions from the evaluation and decision-making module, parse and convert them into low-level control instructions conforming to industrial communication protocols such as Modbus, BACnet, or OPC UA, and distribute them to the programmable logic controller (PLC) or building automation system (BAS) in the field to achieve precise regulation of physical equipment such as heat pumps, water pumps, and valves.

[0094] like Figure 7 As shown, the computer device that carries these software modules is typically an industrial-grade edge controller or server. It contains a high-performance processor (CPU / GPU) to handle complex MILP solutions, a large enough memory (RAM) to temporarily store real-time data and optimization variables, non-volatile memory (SSD / HDD) to permanently store program code and historical data, and rich communication interfaces (Ethernet, RS485, etc.) to connect to internal and external networks.

[0095] This invention's technical solution is based on mixed-integer linear programming theory and multi-source heterogeneous data of building energy systems. It constructs a system operation optimization model with the objective function of minimizing the total system operating cost within the prediction time domain. Based on real-time electricity price data within the prediction time domain and a preset flexibility intensity adjustment factor, it dynamically generates high and low electricity price thresholds to trigger demand response actions. It solves the system operation optimization model in parallel under the baseline operating scenario and the flexibility response scenario. Based on the solution results, it calculates the flexibility-specific marginal cost. It obtains in real-time the current clearing price of the electricity ancillary services market or the user's preset economic benefit payment willingness threshold and compares it with the flexibility-specific marginal cost. If the flexibility-specific marginal cost is lower than the clearing price or payment willingness threshold, it determines to execute the optimal control command sequence corresponding to the flexibility response scenario; otherwise, it executes the control command sequence corresponding to the baseline operating scenario. This effectively solves the problem of low reliability in dynamic collaborative optimization control of integrated energy systems in building complexes caused by existing technologies, and effectively improves the reliability of dynamic collaborative optimization control of integrated energy systems in building complexes.

[0096] This invention introduces and precisely solves the 5R1C dynamic thermal network model, incorporating the inherent thermal inertia (i.e., heat capacity) of the building envelope as a massive, zero-marginal-cost virtual thermal battery into the optimization scheduling model. Compared to solutions relying solely on explicit energy storage devices such as hot water tanks, phase-change energy storage, or electrochemical batteries, this invention significantly expands the system's regulation capacity and flexible supply potential, substantially enhancing the system's ability to peak and valley loads and absorb fluctuating renewable energy sources without any additional hardware investment.

[0097] This invention fully considers heat pumps in its optimization model. The model utilizes advanced linearization techniques, such as special ordered sets (SOS2), to achieve a high-fidelity approximation of the strong nonlinear characteristics of the system as the temperature dynamically changes with both the load and source sides, within the MILP framework. This means that when the system performs thermal storage (e.g., increasing the supply water temperature), the model can automatically and accurately account for the factors... The additional power consumption cost resulting from the reduction. This high-fidelity modeling avoids the overly optimistic optimization results and the disconnect between theory and practice caused by traditional simplified models, ensuring that the final generated control strategy is physically feasible and economically accurate, thereby greatly improving the robustness of control and the actual energy-saving effect.

[0098] The dynamic threshold generation mechanism based on statistical characteristics (moving average + standard deviation) employed in this invention can intelligently track the overall trend (benchmark drift) of the electricity price curve and adapt to its volatility. By setting a non-responsive zone near the mean, this mechanism can effectively filter out high-frequency, small-amplitude price noise, preventing critical equipment such as heat pump compressors and water pumps from frequently starting and stopping due to negligible electricity price fluctuations. This not only ensures that the system only responds to significant price signals with arbitrage value, but also greatly extends the service life of the equipment and reduces the overall lifecycle maintenance costs.

[0099] The technical solution of this invention proposes a flexibility-specific marginal cost ( This core metric, flexibility, is presented with a complete calculation paradigm. It transforms the previously vague concept of flexibility into a quantifiable and comparable cost per kilowatt-hour. Users or system operators can then clearly understand the economic cost or benefit of each flexibility deployment. By... By comparing with real-time market prices, this invention constructs a closed-loop, economically based decision-making mechanism that ensures building users only participate in demand response when it is profitable, thereby fundamentally solving the problems of blind response and response losses, and providing a solid scientific pricing basis for buildings to participate in the electricity ancillary services market as a commercial resource.

[0100] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A dynamic collaborative optimization control method for an integrated energy system of a building complex, characterized in that, include: Collect and preprocess multi-source heterogeneous data from building energy systems; Based on mixed-integer linear programming theory and multi-source heterogeneous data of building energy systems, a system operation optimization model is constructed with the objective function of minimizing the total operating cost of the system in the prediction time domain. Based on real-time electricity price data in the prediction time domain and preset flexibility intensity adjustment factors, high and low electricity price thresholds are dynamically generated to trigger demand response actions. Parallel solution of system operation optimization models under baseline and flexible response scenarios; Based on the solution results, the flexibility-specific marginal cost is calculated; the flexibility-specific marginal cost is the ratio of the total operating cost increased or decreased in the flexibility response scenario compared to the baseline operating scenario to the total amount of electrical energy successfully transferred or reduced in that scenario. The system acquires the current clearing price of the power ancillary services market or the user's preset willingness-to-pay threshold for economic benefits in real time, and compares it with the flexibility-specific marginal cost. If the flexibility-specific marginal cost is lower than the clearing price or willingness-to-pay threshold, the system determines the optimal control instruction sequence corresponding to the flexibility response scenario; otherwise, it executes the control instruction sequence corresponding to the baseline operating scenario.

2. The dynamic collaborative optimization control method for a building complex integrated energy system according to claim 1, characterized in that, Multi-source heterogeneous data for building energy systems include real-time outdoor meteorological parameters of the building complex area, future weather forecast data, real-time electricity price data and forecast curves of the electricity market, building interior environmental status parameters, user-defined thermal comfort setting ranges, and operating status parameters of various underlying equipment within the integrated energy system.

3. The dynamic collaborative optimization control method for a building complex integrated energy system according to claim 1, characterized in that, The objective function in the system operation optimization model for: in, The power purchased from the grid during time period t. This refers to the real-time electricity price at time t. The cost of starting and stopping the equipment during time period t. Let t be the equivalent depreciation cost of the equipment during time period t. The penalty cost for the indoor temperature deviating from the user's most desired value during time period t. To predict the total number of time periods within the time domain.

4. The dynamic collaborative optimization control method for a building complex integrated energy system according to claim 1, characterized in that, The system operation optimization model's constraints include a 5R1C dynamic thermal network model with linear equality constraints. This 5R1C dynamic thermal network model is used to construct the building envelope as including at least five thermal resistances and one thermal capacity to characterize the building envelope's thermal inertia. The discretized state-space equations of the 5R1C dynamic thermal network model include at least: equations describing the temperature dynamics of the building structure's thermal mass nodes and equations describing the thermal balance of the indoor air nodes. Specifically, the equations describing the temperature dynamics of the building structure's thermal mass nodes are: ； The equation describing the indoor air node thermal equilibrium is: ； in, For a period of time, For time intervals; For the effective heat capacity of the main building structure, The temperature of the thermal mass node of the building structure at time t; The indoor air temperature at time t; The outdoor temperature at time t is the overall temperature. The heat transfer coefficient between the thermal mass node and the indoor air; The heat transfer coefficient between the thermal mass node and the outdoor environment; The heat gain projected onto the thermal mass inside the building during time period t; The heat capacity of indoor air; The inner surface temperature during time period t; The heat transfer coefficient between the inner surface and the indoor air; The ventilation heat transfer coefficient; The outdoor dry-bulb temperature at time t; The power supplied to the room by the heating or cooling system during time period t; The internal heat gain generated by indoor personnel and equipment during time period t.

5. The dynamic collaborative optimization control method for a building complex integrated energy system according to claim 4, characterized in that, The constraints of the system operation optimization model also include a set of linear inequality constraints after transforming the nonlinear function. The nonlinear function is used to construct the energy efficiency ratio of the heat pump device as a nonlinear function that varies with the source-side temperature and the load-side temperature. Specifically, the set of linear inequality constraints after transforming the nonlinear function is as follows: The joint operating domain of source-side temperature and load-side temperature is discretized into a two-dimensional grid point set; the two-dimensional grid point is the operating point and its corresponding energy efficiency ratio value; Introduce a set of continuous weight variables and impose an SOS2 constraint on the introduced set of weight variables, such that at most two adjacent variables can be non-zero among all weight variables.

6. The dynamic collaborative optimization control method for a building complex integrated energy system according to claim 1, characterized in that, Based on real-time electricity price data within the prediction time domain and a preset flexibility intensity adjustment factor, high and low electricity price thresholds for triggering demand response actions are dynamically generated as follows: ； ； in, The high electricity price trigger threshold for time period t. The low electricity price trigger threshold for time period t; It is the moving average of electricity prices within a sliding window centered at time t, covering a period of the past and future. The standard deviation of electricity prices within a sliding window centered at time t, covering the past and a period of time to the future; A preset flexibility intensity adjustment factor is used; when the electricity price is predicted. At that time, a signal is generated to reduce load or utilize energy storage for energy release; when At that time, a signal is generated to increase the load for energy storage or preheating.

7. The dynamic collaborative optimization control method for a building complex integrated energy system according to claim 1, characterized in that, In the parallel solution of the system operation optimization model under the baseline operating scenario and the flexible response scenario, the constraint of the baseline operating scenario is to strictly maintain the indoor air temperature at a single temperature set by the user; the constraint of the flexible response scenario is to allow the indoor air temperature to fluctuate within the upper and lower limits of the user-set comfort level, and to impose a penalty cost in the objective function or introduce a mandatory power adjustment command in the constraint when the real-time electricity price reaches the high or low electricity price threshold; wherein, the mandatory power adjustment command is specifically as follows: When the predicted real-time electricity price exceeds the high electricity price threshold, at time t, the power purchased from the grid will be forcibly reduced to the maximum power purchase limit allowed during the demand response period. When the predicted real-time electricity price is lower than the low electricity price threshold, during the time period t, the input power of the heat pump is forcibly set to the minimum target power required for the heat pump to operate during the low electricity price period, so as to utilize low-priced electricity for energy storage by forcibly increasing the load.

8. The dynamic collaborative optimization control method for a building complex integrated energy system according to claim 1, characterized in that, Flexibility-specific marginal cost The calculation formula is: in, and These represent the instantaneous operating costs of the flexible response scenario and the baseline operation scenario at time t, respectively. and The grid power purchases during time period t represent the flexible response scenario and the baseline operation scenario, respectively. To predict the total number of time periods within the time domain; The time step is t; the instantaneous operating cost includes at least the cost of electricity purchase, equipment start-up and shutdown costs, equipment operating depreciation costs, and virtual penalty costs incurred due to deviation from the comfort setting center value.

9. The dynamic collaborative optimization control method for a building complex integrated energy system according to claim 8, characterized in that, Based on the solution results, after calculating the flexibility-specific marginal cost, the following steps are also included: Based on the solution results, the available power flexibility is calculated; where available power flexibility is the total amount of power that can be reduced or transferred through flexibility adjustment within the scheduling cycle; the calculation method is the integral of the absolute value of the difference between the power curves under the baseline operating scenario and the flexibility response scenario.

10. A dynamic collaborative optimization control system for an integrated energy system of a building complex, characterized in that, include: The data acquisition module collects and preprocesses multi-source heterogeneous data from the building energy system; The modeling and prediction module, based on mixed-integer linear programming theory and multi-source heterogeneous data of building energy systems, constructs a system operation optimization model with the objective function of minimizing the total operating cost of the system in the prediction time domain. The dynamic threshold generation module dynamically generates high and low electricity price thresholds to trigger demand response actions based on real-time electricity price data in the prediction time domain and preset flexibility intensity adjustment factors. A two-layer optimization solution module solves the system operation optimization model in parallel under the baseline operation scenario and the flexible response scenario; The evaluation and decision-making module calculates the flexibility-specific marginal cost based on the solution results; the flexibility-specific marginal cost is the ratio of the total operating cost increased or decreased in the flexibility response scenario compared to the baseline operating scenario to the total amount of electrical energy successfully transferred or reduced in that scenario. The execution control module acquires the current clearing price of the power ancillary services market or the user's preset willingness-to-pay threshold for economic benefits in real time, and compares it with the flexibility-specific marginal cost. If the flexibility-specific marginal cost is lower than the clearing price or willingness-to-pay threshold, the optimal control instruction sequence corresponding to the flexibility response scenario is determined; otherwise, the control instruction sequence corresponding to the baseline operating scenario is executed.

Images