A method and system for determining a regulation boundary for a building participating in grid dispatch

Abstract

The application discloses a method and system for determining the adjustment boundary of a building participating in power grid dispatching, the method comprising obtaining physical parameters, operation prediction data and operation error data of a target intelligent building, obtaining equipment operation constraints, power balance constraints, equipment constraints and energy constraints of the target intelligent building in a power grid dispatching period, and then obtaining a linear energy consumption model corresponding to the target intelligent building; obtaining adjustment power boundary data and amplitude constraints of each flexible device in the target intelligent building in the power grid dispatching period according to the physical parameters, so as to obtain an initial boundary envelope of the target intelligent building; performing iterative checking and cutting on the initial boundary envelope according to the linear energy consumption model, so as to obtain a target boundary envelope corresponding to the target intelligent building; and obtaining power boundary data of the target intelligent building in the power grid dispatching period according to the target boundary envelope and a preset target power function of the target intelligent building, so as to improve the accuracy of the power boundary data.

Description

Technical Field

[0001] This invention relates to the field of power grid dispatching technology, and specifically to a method and system for determining the adjustment boundary of a building's participation in power grid dispatching. Background Technology

[0002] With the continuous improvement of terminal electrification levels, intelligent buildings, integrating a large number of controllable and flexible devices such as HVAC, energy storage systems, and electric vehicles, have become key flexible resources for grid dispatch. To provide regulation capabilities to the grid or virtual power plants, existing technologies typically predict the building's baseline total power without demand response intervention by collecting historical electricity consumption data and the operating status of individual loads. This data, combined with distribution transformer capacity, equipment safety limits, and empirical coefficients, is used to calculate the maximum upward or downward adjustment boundary of the total power meter for each future dispatch period. Finally, a series of independent hourly power upper and lower bounds are reported to the grid as rigid constraints. Other methods abstract the building's elastic energy-consuming resources into a unified virtual battery model with an equivalent state of charge. Through data-driven methods, equivalent charging / discharging and state recovery parameters are identified, and the power increase / decrease trajectory that the building can withstand without exceeding state limits is deduced. A quantified set of flexible feasible options with uncertainties is then reported to the dispatch center.

[0003] However, the aforementioned hourly power upper and lower bound methods focus on characterizing the reachability of isolated single points in each time period, failing to fully reflect the cross-time-series coupling characteristics of continuous charging and discharging of energy storage devices inside buildings, which are constrained by the state of charge boundary, and the cumulative effects of building thermal inertia and temperature comfort zone on HVAC operation. This makes it difficult to realize the multi-time-period scheduling trajectory of the power grid plan in actual execution due to physical hard constraints such as triggering energy depletion or exceeding room temperature limits. At the same time, this method often uses fixed empirical margins to handle multi-source uncertainties, making it difficult to balance scheduling reliability and full utilization of resources. The virtual battery method uses simplified macroscopic scalars and linear empirical parameters to fit the complex heterogeneous physical processes at the underlying level, resulting in mapping bias and a high dependence on the fitting quality of historical data, making it difficult to accurately capture the nonlinear physical hard constraints under extreme conditions.

[0004] To address the aforementioned issues, existing technologies have proposed an external approximation projection method based on dual transformation and the Big M method. This method first establishes a continuous linear operational constraint model based on the physical parameters of distributed energy resources and power distribution networks. Second, it transforms the boundary extraction of the time-domain coupled feasible region into a mathematical optimization problem of determining whether a feasible solution exists within it. Then, it introduces auxiliary binary variables using dual transformation and the Big M method, transforming the original continuous projection problem into a mixed-integer linear programming problem. Subsequently, a parallel umbrella constraint algorithm is used to filter and eliminate redundant conditions to reduce computational scale. Finally, an external approximation algorithm iteratively solves the mixed-integer programming problem in space, generating a tangent plane by searching for infeasible solutions in the worst-case scenario, gradually approximating and characterizing the final low-dimensional external power boundary. However, this scheme requires the introduction of a large number of binary variables and large M parameters when solving the feasible region projection for multiple time periods. The setting of large M parameters has an inherent dilemma that leads to instability in numerical solutions, which may result in insufficient accuracy of the final feasible region. In the integer programming framework, if the uncertainty of the environment and load needs to be integrated, it is usually necessary to rely on the multi-scenario generation method, which further exacerbates the expansion of the computational dimension and makes it difficult to achieve a balance between computational efficiency and boundary accuracy. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention discloses a method and system for determining the adjustment boundary of a building's participation in power grid dispatch, thereby improving the accuracy of the adjustment boundary.

[0006] To achieve the above objectives, the present invention discloses a method for determining the regulation boundary of a building's participation in power grid dispatch, comprising: Obtain the physical parameters of the target intelligent building and the predicted operation data and operation error data of the target intelligent building during the power grid dispatch cycle; Based on the physical parameters, the operational prediction data, and the operational error data, the equipment operation constraints, power balance constraints, equipment constraints, and energy constraints of the target intelligent building within the power grid dispatch cycle are obtained. Based on the equipment operation constraints, the power balance constraints, the equipment constraints, and the energy constraints, obtain the linear energy consumption model corresponding to the target intelligent building; Based on the physical parameters, obtain the adjustment power boundary data and amplitude constraints of each flexible device in the target intelligent building during the power grid dispatch cycle; The initial boundary envelope of the target intelligent building is obtained based on the adjusted power boundary data and the amplitude constraint; The initial boundary envelope is iteratively verified and cut according to the linear energy consumption model to obtain the target boundary envelope corresponding to the target intelligent building. Based on the target boundary envelope and the preset target power function of the target intelligent building, the power boundary data of the target intelligent building within the power grid dispatch cycle is obtained.

[0007] This invention discloses a method for determining the regulation boundaries of a building's participation in power grid dispatch. The method involves acquiring physical parameters, operational prediction data, and operational error data, and then performing constraint construction processing on these data to obtain equipment operation constraints, power balance constraints, equipment constraints, and energy constraints characterizing the physical operational limitations of each device within the building. This constraint data is then integrated into a model to obtain linear energy consumption model data characterizing the complete physical operational rules within the building. Simultaneously, boundary extraction processing is performed on the physical parameters to obtain regulation power boundary data and amplitude constraint data characterizing the independent physical limits of each flexible device. Finally, envelope integration processing is performed on these boundary data to obtain… The system obtains initial boundary envelope data representing the initial adjustable capacity range of a building; then iteratively verifies and segments the initial boundary envelope data and the linear energy consumption model data to obtain target boundary envelope data after verification and elimination of internal physical violation trajectories. The target boundary envelope data is then subjected to power projection processing to obtain power boundary data representing the adjustable power range of the building in each time period within the scheduling cycle. This ensures that the final output power boundary data internalizes the uncertainties represented by the operational error data and eliminates unexecutable false adjustment spaces through iterative verification, achieving an accurate depiction of the building's true adjustable capacity and improving the accuracy of the power boundary data.

[0008] As a preferred example, the acquisition of the physical parameters of the target intelligent building and the operational prediction data and operational error data of the target intelligent building during the power grid dispatch cycle includes: The building parameters, historical operating data, and equipment parameters of each flexible device in the target intelligent building are obtained to obtain the physical parameters of the target intelligent building based on the building parameters and the equipment parameters; wherein, the historical operating data includes historical electricity consumption time series data, historical uncontrollable load time series data, a first error sequence between historical outdoor temperature prediction values ​​and historical outdoor temperature measurement values, and a second error sequence between historical rigid load prediction values ​​and historical rigid load measurement values; Based on the historical electricity consumption time series data, the baseline power prediction sequence of the target smart building within the power grid dispatch cycle is obtained through a preset electricity consumption prediction model. Based on the historical uncontrollable load time series data, a rigid load prediction sequence for the target intelligent building within the power grid dispatch cycle is obtained through a preset load prediction model. The maximum absolute value of each error in the first error sequence is taken as the maximum prediction error bound of outdoor temperature, and the maximum absolute value of each error in the second error sequence is taken as the maximum prediction error bound of rigid load. The maximum prediction error bound of outdoor temperature and the maximum prediction error bound of rigid load are used as the operating error data. Obtain the outdoor temperature prediction sequence of the target smart building during the power grid dispatch cycle, and use the baseline power prediction sequence, the rigid load prediction sequence, and the outdoor temperature prediction sequence as the operation prediction data of the target smart building during the power grid dispatch cycle.

[0009] The above scheme obtains building parameters, historical operating data, and equipment parameters, and performs predictive processing on the historical electricity consumption time series data to obtain baseline power prediction sequence data characterizing the natural electricity consumption trend of the building without scheduling intervention; it also performs predictive processing on the historical uncontrollable load time series data to obtain rigid load prediction sequence data characterizing the electricity consumption trend of uncontrollable loads within the building; it performs statistical analysis on the historical prediction error sequence data to obtain prediction error boundary data characterizing the maximum deviation range of outdoor temperature and rigid load; and it classifies and integrates the baseline power prediction sequence data, rigid load prediction sequence data, outdoor temperature prediction sequence data, and prediction error boundary data to obtain operating prediction data and operating error data characterizing the building's operating prediction and error distribution characteristics. This ensures that the acquisition of the above data relies entirely on the building's own historical operating records, avoiding subjective biases introduced by manual experience coefficients, and thus providing a data foundation that objectively reflects the actual operating laws of the building for subsequent boundary construction.

[0010] As a preferred example, the building parameters include the building's equivalent heat capacity and equivalent thermal resistance; the equipment parameters include the heating efficiency ratio, cooling efficiency ratio, maximum heating power and maximum cooling power of each HVAC system, the maximum charging and discharging power, charging and discharging efficiency and ultimate energy capacity of each electric energy storage system, the ultimate charging and discharging power of each electric vehicle cluster, and the maximum operating power and ultimate total energy demand of each transferable load. The step of obtaining the equipment operation constraints, power balance constraints, equipment constraints, and energy constraints of the target intelligent building within the power grid dispatch cycle based on the physical parameters, the operation prediction data, and the operation error data includes: Based on the building's equivalent heat capacity, the building's equivalent thermal resistance, the heating energy efficiency ratio, the cooling energy efficiency ratio, the outdoor temperature prediction sequence, and the maximum prediction error boundary of the outdoor temperature, the equipment operation constraints of the target smart building within the power grid dispatch cycle are obtained. The predicted power generation sequence of each distributed power source in the target smart building during the grid dispatch cycle is obtained, and the power balance constraint of the target smart building during the grid dispatch cycle is obtained based on the predicted power generation sequence, the baseline power prediction sequence and the rigid load prediction sequence. Based on the maximum heating power, the maximum cooling power, the maximum charging and discharging power, and the ultimate charging and discharging power, the equipment constraints of the target intelligent building within the power grid dispatch cycle are obtained; Based on the charging and discharging efficiency, the maximum energy capacity, the maximum operating power, and the maximum total energy demand, the energy constraints of the target smart building within the power grid dispatch cycle are obtained.

[0011] The above scheme divides physical parameters into data representing the building's equivalent heat capacity and equivalent thermal resistance, and data representing the rated performance of each device, such as heating efficiency ratio, cooling efficiency ratio, power limit, charge / discharge efficiency, energy capacity, and ultimate total energy demand. It then combines these building parameter data with outdoor temperature prediction sequence data and outdoor temperature maximum prediction error boundary data to construct thermodynamic constraints, obtaining device operation constraint data representing the operating temperature limits of HVAC systems. Power limit data within the device parameters is processed to obtain device constraint data representing the upper power limit of each device. Efficiency and capacity data within the device parameters are processed using energy recursion to obtain energy constraint data representing the energy state transition limits of energy storage and transferable loads. Finally, power balance processing is performed on distributed power generation prediction sequence data, baseline power prediction sequence data, and rigid load prediction sequence data to obtain power balance constraint data representing the power balance relationship at the building's grid connection point. This ensures a precise correspondence between the construction of each constraint and specific parameters, avoiding constraint deviations caused by parameter mixing, and thus improving the accuracy of the linear energy model in depicting the operating characteristics of various devices within the building.

[0012] As a preferred example, obtaining the linear energy consumption model corresponding to the target intelligent building based on the equipment operation constraints, the power balance constraints, the equipment constraints, and the energy constraints includes: The equipment operation constraints, the power balance constraints, the equipment constraints, and the energy constraints are divided into a set of linear equality constraints and a set of linear inequality constraints. For any constraint in the set of linear equality constraints, the constraint is split into a pair of linear inequality constraints with opposite directions. Based on all the linear inequality constraints corresponding to the linear equality constraint set and the linear inequality constraint set, obtain the inequality constraint set of the target intelligent building within the power grid dispatch cycle; Obtain the decision variables, constant terms, and variable coefficients of each inequality constraint in the inequality constraint set. Concatenate all the decision variables corresponding to the inequality constraint set into a decision variable column vector over the power grid dispatching cycle; The matrix dimension is determined by the number of inequality constraints in the inequality constraint set and the dimension of the column vector of the decision variables. The matrix elements are obtained based on the set of inequality constraints, the column vector of decision variables, and the coefficients of the variables; Obtain the constant term in each inequality constraint in the inequality constraint set, and obtain the linear energy consumption model of the target smart building in the power grid dispatch cycle based on the constant term, the matrix dimension, the matrix elements and the decision variable column vector.

[0013] The above scheme classifies and processes equipment operation constraints, power balance constraints, equipment constraints, and energy constraints to obtain linear equality constraint sets and linear inequality constraint sets. It then decomposes and transforms the constraints in the linear equality constraint sets to obtain inequality constraint data that is consistent with the original linear inequality constraint forms. Finally, it performs full-time splicing processing on the decision variables in all inequality constraints to obtain decision variable column vector data. Statistical processing is performed on the number of inequality constraints and the dimension of the decision variable column vectors to determine the matrix dimension data. Finally, the variable coefficients and constant terms in each inequality constraint are extracted one by one to obtain matrix element data and constant term data, thus forming linear energy consumption model data expressed by coefficient matrices and constant vectors. This allows the originally diverse and scattered constraints to be integrated into a standardized matrix inequality system without information loss, avoiding approximations or simplifications in the constraint transformation process, thereby improving the accuracy of the mathematical expression of the linear energy consumption model.

[0014] As a preferred example, the device parameters include the limit physical ramp rate for each flexible device; The step of obtaining the adjustment power boundary data and amplitude constraints of each flexible device in the target intelligent building within the power grid dispatch cycle based on the physical parameters includes: For any time period within the power grid dispatch cycle, the maximum heating power, the maximum charging power among the maximum charging and discharging power, and the maximum operating power are summed to obtain the upper bound of the regulation power corresponding to that time period; The lower bound of the adjustment power corresponding to this time period is obtained by summing the maximum cooling power, the maximum discharge power in the maximum charge and discharge power and the ultimate discharge power in the ultimate charge and discharge power and taking the amplitude. Based on the upper bound and the lower bound of the regulation power, the regulation power boundary data corresponding to each time period within the power grid dispatch cycle are obtained. The ramp rate value of the target intelligent building is obtained by summing all the extreme physical ramp rates corresponding to the target intelligent building, and the ramp rate value is used as the amplitude constraint between any pair of adjacent time periods within the power grid dispatch cycle.

[0015] The above scheme obtains the upper limit of the regulating power, representing the building's maximum power increase capability within a single time period, by summing the maximum heating power, maximum charging power, and maximum operating power; it obtains the lower limit of the regulating power, representing the building's maximum power reduction capability within a single time period, by summing the maximum cooling power, maximum discharging power, and ultimate discharging power and taking the amplitude; it then processes the upper and lower limit regulating power data for each time period to obtain the regulating power boundary data, representing the building's independent physical reachability for each time period; and finally, it sums the ultimate physical ramp rates of all flexible equipment to obtain ramp rate values, representing the limit of power change between adjacent time periods, and uses these as amplitude constraint data. This ensures that the acquisition of the boundary and constraint data is strictly based on the rated and ultimate values ​​of the equipment parameters, avoiding boundary deviations introduced by empirical estimations, and thus providing a precise initial search range that can completely cover all physically reachable points of the building for subsequent iterative verification.

[0016] As a preferred example, obtaining the initial boundary envelope of the target intelligent building based on the adjusted power boundary data and the amplitude constraint includes: Based on the adjustment power boundary data corresponding to each time period, the hourly power boundary inequality of the target smart building in the power grid dispatch cycle is obtained. Based on the amplitude constraints between any pair of adjacent time periods, the ramp constraint inequality of the smart building within the power grid dispatch cycle is obtained; Based on the adjusted power sequence of the target intelligent building in the power grid dispatch cycle, all the hourly power boundary inequalities, and all the ramp constraint inequalities, the initial boundary envelope of the target intelligent building in the power grid dispatch cycle is obtained.

[0017] The above scheme performs inequality transformation on the regulating power boundary data corresponding to each time period to obtain hourly power boundary inequality data that characterizes the regulating power of that time period as not lower than the lower bound and not higher than the upper bound; it performs inequality transformation on the amplitude constraints between any pair of adjacent time periods to obtain ramp constraint inequality data that characterizes the change in regulating power between adjacent time periods as not exceeding the ramp rate value; it then integrates the regulating power sequence data, all hourly power boundary inequality data, and all ramp constraint inequality data to obtain initial boundary envelope data that characterizes the initial adjustable capacity range of the building within the scheduling cycle. This unifies the power limit constraints originally scattered in each independent time period and the ramp constraints scattered in each pair of adjacent time periods into a complete envelope structure, avoiding omissions or overlaps caused by constraint dispersion, thereby improving the completeness of the initial boundary envelope in characterizing the physical limits and temporal constraints of the building's various flexible equipment.

[0018] As a preferred example, the step of iteratively verifying and cutting the initial boundary envelope according to the linear energy consumption model to obtain the target boundary envelope corresponding to the target smart building includes: Obtain the initial constraint coefficient matrix and initial constraint constant vector corresponding to the initial boundary envelope; Obtain the coefficient matrix and constant vector corresponding to the linear energy consumption model, and combine the coefficient matrix, the constant vector, the initial constraint coefficient matrix and the initial constraint constant vector through power balance constraints to obtain the joint constraint coefficient matrix and the joint constraint constant vector. Extract dual variable data based on the joint constraint coefficient matrix and the joint constraint constant vector; Based on the candidate adjustable power sequence corresponding to the initial constraint coefficient matrix and the initial constraint constant vector, the dual variable data, and the maximum prediction error bound of the rigid load, obtain the executability gap data of the candidate adjustable power sequence within the maximum prediction error bound of the rigid load; When the executability gap data is less than or equal to a preset tolerance threshold, the initial constraint coefficient matrix and the initial constraint constant vector are used as the target constraint coefficient matrix and the target constraint constant vector of the target boundary envelope.

[0019] The above scheme obtains joint constraint coefficient matrix and joint constraint constant vector data, representing the joint constraint relationship between the candidate boundary and the internal model, by simultaneously processing the initial constraint coefficient matrix and initial constraint constant vector data with the coefficient matrix and constant vector data of the linear energy consumption model; extracts dual variables from the joint constraint data to obtain dual variable data; performs executability quantification calculation on the candidate adjustable power sequence data, dual variable data, and rigid load maximum prediction error boundary data to obtain executability gap data, representing the degree of physical default under the worst-case scenario within the candidate boundary envelope; compares the executability gap data with a preset tolerance threshold, and when the executability gap data is less than or equal to the tolerance threshold, the current constraint matrix and constant vector are determined as the target constraint coefficient matrix and target constraint constant vector data. This allows the physical executability verification of the boundary envelope to be completed through deterministic quantitative indicators, rather than relying on empirical boundary margin estimation, thereby obtaining a target boundary envelope confirmed by numerical verification to have no physical default trajectory within it, and achieving precise definition of the building's adjustable capacity range.

[0020] As a preferred example, the step of iteratively verifying and cutting the initial boundary envelope according to the linear energy consumption model to obtain the target boundary envelope corresponding to the target smart building further includes: When the executability gap data is greater than the tolerance threshold, the candidate adjustment power sequence is used as the default adjustment power sequence and the dual variable data is used as the default dual variable data. Based on the default dual variable data and the default adjustment power sequence, obtain the tangent plane coefficient vector and the tangent plane constant term; Based on the tangent plane coefficient vector and the tangent plane constant term, the initial constraint coefficient matrix and the initial constraint constant vector are updated to obtain the updated constraint coefficient matrix and the updated constraint constant vector. The updated constraint coefficient matrix is ​​then used as the initial constraint coefficient matrix and the updated constraint constant vector is used as the initial constraint constant vector.

[0021] The above scheme judges and processes the results of comparing the executability gap data with the tolerance threshold. When the executability gap data is greater than the tolerance threshold, the candidate adjustment power sequence is determined as the default adjustment power sequence data, and the dual variable data is determined as the default dual variable data. The default dual variable data and the default adjustment power sequence data are processed to generate tangent planes to obtain tangent plane coefficient vectors and tangent plane constant terms representing the default regions that need to be removed in the current candidate boundary envelope. The initial constraint coefficient matrix and the initial constraint constant vector data are updated by appending the tangent plane coefficient vectors and tangent plane constant terms to obtain the updated constraint coefficient matrix and constraint constant vector data, which are used as new initial constraint data in the next iteration. This ensures that the region removed in each iteration accurately corresponds to the location of the verified default trajectory, avoiding the accidental removal of feasible regions due to indiscriminate contraction and the residual default regions due to insufficient removal. As a result, the updated boundary envelope gradually approaches the precise boundary of the building's actual executable adjustment range.

[0022] As a preferred example, obtaining the power boundary data of the target intelligent building within the power grid dispatch cycle based on the target boundary envelope and the preset target power function of the target intelligent building includes: Based on the target constraint coefficient matrix and target constraint constant vector of the target boundary envelope, the feasible space for the adjustment power of the target intelligent building within the power grid dispatch cycle is determined; For any time period within the power grid dispatch cycle, a preset target power function corresponding to that time period is obtained; wherein, the preset target power function includes a first objective function aimed at maximizing the regulation power of that time period and a second objective function aimed at minimizing the regulation power of that time period; Using the feasible space for adjustable power as a constraint, the maximum and minimum adjustable power of the target intelligent building during this time period are obtained according to the preset target power function; The maximum adjustable power and the minimum adjustable power for all time periods within the power grid dispatch cycle are arranged in chronological order to obtain the power boundary data of the target intelligent building within the power grid dispatch cycle.

[0023] The above scheme obtains feasible space data of adjustable power, representing the verified adjustable capacity range of the building, by performing feasible space determination processing on the target constraint coefficient matrix and target constraint constant vector data. For each time period, a first objective function aiming to maximize adjustable power and a second objective function aiming to minimize adjustable power are obtained. The first and second objective functions are then optimized and solved respectively using the feasible space of adjustable power as constraints to obtain the first optimal target value data representing the maximum power increase capability and the second optimal target value data representing the maximum power decrease capability for that time period. The entire... The first and second optimal target values ​​for each time period are processed sequentially to obtain power boundary data that characterizes the adjustable power range of the building in each time period within the scheduling cycle. This ensures that the final output power boundary values ​​for each time period are extracted by projecting from a physically executable high-dimensional temporal coupling feasible space. The boundaries of each time period maintain a constraint relationship consistent with the cross-time coupling characteristics such as the continuity of energy storage and thermal inertia within the building. This avoids the contradictions between adjacent time period boundaries caused by the independent calculation of the boundaries of each time period in the existing technology, which breaks the temporal coupling. This achieves the accuracy and self-consistency of the power boundary data in the temporal dimension.

[0024] On the other hand, the present invention discloses a system for determining the adjustment boundary of a building's participation in power grid dispatch, including an operation acquisition module, a multi-constraint module, an energy consumption assessment module, a dispatch constraint module, a boundary candidate module, a boundary update module, and a boundary generation module; The operation acquisition module is used to acquire the physical parameters of the target intelligent building and the operation prediction data and operation error data of the target intelligent building during the power grid dispatch cycle; The multi-constraint module is used to obtain the equipment operation constraints, power balance constraints, equipment constraints and energy constraints of the target intelligent building within the power grid dispatch cycle based on the physical parameters, the operation prediction data and the operation error data. The energy consumption assessment module is used to obtain the linear energy consumption model corresponding to the target intelligent building based on the equipment operation constraints, the power balance constraints, the equipment constraints, and the energy constraints. The scheduling constraint module is used to obtain the adjustment power boundary data and amplitude constraints of each flexible device in the target intelligent building during the power grid scheduling cycle based on the physical parameters. The boundary candidate module is used to obtain the initial boundary envelope of the target intelligent building based on the adjusted power boundary data and the amplitude constraint. The boundary update module is used to iteratively verify and cut the initial boundary envelope according to the linear energy consumption model to obtain the target boundary envelope corresponding to the target smart building. The boundary generation module is used to obtain the power boundary data of the target intelligent building within the power grid dispatch cycle based on the target boundary envelope and the preset target power function of the target intelligent building.

[0025] This invention discloses a system for determining the regulation boundaries of a building's participation in power grid dispatch. It acquires physical parameters, operational prediction data, and operational error data, and performs constraint construction processing on these data to obtain equipment operation constraints, power balance constraints, equipment constraints, and energy constraints characterizing the physical operational limitations of each device within the building. The system then performs model integration processing on these constraint data to obtain linear energy consumption model data characterizing the complete physical operational rules within the building. Simultaneously, it performs boundary extraction processing on the physical parameters to obtain regulation power boundary data and amplitude constraint data characterizing the independent physical limits of each flexible device. Finally, it performs envelope integration processing on these boundary data to obtain... The system obtains initial boundary envelope data representing the initial adjustable capacity range of a building; then iteratively verifies and segments the initial boundary envelope data and the linear energy consumption model data to obtain target boundary envelope data after verification and elimination of internal physical violation trajectories. The target boundary envelope data is then subjected to power projection processing to obtain power boundary data representing the adjustable power range of the building in each time period within the scheduling cycle. This ensures that the final output power boundary data internalizes the uncertainties represented by the operational error data and eliminates unexecutable false adjustment spaces through iterative verification, achieving an accurate depiction of the building's true adjustable capacity and improving the accuracy of the power boundary data. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating a method for determining the adjustment boundary of a building participating in power grid dispatch, as disclosed in an embodiment of the present invention. Figure 2 This is a simplified schematic diagram of the boundary of a smart building participating in power grid dispatch, as disclosed in an embodiment of the present invention; Figure 3 This is a schematic diagram of a system for determining the adjustment boundary of a building's participation in power grid dispatch, as disclosed in an embodiment of the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Reference Figure 1To improve the accuracy of regulation boundary generation, this embodiment discloses a method for determining the regulation boundary of a building's participation in power grid dispatch, mainly including: Step 101: Obtain the physical parameters of the target intelligent building and the operation prediction data and operation error data of the target intelligent building during the power grid dispatch cycle; Step 102: Based on the physical parameters, the operational prediction data, and the operational error data, obtain the equipment operation constraints, power balance constraints, equipment constraints, and energy constraints of the target intelligent building within the power grid dispatch cycle; Step 103: Obtain the linear energy consumption model corresponding to the target intelligent building based on the equipment operation constraints, the power balance constraints, the equipment constraints, and the energy constraints; Step 104: Obtain the adjustment power boundary data and amplitude constraints of each flexible device in the target intelligent building during the power grid dispatch cycle based on the physical parameters; Step 105: Obtain the initial boundary envelope of the target intelligent building based on the adjusted power boundary data and the amplitude constraint; Step 106: Iteratively verify and cut the initial boundary envelope according to the linear energy consumption model to obtain the target boundary envelope corresponding to the target intelligent building; Step 107: Based on the target boundary envelope and the preset target power function of the target intelligent building, obtain the power boundary data of the target intelligent building within the power grid dispatch cycle.

[0029] In this embodiment, the target intelligent building refers to a building body that integrates flexible equipment such as HVAC, electric energy storage systems, electric vehicle clusters, and transferable loads, and is capable of participating in grid dispatch. The grid dispatch cycle refers to the time range within which the grid dispatches the building, consisting of multiple discrete time periods. The physical parameters refer to parameter data describing the thermal characteristics of the target intelligent building and the intrinsic performance of its internal flexible equipment. The operational prediction data refers to the estimated operational status data of the target intelligent building for each time period within the grid dispatch cycle. The operational error data refers to data describing the maximum possible deviation range of each predicted value in the operational prediction data. The equipment operation constraints refer to constraints that limit the operational status of the HVAC system within the target intelligent building. The power balance constraints refer to constraints that limit the balance between the interactive power at the grid connection point and the power consumption and generation of each internal device. The equipment constraints refer to constraints that limit the power output range of each flexible device within the target intelligent building. The energy constraints refer to constraints that limit the energy state of devices within the target intelligent building that have energy storage or energy accumulation characteristics. The above-mentioned constraints are obtained based on the physical parameters, the operational prediction data, and the operational error data, through corresponding physical modeling and constraint derivation.

[0030] Next, the linear energy consumption model refers to a mathematical model that integrates the equipment operation constraints, power balance constraints, equipment constraints, and energy constraints to describe all energy consumption rules of the target intelligent building within the power grid dispatch cycle. This linear energy consumption model expresses the operating status of each device within the building in the form of decision variables, and expresses each constraint condition in the form of linear equations and linear inequalities, thus forming a complete linear constraint system. The regulating power boundary data refers to the maximum range of regulating power that each flexible device in the target intelligent building can provide in each time period within the power grid dispatch cycle, considering its own physical power limit independently, including the upper and lower bounds of the regulating power. The amplitude constraint refers to the constraint condition that limits the maximum allowable amplitude of the change in regulating power of the target intelligent building between adjacent time periods, determined based on the physical ramp rate limit of each flexible device.

[0031] Then, the initial boundary envelope refers to the data set defined by the regulating power boundary data and the amplitude constraints, used to characterize the initial feasible range of regulating power of the target intelligent building within the power grid dispatch cycle. This initial boundary envelope only considers the independent physical power limit and ramp rate limit of each flexible device, serving as the initial input for subsequent iterative processing; while the iterative verification and cutting refers to a cyclical process that starts with the initial boundary envelope, uses the linear energy consumption model to verify whether there are physically unexecutable regulating trajectories within the current boundary envelope, and cuts and shrinks the current boundary envelope to update it and performs verification again when an unexecutable trajectory is found. The target boundary envelope refers to the boundary envelope data finally obtained after the iterative verification and cutting, which has been verified to have no physically defaulted regulating trajectories within it, and represents the adjustable capability range output by the target intelligent building to the power grid dispatch terminal.

[0032] Finally, the preset target power function refers to a pre-defined function that optimizes the regulating power of the target intelligent building in each time period within the power grid dispatch cycle. This includes functions aiming to maximize regulating power and functions aiming to minimize regulating power. The power boundary data refers to the adjustable power range that the target intelligent building can safely deliver in each time period within the power grid dispatch cycle. This includes the maximum upward and maximum downward adjustable power for each time period, and is the final output of the method used to report the regulating capacity to the power grid dispatch center or the electricity market.

[0033] In summary, the above method acquires physical parameters, operational prediction data, and operational error data, and performs constraint construction processing on these data to obtain equipment operation constraints, power balance constraints, equipment constraints, and energy constraints characterizing the physical operational limitations of each device within the building. It then performs model integration processing on these constraint data to obtain linear energy consumption model data characterizing the complete physical operational rules within the building. Simultaneously, it performs boundary extraction processing on the physical parameters to obtain adjustable power boundary data and amplitude constraint data characterizing the independent physical limits of each flexible device, and performs envelope integration processing on these boundary data to obtain data characterizing the building's initial adjustable energy. The system obtains initial boundary envelope data of the power range; then iteratively verifies and segments the initial boundary envelope data and the linear energy consumption model data to obtain target boundary envelope data after verification and elimination of internal physical violation trajectories. The target boundary envelope data is then subjected to power projection processing to obtain power boundary data characterizing the adjustable power range of the building in each time period within the scheduling cycle. This ensures that the final output power boundary data internalizes the uncertainties represented by the operational error data and eliminates unexecutable false adjustment spaces through iterative verification, achieving an accurate depiction of the building's true adjustable capacity and improving the accuracy of the power boundary data.

[0034] In this embodiment, step 101 includes: Step 1011: Obtain the building parameters, historical operating data, and equipment parameters of each flexible device in the target intelligent building, so as to obtain the physical parameters of the target intelligent building based on the building parameters and the equipment parameters; wherein, the historical operating data includes historical electricity consumption time series data, historical uncontrollable load time series data, a first error sequence between historical outdoor temperature prediction values ​​and historical outdoor temperature measured values, and a second error sequence between historical rigid load prediction values ​​and historical rigid load measured values; Step 1012: Based on the historical electricity consumption time series data, obtain the baseline power prediction sequence of the target smart building within the power grid dispatch cycle through a preset electricity consumption prediction model; Step 1013: Based on the historical uncontrollable load time series data, obtain the rigid load prediction sequence of the target smart building within the power grid dispatch cycle through a preset load prediction model; Step 1014: Take the maximum absolute value of each error in the first error sequence as the maximum prediction error bound of outdoor temperature, and take the maximum absolute value of each error in the second error sequence as the maximum prediction error bound of rigid load, so as to use the maximum prediction error bound of outdoor temperature and the maximum prediction error bound of rigid load as the operating error data. Step 1015: Obtain the outdoor temperature prediction sequence of the target smart building during the power grid dispatch cycle, so as to use the baseline power prediction sequence, the rigid load prediction sequence and the outdoor temperature prediction sequence as the operation prediction data of the target smart building during the power grid dispatch cycle.

[0035] In this embodiment, the building parameters include the building's equivalent heat capacity and equivalent thermal resistance, used to describe the thermal inertia and heat exchange characteristics between the target intelligent building's envelope and indoor air. The equipment parameters include the heating and cooling energy efficiency ratios, maximum heating and cooling power, maximum charging and discharging power, charging and discharging efficiency, and ultimate energy capacity of each HVAC system, the ultimate charging and discharging power of each electric vehicle cluster, and the maximum operating power and ultimate total energy demand of each transferable load. The building parameters can be obtained from the building energy management system or design drawings, and the equipment parameters can be obtained from the nameplate information of each device. The historical electricity consumption time-series data refers to the time-series data of the total electricity consumption at the grid connection point of the target intelligent building in the past period. The historical uncontrollable load time-series data refers to the time-series data of the electricity consumption of uncontrollable loads inside the building (such as basic lighting, security power, etc.) in the past period. The first error sequence refers to the sequence formed by the deviations between predicted and measured outdoor temperatures at multiple historical moments. The second error sequence refers to the sequence of deviations between predicted and measured values ​​of rigid loads at multiple historical points in time. This historical operational data can be obtained from the historical database of the building energy management system.

[0036] The baseline power prediction sequence represents the natural predicted power trajectory of the target intelligent building when it does not participate in grid interaction and only meets internal load and comfort requirements. The preset electricity consumption prediction model can be a regression model or a neural network model based on historical data. By inputting the historical electricity consumption time-series data and corresponding weather, date type, and other characteristic data, it outputs the predicted values ​​of the baseline power for each time period within the grid dispatch cycle, forming the baseline power prediction sequence. Similarly, the rigid load prediction sequence refers to the predicted electricity consumption of the uncontrollable basic load within the target intelligent building for each time period within the grid dispatch cycle. The preset load prediction model can be a regression model or a neural network model based on historical data. By inputting the historical uncontrollable load time-series data and corresponding period-related characteristic data, it outputs the predicted values ​​of the rigid load for each time period within the grid dispatch cycle, forming the rigid load prediction sequence.

[0037] Finally, the absolute value of each error in the first error sequence is obtained, that is, the absolute value of each error value in the first error sequence is taken, and then the maximum value is found from all the absolute values, which is used as the maximum prediction error bound for outdoor temperature. Similarly, the absolute value of each error in the second error sequence is obtained, and the maximum value is found as the maximum prediction error bound for rigid load. The maximum prediction error bound for outdoor temperature means that the deviation of the actual outdoor temperature value from its predicted value at any time during the power grid dispatching cycle does not exceed the range of this error bound. The maximum prediction error bound for rigid load means that the deviation of the actual rigid load value from its predicted value at any time during the power grid dispatching cycle does not exceed the range of this error bound; and the outdoor temperature prediction sequence can be obtained by accessing the forecast data interface of the meteorological service agency, so that the baseline power prediction sequence, the rigid load prediction sequence, and the outdoor temperature prediction sequence obtained in the aforementioned steps can be used together as the operational prediction data.

[0038] The above implementation method acquires building parameters, historical operating data, and equipment parameters, and performs predictive processing on the historical electricity consumption time series data to obtain baseline power prediction sequence data characterizing the natural electricity consumption trend of the building without scheduling intervention; performs predictive processing on the historical uncontrollable load time series data to obtain rigid load prediction sequence data characterizing the electricity consumption trend of uncontrollable loads within the building; performs statistical analysis on the historical prediction error sequence data to obtain prediction error boundary data characterizing the maximum deviation range of outdoor temperature and rigid load; and performs classification and integration processing on the baseline power prediction sequence data, rigid load prediction sequence data, outdoor temperature prediction sequence data, and prediction error boundary data to obtain operating prediction data and operating error data characterizing the building's operating prediction and error distribution characteristics. This ensures that the acquisition of the above data relies entirely on the building's own historical operating records, avoiding subjective biases introduced by manual experience coefficients, and thus providing a data foundation that objectively reflects the actual operating laws of the building for subsequent boundary construction.

[0039] In this embodiment, the building parameters include the building's equivalent heat capacity and equivalent thermal resistance; the equipment parameters include the heating efficiency ratio, cooling efficiency ratio, maximum heating power and maximum cooling power of each HVAC system, the maximum charging and discharging power, charging and discharging efficiency and ultimate energy capacity of each electric energy storage system, the ultimate charging and discharging power of each electric vehicle cluster, and the maximum operating power and ultimate total energy demand of each transferable load; step 102 includes: Step 1021: Based on the building's equivalent heat capacity, the building's equivalent thermal resistance, the heating energy efficiency ratio, the cooling energy efficiency ratio, the outdoor temperature prediction sequence, and the outdoor temperature maximum prediction error boundary, obtain the equipment operation constraints of the target smart building within the power grid dispatch cycle; Step 1022: Obtain the predicted power generation sequence of each distributed power source in the target smart building during the grid dispatch cycle, so as to obtain the power balance constraint of the target smart building during the grid dispatch cycle based on the predicted power generation sequence, the baseline power prediction sequence and the rigid load prediction sequence; Step 1023: Based on the maximum heating power, the maximum cooling power, the maximum charging and discharging power, and the ultimate charging and discharging power, obtain the equipment constraints of the target smart building within the power grid dispatching cycle; Step 1024: Based on the charging and discharging efficiency, the limit energy capacity, the maximum operating power, and the limit total energy demand, obtain the energy constraint of the target smart building within the power grid dispatch cycle.

[0040] In this embodiment, smart buildings typically incorporate distributed power sources, including distributed photovoltaics and micro wind turbines. Internal building loads can generally be categorized into two types based on controllability: uncontrollable rigid loads (such as basic lighting and security power) and controllable flexible loads. Flexible loads may further include HVAC systems, energy storage systems, electric vehicles, and transferable loads.

[0041] Treating a smart building as a whole, its thermodynamic dynamics are described using a first-order equivalent thermal parameter (ETP) model, in continuous time... Below, indoor building temperature The dynamic differential equation can be expressed as: in, For the building's equivalent heat capacity, Equivalent thermal resistance; Outdoor temperature; The total internal heat power generated by heat dissipation from indoor infrastructure and personnel, solar radiation, etc. The heat power injected into the room by the HVAC system.

[0042] To adapt to all-season operating conditions, the input power of the HVAC system needs to be explicitly decomposed into heating power. With cooling power The physical equation for the energy conversion between these two and the net output heat power is as follows: in, and These are the heating energy efficiency ratio and the cooling energy efficiency ratio of the equipment, respectively.

[0043] Assuming a given scheduling time step Inside, the input power and environmental parameters remain approximately constant. Discretizing the differential equation for the dynamic change of indoor temperature, the time-series variation of indoor temperature can be obtained as follows: Among them, coefficient satisfy , ; , ; Outdoor temperature over discrete time periods; parameters .

[0044] Among them, the building indoor temperature The dynamic change differential equation is a common first-order linear ordinary differential equation, which can be equivalently transformed into: Assume the current scheduling period is t, and the corresponding time interval is [t-Δt, t]. Assume that within the time step Δt, the environmental parameters and various input powers remain constant. Therefore, we can... , , Let be a discrete constant. Taking a definite integral over [t-Δt, t] on both sides of the above equation, we get: After integration, substituting the results into the expression for the energy conversion equation above and simplifying, we obtain: In the formula, The building's indoor temperature self-sustaining coefficient. The outdoor ambient temperature influence coefficient. and These are the equivalent heating temperature rise coefficient and the equivalent cooling temperature drop coefficient of the HVAC system, respectively. To comprehensively account for the equivalent temperature rise due to internal heat gain, the heating and cooling electrical power must satisfy the following: Indoor temperature must meet the user's comfort requirements. To avoid a false regulation capability that deviates from actual physical logic due to simultaneous large-scale heating and cooling, and to avoid introducing binary variables, a duty cycle constraint is introduced for heating and cooling power: ;in, and These are the upper limits for heating and cooling power of the HVAC system, respectively. This constraint ensures that the heating and cooling power does not exceed the corresponding upper limits, while also physically aligning with actual operating conditions: allowing for simultaneous heating and cooling needs in different rooms within a large building, and also allowing for smooth switching between heating and cooling modes within a relatively long timeframe; the aforementioned... These represent the minimum and maximum permissible indoor temperatures, respectively.

[0045] Generally, HVAC systems can only operate in either heating or cooling mode at any given time. Traditional constraints typically introduce binary variables. By modeling, we can obtain: However, this model has two significant drawbacks when constructing the overall adjustable boundary of a smart building: First, binary variables cause the building's flexible feasible domain to exhibit a non-convex, discontinuous, and "fragmented" state. For the upper-level power grid or aggregator, its scheduling and resource integration are highly dependent on a continuous and convex flexible boundary (such as a linear polyhedron); discrete and fragmented boundaries will lead to a more severe curse of dimensionality in aggregation calculations, making it difficult for the upper level to effectively evaluate and utilize them. Second, when a smart building is considered as a whole, different areas within it may simultaneously have mixed demands for cooling and heating. If absolute mutual exclusion of states is forcibly required, or if the net power after cooling and heating is simply used to replace the actual electrical power, the thermodynamic temperature transfer equation will have serious calculation deviations because the energy efficiency ratios of cooling and heating equipment are usually different; in addition, this strict mutual exclusion cannot reflect the smooth switching process of equipment switching between cooling and heating modes within the same long scheduling period.

[0046] To avoid introducing binary variables and to allow the building to exhibit the macroscopic characteristics of simultaneous cooling and heating, while constraining equipment from simultaneously outputting high power for both cooling and heating to prevent the creation of a "false regulatory capacity" that deviates from actual physical logic, the concept of time or space allocation ratio (i.e., duty cycle) is introduced here. Assuming that during a time period... Internal allocation The proportion (space capacity or runtime) used to Heating, distribution The proportion used to If refrigeration is used, and the sum of the two ratios must not exceed 100% of the total available share, then all of the following constraints must be met: in, and These are the maximum heating power and the maximum cooling power, respectively. It's important to note that these constraints can be simplified and are strictly equivalent to: .

[0047] Assume the predicted value of the rigid load is In order to support the adjustable power output The power interaction between various loads within the building and the building-grid needs to meet the power balance requirements: in, The power generation capacity of distributed power sources within intelligent buildings; and These represent the discharge power and charging power of the energy storage system, respectively; the power consumption for ventilation in HVAC systems is included in the rigid load. , and These are the input electrical power for heating and cooling, respectively; and These represent the charging power and discharging power of the electric vehicle cluster, respectively. This refers to the electrical power consumed by transferable loads.

[0048] For energy storage systems, the discharge power and charging power must meet the following requirements: Similar to the handling of heating and cooling power in HVAC systems, a duty cycle constraint for the charging and discharging power of electrical energy storage is introduced: The energy transfer equation for the electric energy storage system is as follows: in, Energy for electric energy storage systems; and Their discharge efficiency and charging efficiency are respectively, and both satisfy the following conditions: ;in, and These represent the minimum and maximum energy values ​​of the energy storage system, respectively.

[0049] Substituting the energy transfer equation into the calculation expressions for discharge efficiency and charging efficiency, we obtain: .

[0050] Unlike stationary energy storage systems, the charging and discharging capabilities and energy storage capacity of a cluster of electric vehicles parked in a building dynamically change over time as the vehicles arrive and leave. Specifically, the dynamic upper and lower limits of the charging and discharging power of the electric vehicle cluster are limited by the total number of electric vehicles currently parked in the smart building, representing the maximum available charging and discharging power of the cluster.

[0051] set up and They are respectively A collection of electric vehicles arriving at and departing from the building during designated time periods. for A collection of electric vehicles that stop during designated time periods, which meets the requirements The maximum charging and discharging power of the electric vehicle cluster is: in, and Electric vehicles The upper limit of charging and discharging power; and These represent the upper limits of charging and discharging power for the cluster, respectively. The charging and discharging power of the electric vehicle cluster must meet the following requirements: Similarly, a duty cycle constraint is introduced for the charging and discharging power of the electric vehicle cluster: .

[0052] The process of obtaining the dynamic upper and lower limits of cluster energy is as follows: To ensure that each vehicle within the cluster can meet its individual energy needs during the control process (i.e., reach the target energy level upon departure), the energy boundary of a single electric vehicle is first characterized. For any electric vehicle... Its energy upper limit This corresponds to "charging as early as possible," meaning that vehicles are charged to their maximum capacity upon arrival at the depot and maintained until departure; its lower energy limit. This corresponds to "latest charging," meaning that after the vehicle arrives, it first discharges to the grid at maximum power to the lower limit, and then charges at maximum power just before leaving the site, so that its charging needs are met at the moment of departure.

[0053] By superimposing the energy boundaries of each electric vehicle, the dynamic upper and lower limits of the cluster energy can be obtained: in, and Electric vehicles During the period The upper and lower limits of energy; and These are the upper and lower limits of the cluster's energy; the cluster's energy should not exceed these limits. ;in, Energy for electric vehicle clusters.

[0054] Furthermore, the process of obtaining the cluster energy transfer equation is as follows: the evolution of cluster energy is affected not only by charging and discharging behavior, but also by the energy changes caused by vehicles entering and leaving the cluster. The equation is as follows: in, and These are the equivalent charge and discharge power efficiencies of the cluster, respectively. for The energy brought by electric vehicles entering the depot at any given moment is [value missing]. ,in For electric vehicles Initial energy upon arrival at the station; for The energy carried away by the electric vehicle as it leaves the depot is [value missing]. ,in This represents the expected charge level when the vehicle leaves the site. It's important to note that this assumes all electric vehicles have the same charging efficiency. The discharge efficiency of all electric vehicles is (e.g., 0.95). Thus, at the electric vehicle cluster level, they represent the cluster's charging and discharging efficiency, respectively.

[0055] Substituting the cluster energy transfer equation into the upper and lower bound constraints of the cluster energy yields: .

[0056] Power consumption of transferable loads in smart buildings Subject to the physical operating capacity of the equipment Limitations: Its accumulated electrical energy consumption satisfy: .

[0057] To ensure the feasibility of the transferable load and guarantee its successful completion before the deadline, the cumulative energy consumption of the transferable load should meet the following requirements: ;in, and These represent the lower and upper limits of cumulative energy consumption, reflecting the electricity demand of the transferable load. Substituting the conditions for meeting the cumulative energy consumption into the cumulative energy consumption of the transferable load yields the operating constraints of the transferable load: .

[0058] The above implementation method divides physical parameters into building equivalent heat capacity and building equivalent thermal resistance data, which characterize the building's thermal performance, and equipment parameter data, such as heating efficiency ratio, cooling efficiency ratio, power limit, charge and discharge efficiency, energy capacity, and ultimate total energy demand, which characterize the rated performance of each device. The building parameter data is then combined with outdoor temperature prediction sequence data and outdoor temperature maximum prediction error boundary data to construct thermodynamic constraints, thereby obtaining equipment operation constraint data characterizing the operating temperature limits of HVAC systems. Power limit data in the equipment parameters is then subjected to power limiting processing to obtain equipment constraint data characterizing the upper limit of each device's power. Efficiency and capacity data in the equipment parameters are then subjected to energy recursion processing to obtain energy constraint data characterizing the energy state transition limits of energy storage and transferable loads. Power balance processing is then performed on distributed power generation prediction sequence data, baseline power prediction sequence data, and rigid load prediction sequence data to obtain power balance constraint data characterizing the power balance relationship at the building's grid connection point. This ensures a precise correspondence between the construction of each constraint and specific parameters, avoiding constraint deviations caused by parameter mixing, and thus improving the accuracy of the linear energy consumption model in depicting the operating characteristics of various devices within the building.

[0059] In this embodiment, step 103 includes: Step 1031: Divide the equipment operation constraints, the power balance constraints, the equipment constraints, and the energy constraints into a set of linear equality constraints and a set of linear inequality constraints; Step 1032: For any constraint in the set of linear equality constraints, split the constraint into a pair of linear inequality constraints with opposite directions. Step 1033: Based on all the linear inequality constraints corresponding to the linear equality constraint set and the linear inequality constraint set, obtain the inequality constraint set of the target intelligent building within the power grid dispatch cycle; Step 1034: Obtain the decision variables, constant terms, and variable coefficients of each inequality constraint in the inequality constraint set; Step 1035: Concatenate all the decision variables corresponding to the inequality constraint set into a decision variable column vector over the power grid dispatching cycle; Step 1036: Determine the matrix dimension based on the number of inequality constraints in the inequality constraint set and the dimension of the decision variable column vector; Step 1037: Obtain matrix elements based on the inequality constraint set, the decision variable column vector, and the variable coefficients; Step 1038: Obtain the constant term in each inequality constraint in the inequality constraint set, so as to obtain the linear energy consumption model of the target smart building in the power grid dispatch cycle based on the constant term, the matrix dimension, the matrix elements and the decision variable column vector.

[0060] In this embodiment, the constraints are categorized according to their mathematical forms. Specifically, the power balance constraint is an equality constraint, the recursive equation in the energy transfer equation of the energy storage system is an equality constraint, and the recursive equation for the cumulative energy consumption of the transferable load is an equality constraint. These equality constraints are all classified into the linear equality constraint set. The room temperature comfort constraint in the equipment operation constraints, the duty cycle constraints and power upper limit constraints in the equipment constraints, and the energy upper and lower bound constraints in the energy constraints are all inequality constraints. These inequality constraints are all classified into the linear inequality constraint set.

[0061] For each equality constraint in the set of linear equality constraints, it is equivalently transformed into two inequality constraints. For example, the equality "A = B" is split into two inequalities "A ≤ B" and "A ≥ B", and then multiplying both sides of "A ≥ B" by -1 transforms it into "-A ≤ -B", thus unifying it into a linear inequality constraint of the form "≤".

[0062] All linear inequality constraints of the form "≤" obtained by decomposing the equality constraints are merged with the original linear inequality constraints in the set of linear inequality constraints to form the set of inequality constraints. After the above processing, all constraints in the set of inequality constraints retain the characteristic of continuous linearity in mathematics and contain only inequality constraints.

[0063] For each inequality constraint in the inequality constraint set, identify the decision variables contained therein, the coefficients corresponding to each decision variable, and the constant term on the right side of the inequality. The decision variables are the operating state variables of each flexible device within the building at different time periods, including the heating and cooling power of HVAC systems, the charging and discharging power of energy storage systems, the charging and discharging power of electric vehicle clusters, the power consumption of transferable loads, the energy of energy storage systems, the energy of electric vehicle clusters, and the cumulative energy consumption of transferable loads. All decision variables involved in the inequality constraint set are concatenated into a column vector, denoted as x, in a uniform variable arrangement order across all time periods within the power grid dispatch cycle. This column vector x is a column vector formed by concatenating all continuous decision variables within the building over the entire time period. Let the total number of inequality constraints in the inequality constraint set be M, and the dimension of the decision variable column vector x be D; then the matrix is ​​determined to have M rows and D columns. For the i-th constraint in the inequality constraint set, the coefficient of the j-th decision variable in the column vector x of the decision variables in that constraint is taken as the element in the i-th row and j-th column of the matrix; if the j-th decision variable is not included in the constraint, the corresponding element is zero. This element assignment is performed one by one for i from 1 to M and j from 1 to D to determine all elements of the matrix. The constant term on the right-hand side of the i-th inequality constraint in the inequality constraint set is taken as the i-th element of the constant vector.

[0064] Specifically, the operating environment of intelligent buildings faces multi-source uncertainties, such as temperature prediction errors and load prediction errors, which are assumed to be characterized by a box-type uncertainty set. Considering the relative independence of the uncertainty sources of HVAC, electric vehicles, and transferable loads, the worst-case boundary contraction in analytical form can be directly implemented on their local constraint boundaries to resist the accumulation of uncertainties within the box-type uncertainty set and ensure the conservatism of the boundary of intelligent building participation in grid interaction. Therefore, the boundary contraction of the HVAC system operating constraints is as follows: Considering the uncertainty of outdoor temperature Uncertainty regarding heat gain within the building Define the cumulative term based on predicted values ​​in the time-series variation of indoor temperature. for: Since the thermal dynamic system is purely linear and the transfer coefficient is non-negative, for a bounded box perturbation, the above cumulative term at time... The maximum absolute deviation can be obtained directly through analytical summation: To ensure that the room temperature comfort constraint holds for any box-shaped perturbation, the comfort boundary can be conservatively contracted: .

[0065] The boundary contraction process of the electric vehicle cluster operation constraints is as follows: considering the energy prediction errors caused by electric vehicles entering and leaving the cluster. The equation obtained by substituting the cluster energy transfer equation into the upper and lower bound constraints of the cluster energy can be reduced to: The process of boundary contraction of transferable load operation constraints is as follows: Definition Deadline The upper bound of the cumulative energy demand uncertainty of the aggregated load is determined by factors such as some tasks arriving early or individual task loads slightly exceeding expectations. To ensure that the building's internal control system can still reserve sufficient power margin to complete all tasks on time under the worst-case scenario of sudden demand increases or severe task backlog, without violating the maximum full-load operating limits of the equipment, the upper and lower edges of the cumulative energy boundary of the transferable load are directly reduced inward simultaneously. The operating constraints of the transferable load are rewritten as follows: After local analytical contraction for each heterogeneous physical resource, each constraint mathematically retains its continuous linear characteristics and contains only inequality constraints. Based on this, the integrated smart building energy consumption model can be uniformly abstracted into the following matrix inequality form: in, A column vector composed of all continuous decision variables within the building over the entire time period; a constant matrix. and constant vector This corresponds to all the constraints of the smart building energy consumption model after local robust shrinkage. It is particularly noteworthy that the vector... The uncertainty deviation term of local boundary contraction has been absorbed.

[0066] It should be noted that this implementation provides a partial example for illustrative purposes only. Specifically, it assumes that only the boundary contraction of equipment operation constraints and HVAC system operation constraints is considered, along with the corresponding variables (the complete constraints can be extended in the same way for further consideration), as follows: Assuming only consider The time period (which can also be extended to 12, 24, etc.) is then represented by a column vector composed of all decision variables over the entire time period. for: Based on the boundary contraction of the equipment operation constraints and the HVAC system operation constraints, write out the corresponding complete matrix form: Representing a complete matrix or vector in a compact form is as follows: ; It should be noted that when further constraints are included, a matrix needs to be added. The rows, and the corresponding extended vectors. If other variables are included, vectors need to be added. OK.

[0067] The above implementation method classifies and processes equipment operation constraints, power balance constraints, equipment constraints, and energy constraints to obtain linear equality constraint sets and linear inequality constraint sets. It then decomposes and transforms the constraints in the linear equality constraint sets to obtain inequality constraint data that is consistent with the original linear inequality constraint forms. Finally, it performs full-time splicing processing on the decision variables in all inequality constraints to obtain decision variable column vector data. Statistical processing is performed on the number of inequality constraints and the dimension of the decision variable column vectors to determine the matrix dimension data. Finally, it extracts the variable coefficients and constant terms from each inequality constraint to obtain matrix element data and constant term data, thereby forming linear energy consumption model data expressed by coefficient matrices and constant vectors. This allows multiple types of constraints, originally of varying forms and with scattered variables, to be integrated into a standardized matrix inequality system without information loss, avoiding approximations or simplifications during constraint transformation and thus improving the accuracy of the mathematical expression of the linear energy consumption model.

[0068] In this embodiment, the equipment parameters include the limit physical ramp rate of each flexible device, and step 104 includes: Step 1041: For any time period within the power grid dispatch cycle, sum the maximum heating power, the maximum charging power among the maximum charging and discharging power, and the maximum operating power to obtain the upper bound of the regulation power corresponding to that time period; Step 1042: Sum the maximum cooling power, the maximum discharge power in the maximum charge and discharge power and the limit discharge power in the limit charge and discharge power and take the amplitude to obtain the lower limit of the adjustment power corresponding to this time period; Step 1043: Based on the upper bound and the lower bound of the regulation power, obtain the regulation power boundary data corresponding to each time period within the power grid dispatching cycle; Step 1044: Sum all the extreme physical ramp rates corresponding to the target intelligent building to obtain the ramp rate value of the target intelligent building, and use the ramp rate value as the amplitude constraint between any pair of adjacent time periods within the power grid dispatch cycle.

[0069] In this embodiment, before accurately characterizing the dynamic boundary of smart building participation in grid interaction, a broad initial polyhedral envelope is first constructed based on the maximum physical interaction capability of the grid connection point. This initial envelope only considers the static algebraic sum of the physical power limits of each flexible resource, and temporarily ignores complex temporal energy state coupling and global uncertainty disturbances.

[0070] Set in time period Intelligent building external power regulation The absolute physical lower limit and upper limit are respectively and The extreme value of the maximum gradient in adjacent time periods is The constraints of the initial polyhedral envelope can be expressed as: Among them, absolute upper and lower limits and The maximum gradeability can be obtained by directly summing the ultimate capabilities of each flexible device within the building during the corresponding time period (such as the maximum charge / discharge limit of the energy storage system, the rated operating power of the HVAC system, and the maximum apparent capacity of the electric vehicle depot's basic hardware facilities); This is approximated by the sum of the upper limits of the physical ramp rates of each device. This static relaxation process provides the outermost basic boundary guarantee for the overall table interaction trajectory. Specifically, for any time period t within the grid dispatch cycle, the maximum heating power of HVAC, the maximum charging power of the electric energy storage system, the maximum charging power of the electric vehicle cluster, and the maximum operating power of the transferable load are summed. The sum is used as the upper bound of the regulation power for that time period, which represents the maximum increment of power absorbed by the building from the grid during that time period. For any time period t within the grid dispatch cycle, the maximum cooling power of HVAC, the maximum discharge power of the electric energy storage system, and the limit discharge power of the electric vehicle cluster are summed, and the negative value of the sum is used as the lower bound of the regulation power for that time period, which represents the maximum extent to which the building reduces the power absorbed from the grid or the power fed back to the grid during that time period.

[0071] In the above summation process, the maximum charge / discharge limit of the energy storage system, the rated operating power of the HVAC system, and the maximum apparent capacity of the electric vehicle depot's basic hardware facilities are all included as the limit capabilities of each device in the corresponding time period. Then, for each time period within the power grid dispatch cycle, the above summation operation is performed to obtain the upper and lower bounds of the regulation power for each time period. These upper and lower bounds are used as the regulation power boundary data. The maximum ramp rate is approximated by the sum of the upper limits of the physical ramp rates of each device. Specifically, the limit physical ramp rates corresponding to each flexible device in the target intelligent building are obtained, and all limit physical ramp rates are summed to obtain the ramp rate value. The ramp rate value is used as the amplitude constraint between any pair of adjacent time periods within the power grid dispatch cycle, indicating that the absolute value of the change in regulation power between two adjacent time periods must not exceed the ramp rate value. This static relaxation process provides the outermost basic boundary guarantee for the overall table interaction trajectory.

[0072] The above implementation method sums up the maximum heating power, maximum charging power, and maximum operating power to obtain the upper limit data of the regulating power, which characterizes the building's maximum power increase capability in a single time period; sums up the maximum cooling power, maximum discharging power, and ultimate discharge power and takes the amplitude to obtain the lower limit data of the regulating power, which characterizes the building's maximum power reduction capability in a single time period; the upper and lower limit data of the regulating power are processed on a time-by-time basis to obtain the boundary data of the regulating power, which characterizes the independent physical reachability of the building in each time period; and the ultimate physical ramp rate of all flexible equipment is summed up to obtain the ramp rate value, which characterizes the power change limit between adjacent time periods of the building, and is used as the amplitude constraint data. This ensures that the acquisition of the above boundary data and constraint data is strictly based on the rated and ultimate values ​​of the equipment parameters, avoiding the boundary deviation introduced by empirical estimation, and thus providing a precise initial search range that can completely cover all physical reachability points of the building for subsequent iterative verification.

[0073] In this embodiment, step 105 includes: Step 1051: Based on the adjustment power boundary data corresponding to each time period, obtain the hourly power boundary inequality of the target smart building within the power grid dispatch cycle; Step 1052: Based on the amplitude constraints between any pair of adjacent time periods, obtain the ramp constraint inequality of the smart building within the power grid dispatch cycle; Step 1053: Based on the adjustment power sequence of the target intelligent building in the power grid dispatching cycle, all the hourly power boundary inequalities, and all the ramp constraint inequalities, obtain the initial boundary envelope of the target intelligent building in the power grid dispatching cycle.

[0074] Before precisely characterizing the dynamic boundaries of smart building interaction with the power grid, this embodiment first constructs a broad initial polyhedral envelope based on the maximum physical interaction capability of the grid connection point. The initial envelope only considers the static algebraic sum of the physical power limits of each flexible resource, and does not take into account complex temporal energy state coupling and global uncertainty disturbances.

[0075] Set in time period Intelligent building external power regulation The absolute physical lower limit and upper limit are respectively and The extreme value of the maximum gradient in adjacent time periods is The constraints of the initial polyhedral envelope can be expressed as: Among them, absolute upper and lower limits and The maximum gradeability can be obtained by directly summing the ultimate capabilities of each flexible device within the building during the corresponding time period (such as the maximum charge / discharge limit of the energy storage system, the rated operating power of the HVAC system, and the maximum apparent capacity of the electric vehicle depot's basic hardware facilities); This can be approximated by the sum of the upper limits of the physical ramp rates of each device. This static relaxation process provides the outermost basic boundary guarantee for the overall table interaction trajectory.

[0076] The above hourly power extrema and ramping constraints are uniformly converted into a standard system of linear matrix inequalities, denoted as the initial boundary envelope: Among them, matrix With column vectors These are the constraint coefficient matrix and constant vector corresponding to the construction of the initial static boundary, respectively.

[0077] In the subsequent algorithm process, the polyhedral envelope will serve as the basic feasible region for iterative optimization. It will be gradually cut and shrunk through global duality verification and tangent plane generation until all potential temporal physical state conflicts and uncertain default risks are completely eliminated.

[0078] The above implementation method performs inequality transformation on the regulating power boundary data corresponding to each time period to obtain hourly power boundary inequality data that characterizes the regulating power of that time period as not lower than the lower bound and not higher than the upper bound; performs inequality transformation on the amplitude constraints between any pair of adjacent time periods to obtain ramp constraint inequality data that characterizes the change in regulating power between adjacent time periods as not exceeding the ramp rate value; and integrates the regulating power sequence data, all hourly power boundary inequality data, and all ramp constraint inequality data to obtain initial boundary envelope data that characterizes the initial adjustable capacity range of the building within the scheduling cycle. This unifies the power limit constraints originally scattered in each independent time period and the ramp constraints scattered in each pair of adjacent time periods into a complete envelope structure, avoiding omissions or overlaps caused by constraint dispersion, thereby improving the completeness of the initial boundary envelope in characterizing the physical limits and temporal constraints of the building's various flexible equipment.

[0079] In this embodiment, step 106 includes: Step 1061: Obtain the initial constraint coefficient matrix and initial constraint constant vector corresponding to the initial boundary envelope; Step 1062: Obtain the coefficient matrix and constant vector corresponding to the linear energy consumption model, and combine the coefficient matrix, the constant vector, the initial constraint coefficient matrix and the initial constraint constant vector through power balance constraints to obtain the joint constraint coefficient matrix and the joint constraint constant vector; Step 1063: Extract the dual variable data based on the joint constraint coefficient matrix and the joint constraint constant vector; Step 1064: Based on the candidate adjustable power sequence corresponding to the initial constraint coefficient matrix and the initial constraint constant vector, the dual variable data, and the maximum prediction error bound of the rigid load, obtain the executability gap data of the candidate adjustable power sequence within the maximum prediction error bound of the rigid load; Step 1065: When the executability gap data is less than or equal to a preset tolerance threshold, the initial constraint coefficient matrix and the initial constraint constant vector are used as the target constraint coefficient matrix and the target constraint constant vector of the target boundary envelope; Step 1066: When the executability gap data is greater than the tolerance threshold, the candidate adjustment power sequence is used as the default adjustment power sequence and the dual variable data is used as the default dual variable data; Step 1067: Obtain the tangent plane coefficient vector and tangent plane constant term based on the default dual variable data and the default adjustment power sequence; Step 1068: Update the initial constraint coefficient matrix and the initial constraint constant vector according to the tangent plane coefficient vector and the tangent plane constant term to obtain the updated constraint coefficient matrix and the updated constraint constant vector, and use the updated constraint coefficient matrix as the initial constraint coefficient matrix and the updated constraint constant vector as the initial constraint constant vector.

[0080] In this embodiment, the rigid loads of the building (such as basic lighting, security power, etc.) inevitably have prediction deviations. This is assumed to be true throughout the entire time period. Within, the actual instantaneous prediction error sequence of rigid load is as follows: And it is known that the error is bounded, that is, it satisfies ,in Given the vector of the absolute values ​​of the maximum error fluctuation.

[0081] Substituting the actual rigid load into the aforementioned power balance constraint, the power balance constraint can be rewritten as follows: For the current iteration step Candidate feasible domain envelope (in, and The first (Constraint coefficient matrix and constant vector of the envelope in the next iteration) It is necessary to verify whether the envelope has full envelope physical executability, that is, to determine: for any valid timing adjustment trajectory within the current envelope Faced with any of the above possible instantaneous errors of rigid loads At that time, does the building always contain at least one set of constraints that satisfy the local model? Physically executable control policy vector .

[0082] To perform this verification, the rewritten power balance constraints are abstracted into a unified matrix form: Among them, the For internal decision vectors Extract the coefficient matrix of the net power of each flexible device. For the corresponding constant vector; Predict the power vector for a given baseline of the building; This is the vector of nominal predicted values ​​for rigid loads; This is a vector representing the predicted output value of new energy sources.

[0083] At this point, the problem is transformed into determining the linear system. The feasibility of this is determined by Farkas' lemma, given a trajectory. With a given error Under the premise that the system has a feasible solution The necessary and sufficient condition is: for any condition satisfying dual multiplier vectors (Corresponding to local inequality constraints) and free dual vectors (Corresponding to the global power balance equation), its dual objective function must always be greater than or equal to zero, that is: .

[0084] Given the robustness requirement of this scheme, which ensures safety under all extreme perturbations, the above inequalities must be applied to the error space. any within All are true. Due to the unknown perturbation term. Existing only on the constant side of the inequality, its adverse effects on the dual space are manifested as follows: This term. To ensure the system still has a solution even in the worst-case scenario, it is only necessary to ensure that the minimum value of the dual objective function under the worst-case perturbation is greater than or equal to zero. According to the property of the absolute value inequality, The minimum value that can be obtained is To avoid absolute values Introducing nonlinear computation to transform the unsigned equality dual multipliers It can be decomposed into the difference of two non-negative vectors, i.e. ,in , And satisfy Therefore, absolute value can be equivalently replaced by The worst-case reduction in the objective function caused by instantaneous disturbances is linearized to be equivalent to .

[0085] Furthermore, to prevent the dual multiplier ray from approaching infinity and causing the projection model to become unbounded, the multiplier space needs to be truncated and normalized. Here, a compact Farkas multiplier feasible region is defined. : In the formula, The vector is composed entirely of 1s; the last equality constraint forces all multipliers to have a norm of 1, thus ensuring that the subsequently extracted dual extremum rays are bounded. Finally, to verify the current envelope... Whether there are trajectories that could lead to internal conflict needs to be determined across the entire envelope space. and the feasible region of multipliers The worst-case scenario is sought within the dual objective function (i.e., finding the global minimum of the dual objective function). This leads to the following full-envelope verification optimization model: Solving this bilinear model, if the optimal objective value is 0, it means that even under the worst rigid load disturbance and the most extreme scheduling trajectory, there must be a feasible power allocation scheme within the building, and the current candidate envelope has absolute physical executability and robustness. If the optimal objective value is less than 0, it indicates that the range defined by the current envelope is too large (including unexecutable trajectories). In this case, it is necessary to extract the worst-case trajectory counterexample that caused the conflict from the solution results. and the corresponding extreme dual rays In numerical implementation, tolerances can be introduced. ,like The boundary construction can be considered complete if... Then iterative calculations still need to be continued.

[0086] When an unexecutable trajectory is detected within a candidate envelope, a tangent plane is generated using the extracted dual extremum ray and added to the current envelope set to remove the violation region. The new tangent plane inequality constructed based on duality theory is as follows: The mechanism of action of this tangent plane lies in the dual multiplier. The coupling effects of physical states such as temperature and energy storage across time periods are automatically quantified. The summation term on the right-hand side of the inequality... Instead of rigidly deducting a fixed power margin at each time period, the system uses multipliers as weights to automatically accumulate the errors from the time periods that have the greatest impact on the current operational bottleneck, adaptively transforming them into capacity concession margins at the envelope boundary. Furthermore, based on duality theory, this tangent plane is a necessary condition for the system to meet robust feasibility requirements. Therefore, iteratively adding this constraint does not exclude any trajectories that are feasible within the preset error bounds; it only eliminates trajectories that lead to robust infeasibility.

[0087] To facilitate iterative updates, the coefficient row vector of the tangent plane is explicitly defined. With the constant term on the right They are respectively: Extracted and Append the appended data to the bottom of the current envelope matrix and constant vector to form the constraint set for the next iteration, i.e., update it to... and Update the iteration counter Then, a full envelope robust executability check is performed again until no conflicting rays are found. Finally, Output the boundaries of smart building interaction with the power grid: To clearly illustrate the iterative segmentation and tightening process described above, the overall algorithm's logical flow is as follows: Input parameter: Baseline power of the building prediction Rigid load Disturbance boundary And the physical parameters of each device. Define the convergence tolerance. ; Constructing an internal model: Generating a coefficient matrix that includes local boundary contraction based on the matrix inequalities of the intelligent building energy consumption model. With constant vector Based on the device's physical parameters, an initial candidate envelope matrix is ​​constructed. and Initialize the iteration counter Start iteration: Solve the current iteration. The full envelope verification model (46) of the step is used to obtain the minimized objective value. ,if Then, the full envelope physical verification passes or the optimal solution leading to the default is extracted: trajectory counterexample. Dual rays with extrema Then, based on the new tangent plane inequality constructed based on duality theory, a tangent plane for removing illegal regions is generated, and the coefficient row vector of the tangent plane is used to determine the tangent plane. With the constant term on the right Extract the coefficient row vector of the new tangent plane With constant term This is then appended to the bottom of the current envelope matrix to form the constraint set for the next iteration: , and update the iteration counter. To output the final converged polyhedral matrix interface This serves as the target boundary protection for smart buildings when submitting their applications to the power grid.

[0088] The above implementation method obtains joint constraint coefficient matrix and joint constraint constant vector data representing the joint constraint relationship between the candidate boundary and the internal model by simultaneously processing the initial constraint coefficient matrix and initial constraint constant vector data with the coefficient matrix and constant vector data of the linear energy consumption model; extracts dual variables from the joint constraint data to obtain dual variable data; performs executability quantification calculation on the candidate adjustable power sequence data, dual variable data, and rigid load maximum prediction error boundary data to obtain executability gap data representing the degree of physical default under the worst-case scenario within the candidate boundary envelope; compares the executability gap data with a preset tolerance threshold, and when the executability gap data is less than or equal to the tolerance threshold, the current constraint matrix and constant vector are determined as the target constraint coefficient matrix and target constraint constant vector data. This allows the physical executability verification of the boundary envelope to be completed through deterministic quantitative indicators, rather than relying on empirical boundary margin estimation, thereby obtaining a target boundary envelope confirmed by numerical verification to have no physical default trajectory within it, and achieving precise definition of the adjustable capacity range of the building.

[0089] In this embodiment, step 107 includes: Step 1071: Determine the feasible space for the adjustment power of the target intelligent building within the power grid dispatch cycle based on the target constraint coefficient matrix and the target constraint constant vector of the target boundary envelope; Step 1072: For any time period within the power grid dispatch cycle, obtain the preset target power function corresponding to that time period; wherein, the preset target power function includes a first objective function aimed at maximizing the regulation power of that time period and a second objective function aimed at minimizing the regulation power of that time period; Step 1073: Using the feasible space for adjustable power as a constraint, obtain the maximum and minimum adjustable power of the target intelligent building during this time period according to the preset target power function; Step 1074: Arrange the maximum adjustable power and the minimum adjustable power for all time periods within the power grid dispatch cycle in chronological order to obtain the power boundary data of the target smart building within the power grid dispatch cycle.

[0090] In this embodiment, the boundary This constitutes a standard, flexible interface for smart buildings to submit applications to the power grid dispatch center or virtual power plants. When arranging day-ahead or intraday unit combinations, economic dispatch, and demand response, the dispatch center can... As a regular linear constraint, it is embedded into its optimization model. This transformation ensures that any scheduling instructions assigned to the building by the upper-level scheduler are... Both possess physical executability and robustness in the underlying control.

[0091] like Figure 2 As shown, due to limitations in visualization capabilities in high-dimensional space, this embodiment extracts the scheduling power trajectory. The three consecutive time-segment components in the figure are used to perform a three-dimensional projection comparison of the polyhedron boundary. The light-colored outer polyhedron in the figure represents the initial envelope considering only the instantaneous physical extreme values ​​of the equipment and the ramp-up in adjacent time periods. The dark polyhedrons inside represent the final boundary obtained after full envelope verification and tangent plane shrinkage using this scheme. Observations show that the final determined boundary space is significantly shrunk compared to the initial envelope. This is mainly because the final boundary fully internalizes the energy state coupling constraints of flexible devices across time periods and reserves a safety margin to resist uncertain disturbances. If the actual trajectory issued by the upper-level scheduler falls into the light-colored area but is outside the dark-colored polyhedron, the building microgrid will face the risk of exceeding physical capacity limits or energy depletion during actual execution at the lower level, resulting in the scheduling command not being fully implemented. This comparison objectively reflects the necessity and practical engineering value of deeply considering temporal coupling constraints and uncertainties to shrink the polyhedron when extracting the building interaction boundary.

[0092] Furthermore, given the difficulty in visually representing high-dimensional polyhedral sets, this scheme further includes a step of dimensionality reduction and projection of the high-dimensional envelope onto a two-dimensional time-series plane to provide operators with an intuitive reference for the maximum response depth in a single time period. Specifically, this is achieved by solving the total along the positive and negative directions of each discrete time axis. Using a sublinear programming approach, calculate the extreme values ​​of the upward and downward responses for each time period: in, and The intelligent building, after dimensionality reduction projection of the high-dimensional envelope, is located in the time period. The maximum upward and downward response power limits can be safely realized. It should be noted that the response quantities in other time periods are free variables and do not need to be fixed. Instead, they are automatically optimized under the premise of satisfying the overall constraint Ar<=b.

[0093] The two-dimensional time-series boundary curves obtained through projection calculations intuitively reflect the single-point limit regulation capacity range of intelligent buildings when independently examining each time period. This visualized projection result can be directly used for capacity declaration in the electricity market, ancillary service pricing assessment, and to provide operators with an intuitive benchmark for monitoring operating conditions.

[0094] The above implementation method performs feasible space determination processing on the target constraint coefficient matrix and target constraint constant vector data to obtain feasible space data of adjustable power representing the verified adjustable capacity range of the building; for each time period, a first objective function aiming to maximize the adjustable power and a second objective function aiming to minimize the adjustable power are obtained, and the first and second objective functions are optimized and solved respectively with the feasible space of adjustable power as constraints to obtain first optimal target value data representing the maximum power increase capability and second optimal target value data representing the maximum power reduction capability of the time period; The first and second optimal target values ​​for all time periods are processed sequentially to obtain power boundary data that characterizes the adjustable power range of the building in each time period within the scheduling cycle. This ensures that the final output power boundary values ​​for each time period are extracted by projecting from a physically executable high-dimensional temporal coupling feasible space. The boundaries of each time period maintain a constraint relationship consistent with the cross-time coupling characteristics such as the continuity of energy storage and thermal inertia within the building. This avoids the contradictions between adjacent time period boundaries caused by independently calculating the boundaries of each time period and severing the temporal coupling in existing technologies, thus achieving the accuracy and self-consistency of the power boundary data in the temporal dimension.

[0095] On the other hand, refer to Figure 3 This embodiment also discloses a system for determining the adjustment boundary of a building's participation in power grid dispatch, including an operation acquisition module 301, a multi-constraint module 302, an energy consumption assessment module 303, a dispatch constraint module 304, a boundary candidate module 305, a boundary update module 306, and a boundary generation module 307. The operation acquisition module 301 is used to acquire the physical parameters of the target intelligent building and the operation prediction data and operation error data of the target intelligent building during the power grid dispatch cycle; The multi-constraint module 302 is used to obtain, based on the physical parameters, the operational prediction data and the operational error data, the equipment operation constraints, power balance constraints, equipment constraints and energy constraints of the target intelligent building within the power grid dispatch cycle; The energy consumption assessment module 303 is used to obtain the linear energy consumption model corresponding to the target intelligent building based on the equipment operation constraints, the power balance constraints, the equipment constraints, and the energy constraints. The scheduling constraint module 304 is used to obtain the adjustment power boundary data and amplitude constraints of each flexible device in the target intelligent building during the power grid scheduling cycle based on the physical parameters. The boundary candidate module 305 is used to obtain the initial boundary envelope of the target intelligent building based on the adjustment power boundary data and the amplitude constraint. The boundary update module 306 is used to iteratively verify and cut the initial boundary envelope according to the linear energy consumption model to obtain the target boundary envelope corresponding to the target smart building. The boundary generation module 307 is used to obtain the power boundary data of the target intelligent building within the power grid dispatch cycle based on the target boundary envelope and the preset target power function of the target intelligent building.

[0096] This embodiment discloses a method and system for determining the adjustment boundary of a building's participation in power grid dispatch. By acquiring physical parameters, operational prediction data, and operational error data, a continuous linear energy consumption model characterized by duty cycle constraints is constructed. Then, an initial boundary envelope considering only static physical limits is constructed. Subsequently, the initial boundary envelope is iteratively verified and tangent planes are cut. Finally, the verified target boundary envelope is dimensionally reduced and projected into adjustable power boundary data for each time period. This method offers the following beneficial technical effects: First, it improves the accuracy of constructing cross-time-period adjustment boundaries and avoids numerical ill-conditioning in model solutions. This application's embodiments, targeting the heating and cooling power of HVAC systems, the charging and discharging power of energy storage and electric vehicles, do not introduce binary variables. Instead, they use duty cycle constraints to characterize their true physical adjustment capabilities, avoiding the "fragmented" state of the feasible region caused by the introduction of binary variables in traditional methods. In step 103, all equality constraints are decomposed into inequality constraints and uniformly integrated into a linear energy consumption model in matrix inequality form. Farkas' lemma is used to transform the feasibility judgment of the internal state into a dual optimization problem and perform iterative verification and segmentation. Because the system maintains a strict continuous linear mathematical structure throughout the global boundary optimization process, it avoids the numerical instability problems caused by the introduction of binary variables and large M parameters in the traditional Big M method, while fully preserving the physical characteristics of heterogeneous devices such as HVAC, energy storage, and electric vehicles, thus improving the accuracy and stability of boundary calculations.

[0097] Second, it improves the physical executability of interactive boundaries under multi-source uncertainty environments. This application's embodiments address local disturbances affecting the cumulative state of equipment, such as outdoor temperature errors. By analytically calculating the worst-case deviation of the cumulative temperature effect term and tightening this deviation inversely to the upper and lower bounds of the room temperature comfort constraint, local analytical contraction is achieved. For instantaneous power fluctuations such as rigid loads, instead of blindly deducting fixed margins, the maximum prediction error boundary of rigid loads is left to the Farkas dual verification stage. The dual multiplier automatically quantifies the cross-period coupling effects of physical states such as temperature and energy storage, and uses the extracted worst-case dual extreme ray to generate a robust tangent plane for adaptive global time-series margin concession. This mechanism uses the dual multiplier as the weight for quantifying cross-period physical coupling, reasonably internalizing the impact of extreme continuous error accumulation on system operating bottlenecks. This ensures that the final output dynamic power envelope can effectively avoid potential physical hard constraint conflicts within the preset disturbance range, providing a more reliable underlying physical fulfillment guarantee for the dispatch commands issued by the power grid.

[0098] Third, it can accurately characterize the adjustable power boundary of the building and eliminate the physical default of dispatch instructions. This application constructs an initial boundary envelope that only considers the physical power limits of each flexible device. Then, through iterative verification and segmentation, Farkas' lemma is used to verify whether there are infeasible trajectories within the current envelope due to temporal coupling constraints and uncertainties. When the verification finds an unexecutable dispatch trajectory within the envelope boundary, the defaulted adjustable power sequence and defaulted dual variable data are extracted from the solution results, and a tangent plane is generated to shrink the envelope until there are no physically defaulted trajectories within the envelope. For the target boundary envelope that passes the iterative verification, it is dimensionality-reduced and projected into the maximum and minimum adjustable power for each time period by solving a linear programming problem. This power boundary data can be directly used for capacity declaration and ancillary service pricing assessment in the electricity market. Because the target boundary envelope fully internalizes the energy state coupling constraints of flexible devices across time periods and reserves a safety margin to resist uncertain disturbances, any dispatch instruction issued by the power grid within this boundary can be physically implemented within the building, thereby eliminating the physical default of dispatch instructions caused by ignoring temporal coupling and uncertainties.

[0099] Fourth, it effectively avoids the generation of false regulation capabilities. This application's embodiments target HVAC systems, introducing duty cycle constraints to allow the building as a whole to exhibit simultaneous cooling and heating macroscopic characteristics while restricting devices from simultaneously outputting high power for both cooling and heating. Similarly, for electric energy storage systems and electric vehicle clusters, duty cycle constraints are introduced to limit the mixed operation of their charging and discharging power. Mathematically, this constraint is mathematically simplified to an inequality constraint between a power upper limit constraint and a proportional sum not exceeding 1. While preserving the model's full continuity and linearity, it effectively prevents "false regulation capabilities" arising from ignoring the actual operating logic of the equipment, resulting in simultaneous large-scale heating and cooling, or simultaneous large-scale charging and discharging. This ensures that the final power boundary data accurately reflects the building's actual physical regulation capabilities.

[0100] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for determining the regulation boundary of a building's participation in power grid dispatch, characterized in that, include: Obtain the physical parameters of the target intelligent building and the predicted operation data and operation error data of the target intelligent building during the power grid dispatch cycle; Based on the physical parameters, the operational prediction data, and the operational error data, the equipment operation constraints, power balance constraints, equipment constraints, and energy constraints of the target intelligent building within the power grid dispatch cycle are obtained. Based on the equipment operation constraints, the power balance constraints, the equipment constraints, and the energy constraints, obtain the linear energy consumption model corresponding to the target intelligent building; Based on the physical parameters, obtain the adjustment power boundary data and amplitude constraints of each flexible device in the target intelligent building during the power grid dispatch cycle; The initial boundary envelope of the target intelligent building is obtained based on the adjusted power boundary data and the amplitude constraint; The initial boundary envelope is iteratively verified and cut according to the linear energy consumption model to obtain the target boundary envelope corresponding to the target intelligent building. Based on the target boundary envelope and the preset target power function of the target intelligent building, the power boundary data of the target intelligent building within the power grid dispatch cycle is obtained.

2. The method for determining the adjustment boundary of a building's participation in power grid dispatch according to claim 1, characterized in that, The acquisition of the physical parameters of the target intelligent building and the operational prediction data and operational error data of the target intelligent building during the power grid dispatch cycle includes: The building parameters, historical operating data, and equipment parameters of each flexible device in the target intelligent building are obtained to obtain the physical parameters of the target intelligent building based on the building parameters and the equipment parameters; wherein, the historical operating data includes historical electricity consumption time series data, historical uncontrollable load time series data, a first error sequence between historical outdoor temperature prediction values ​​and historical outdoor temperature measurement values, and a second error sequence between historical rigid load prediction values ​​and historical rigid load measurement values; Based on the historical electricity consumption time series data, the baseline power prediction sequence of the target smart building within the power grid dispatch cycle is obtained through a preset electricity consumption prediction model. Based on the historical uncontrollable load time series data, a rigid load prediction sequence for the target intelligent building within the power grid dispatch cycle is obtained through a preset load prediction model. The maximum absolute value of each error in the first error sequence is taken as the maximum prediction error bound of outdoor temperature, and the maximum absolute value of each error in the second error sequence is taken as the maximum prediction error bound of rigid load. The maximum prediction error bound of outdoor temperature and the maximum prediction error bound of rigid load are used as the operating error data. The outdoor temperature prediction sequence of the target smart building during the power grid dispatch cycle is obtained, and the baseline power prediction sequence, the rigid load prediction sequence, and the outdoor temperature prediction sequence are used as the operation prediction data of the target smart building during the power grid dispatch cycle.

3. The method for determining the adjustment boundary of a building's participation in power grid dispatch according to claim 2, characterized in that, The building parameters include the building's equivalent heat capacity and equivalent thermal resistance; the equipment parameters include the heating efficiency ratio, cooling efficiency ratio, maximum heating power and maximum cooling power of each HVAC system, the maximum charging and discharging power, charging and discharging efficiency and ultimate energy capacity of each electric energy storage system, the ultimate charging and discharging power of each electric vehicle cluster, and the maximum operating power and ultimate total energy demand of each transferable load. The step of obtaining the equipment operation constraints, power balance constraints, equipment constraints, and energy constraints of the target intelligent building within the power grid dispatch cycle based on the physical parameters, the operation prediction data, and the operation error data includes: Based on the building's equivalent heat capacity, the building's equivalent thermal resistance, the heating energy efficiency ratio, the cooling energy efficiency ratio, the outdoor temperature prediction sequence, and the maximum prediction error boundary of the outdoor temperature, the equipment operation constraints of the target smart building within the power grid dispatch cycle are obtained. The predicted power generation sequence of each distributed power source in the target smart building during the grid dispatch cycle is obtained, and the power balance constraint of the target smart building during the grid dispatch cycle is obtained based on the predicted power generation sequence, the baseline power prediction sequence and the rigid load prediction sequence. Based on the maximum heating power, the maximum cooling power, the maximum charging and discharging power, and the ultimate charging and discharging power, the equipment constraints of the target intelligent building within the power grid dispatch cycle are obtained; Based on the charging and discharging efficiency, the maximum energy capacity, the maximum operating power, and the maximum total energy demand, the energy constraints of the target smart building within the power grid dispatch cycle are obtained.

4. The method for determining the adjustment boundary of a building's participation in power grid dispatch according to claim 1, characterized in that, The step of obtaining the linear energy consumption model corresponding to the target intelligent building based on the equipment operation constraints, the power balance constraints, the equipment constraints, and the energy constraints includes: The equipment operation constraints, the power balance constraints, the equipment constraints, and the energy constraints are divided into a set of linear equality constraints and a set of linear inequality constraints. For any constraint in the set of linear equality constraints, the constraint is split into a pair of linear inequality constraints with opposite directions. Based on all the linear inequality constraints corresponding to the linear equality constraint set and the linear inequality constraint set, obtain the inequality constraint set of the target intelligent building within the power grid dispatch cycle; Obtain the decision variables, constant terms, and variable coefficients of each inequality constraint in the inequality constraint set. Concatenate all the decision variables corresponding to the inequality constraint set into a decision variable column vector over the power grid dispatching cycle; The matrix dimension is determined by the number of inequality constraints in the inequality constraint set and the dimension of the column vector of the decision variables. The matrix elements are obtained based on the set of inequality constraints, the column vector of decision variables, and the coefficients of the variables; Obtain the constant term in each inequality constraint in the inequality constraint set, and obtain the linear energy consumption model of the target smart building in the power grid dispatch cycle based on the constant term, the matrix dimension, the matrix elements and the decision variable column vector.

5. The method for determining the adjustment boundary of a building's participation in power grid dispatch according to claim 3, characterized in that, The equipment parameters include the maximum physical ramp rate for each flexible device. The step of obtaining the adjustment power boundary data and amplitude constraints of each flexible device in the target intelligent building within the power grid dispatch cycle based on the physical parameters includes: For any time period within the power grid dispatch cycle, the maximum heating power, the maximum charging power among the maximum charging and discharging power, and the maximum operating power are summed to obtain the upper bound of the regulation power corresponding to that time period; The lower bound of the adjustment power corresponding to this time period is obtained by summing the maximum cooling power, the maximum discharge power in the maximum charge and discharge power and the ultimate discharge power in the ultimate charge and discharge power and taking the amplitude. Based on the upper bound and the lower bound of the regulation power, the regulation power boundary data corresponding to each time period within the power grid dispatch cycle are obtained. The ramp rate value of the target intelligent building is obtained by summing all the extreme physical ramp rates corresponding to the target intelligent building, and the ramp rate value is used as the amplitude constraint between any pair of adjacent time periods within the power grid dispatch cycle.

6. The method for determining the adjustment boundary of a building's participation in power grid dispatch according to claim 5, characterized in that, The step of obtaining the initial boundary envelope of the target intelligent building based on the adjusted power boundary data and the amplitude constraint includes: Based on the adjustment power boundary data corresponding to each time period, the hourly power boundary inequality of the target smart building in the power grid dispatch cycle is obtained. Based on the amplitude constraints between any pair of adjacent time periods, the ramp constraint inequality of the smart building within the power grid dispatch cycle is obtained; Based on the adjusted power sequence of the target intelligent building in the power grid dispatch cycle, all the hourly power boundary inequalities, and all the ramp constraint inequalities, the initial boundary envelope of the target intelligent building in the power grid dispatch cycle is obtained.

7. A method for determining the adjustment boundary of a building's participation in power grid dispatch according to any one of claims 1-6, characterized in that, The step of iteratively verifying and cutting the initial boundary envelope according to the linear energy consumption model to obtain the target boundary envelope corresponding to the target smart building includes: Obtain the initial constraint coefficient matrix and initial constraint constant vector corresponding to the initial boundary envelope; Obtain the coefficient matrix and constant vector corresponding to the linear energy consumption model, and combine the coefficient matrix, the constant vector, the initial constraint coefficient matrix and the initial constraint constant vector through power balance constraints to obtain the joint constraint coefficient matrix and the joint constraint constant vector. Extract dual variable data based on the joint constraint coefficient matrix and the joint constraint constant vector; Based on the candidate adjustable power sequence corresponding to the initial constraint coefficient matrix and the initial constraint constant vector, the dual variable data, and the maximum prediction error bound of the rigid load, obtain the executability gap data of the candidate adjustable power sequence within the maximum prediction error bound of the rigid load; When the executability gap data is less than or equal to a preset tolerance threshold, the initial constraint coefficient matrix and the initial constraint constant vector are used as the target constraint coefficient matrix and the target constraint constant vector of the target boundary envelope.

8. The method for determining the adjustment boundary of a building's participation in power grid dispatch according to claim 7, characterized in that, The step of iteratively verifying and cutting the initial boundary envelope according to the linear energy consumption model to obtain the target boundary envelope corresponding to the target intelligent building further includes: When the executability gap data is greater than the tolerance threshold, the candidate adjustment power sequence is used as the default adjustment power sequence and the dual variable data is used as the default dual variable data. Based on the default dual variable data and the default adjustment power sequence, obtain the tangent plane coefficient vector and the tangent plane constant term; Based on the tangent plane coefficient vector and the tangent plane constant term, the initial constraint coefficient matrix and the initial constraint constant vector are updated to obtain the updated constraint coefficient matrix and the updated constraint constant vector. The updated constraint coefficient matrix is ​​then used as the initial constraint coefficient matrix and the updated constraint constant vector is used as the initial constraint constant vector.

9. A method for determining the adjustment boundary of a building's participation in power grid dispatch according to claim 7, characterized in that, The step of obtaining the power boundary data of the target intelligent building within the power grid dispatch cycle based on the target boundary envelope and the preset target power function of the target intelligent building includes: Based on the target constraint coefficient matrix and target constraint constant vector of the target boundary envelope, the feasible space for the adjustment power of the target intelligent building within the power grid dispatch cycle is determined; For any time period within the power grid dispatch cycle, a preset target power function corresponding to that time period is obtained; wherein, the preset target power function includes a first objective function aimed at maximizing the regulation power of that time period and a second objective function aimed at minimizing the regulation power of that time period; Using the feasible space for adjustable power as a constraint, the maximum and minimum adjustable power of the target intelligent building during this time period are obtained according to the preset target power function; The maximum adjustable power and the minimum adjustable power for all time periods within the power grid dispatch cycle are arranged in chronological order to obtain the power boundary data of the target intelligent building within the power grid dispatch cycle.

10. A system for determining the regulation boundary of a building's participation in power grid dispatch, characterized in that, It includes a data acquisition module, a multi-constraint module, an energy consumption assessment module, a scheduling constraint module, a boundary candidate module, a boundary update module, and a boundary generation module; The operation acquisition module is used to acquire the physical parameters of the target intelligent building and the operation prediction data and operation error data of the target intelligent building during the power grid dispatch cycle; The multi-constraint module is used to obtain the equipment operation constraints, power balance constraints, equipment constraints and energy constraints of the target intelligent building within the power grid dispatch cycle based on the physical parameters, the operation prediction data and the operation error data. The energy consumption assessment module is used to obtain the linear energy consumption model corresponding to the target intelligent building based on the equipment operation constraints, the power balance constraints, the equipment constraints, and the energy constraints. The scheduling constraint module is used to obtain the adjustment power boundary data and amplitude constraints of each flexible device in the target intelligent building during the power grid scheduling cycle based on the physical parameters. The boundary candidate module is used to obtain the initial boundary envelope of the target intelligent building based on the adjusted power boundary data and the amplitude constraint. The boundary update module is used to iteratively verify and cut the initial boundary envelope according to the linear energy consumption model to obtain the target boundary envelope corresponding to the target smart building. The boundary generation module is used to obtain the power boundary data of the target intelligent building within the power grid dispatch cycle based on the target boundary envelope and the preset target power function of the target intelligent building.

Images