Load control methods, systems, equipment, and media based on building thermal dynamics characteristics

By using a generalized additive model and a joint optimization model, the state of building inertial thermal energy storage is accurately estimated, which solves the problem of state-of-charge deviation in virtual energy storage regulation, realizes efficient utilization of building thermal energy storage and grid optimization, and reduces system costs.

CN122905376BActive Publication Date: 2026-11-17ELECTRIC POWER SCI & RES INST OF STATE GRID TIANJIN ELECTRIC POWER CO +2
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202611399027.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-09-10
Publication Date
2026-11-17
Estimated Expiration
2046-09-10

AI Technical Summary

Technical Problem

Existing virtual energy storage control methods suffer from biased state of charge estimation due to the nonlinearity of building thermal processes and the coupling characteristics of multiple factors, which affects scheduling flexibility and the efficiency of thermal energy storage utilization.

Method used

A generalized additive model is used to establish a nonlinear regression relationship for the state of charge of building inertial thermal energy storage. Combined with user thermal comfort constraints, the flexibility power boundary is calculated, and a joint optimization model of the power network and the district heating network is constructed for power allocation.

Benefits of technology

It improves the accuracy of state of charge estimation, overcomes the problem of state perception lag, realizes the refined perception and collaborative optimization of building virtual energy storage resources, reduces system operating costs, and alleviates the problems of reverse power flow and voltage over-limit in the distribution network.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122905376B_ABST
    Figure CN122905376B_ABST
Patent Text Reader

Abstract

This invention discloses a load regulation method, system, equipment, and medium based on building thermal dynamics, relating to the technical field of new energy power systems. The method collects historical building operation data and uses a generalized additive model to establish a nonlinear regression relationship between external predictive variables and the building's inertial thermal energy storage state of charge (SOC), thereby achieving perception of the building's thermal state. The SOC values ​​are input into the building's thermal dynamics model, and combined with user thermal comfort constraints, the upward and downward flexibility power boundaries currently available to the building are dynamically calculated. This aggregates the flexibility resources of multiple buildings to form a cluster flexibility curve, assisting in the construction of a joint optimization model including the power network and district heating network. With the minimum operating cost as the objective function and the building cluster flexibility curve as the constraint, the power allocation scheme for future periods is solved, and dispatch instructions are issued. This method achieves refined perception and collaborative optimization utilization of building virtual energy storage resources.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention generally relates to the technical field of new energy power systems, and specifically to load regulation methods, systems, equipment and media based on building thermal dynamic characteristics. Background Technology

[0002] With the large-scale integration of distributed photovoltaic (PV) power into the distribution network, reverse power flow and voltage over-limit issues are becoming increasingly prominent. Utilizing building thermal inertia as "virtual energy storage" for load regulation is an important way to alleviate these problems.

[0003] Existing virtual energy storage control methods mainly rely on low-order equivalent thermal parameter models to describe building thermal dynamics. However, building thermal processes are influenced by multiple factors such as outdoor temperature, solar radiation, and occupant activity, exhibiting strong nonlinear characteristics. Physical models with fixed parameters struggle to accurately depict the real thermal response under complex and variable conditions. This results in a significant discrepancy between the estimated State of Energy (SoE) of virtual energy storage and the actual heat storage capacity. Consequently, the dispatchable flexibility boundary calculated based on this state is inaccurate, leading to optimized control schemes either exceeding user thermal comfort limits or being overly conservative and failing to fully utilize the building's thermal energy storage potential.

[0004] Therefore, there is an urgent need for a control method that can sense the real-time status of building virtual energy storage and deeply integrate it with multi-energy flow collaborative optimization. Summary of the Invention

[0005] In view of the above-mentioned defects or deficiencies in the existing technology, it is desirable to provide a load regulation method, system, equipment and medium based on building thermal dynamic characteristics, so as to realize the accurate perception and coordinated optimization utilization of building virtual energy storage resources.

[0006] In a first aspect, the present invention provides a load regulation method based on building thermal dynamic characteristics, comprising: Collect historical operational data for all target buildings. The historical operational data includes at least outdoor temperature, solar irradiance, time-of-day category variables, and week-of-day category variables. A nonlinear regression relationship between the historical operating data and the building's inertial thermal energy storage state of charge is established using a generalized additive model, and the current building's inertial thermal energy storage state of charge is output based on the nonlinear regression relationship. The current building inertial thermal energy storage charge state is input into the preset building thermal dynamics model, and combined with user thermal comfort constraints, the upward flexibility power boundary and downward flexibility power boundary of the target building in the current state are calculated. Based on the upward and downward flexibility power boundaries of each target building, the flexibility curve of the building cluster is obtained; A joint optimization model integrating the power grid and the district heating network is constructed. The objective function is to minimize the operating cost, and the building cluster flexibility curve is used as the constraint. The model combines the photovoltaic power output forecast, the grid time-of-use electricity price, and the heat load demand to solve for the power allocation data in the future forecast time. Based on the power allocation data, the model issues scheduling instructions to the power-to-heat equipment of each target building.

[0007] According to the technical solution provided by the present invention, establishing a nonlinear regression relationship between the historical operating data and the building's inertial thermal energy storage state of charge using a generalized additive model includes: A mapping architecture is established based on the building's inertial thermal energy storage state of charge and historical operating data. An interaction term function is introduced into the mapping architecture to capture the coupling influence between various predictive variables on the building's inertial thermal energy storage state of charge under combined operating conditions. Based on the mapping architecture, spline smoothing functions are used to perform nonlinear fitting on the outdoor temperature and the solar irradiance to generate smoothing function terms corresponding to the outdoor temperature and the solar irradiance, respectively; and class functions are used to encode the time category variable and the week category variable to generate class function terms corresponding to the time category variable and the week category variable, respectively. The smoothing function, category function, and interaction function are linearly summed to construct a building inertial thermal energy storage model, which characterizes the nonlinear regression relationship between the historical operating data and the building inertial thermal energy storage state of charge.

[0008] According to the technical solution provided by the present invention, the training process of the building inertial thermal energy storage model includes: Historical monitoring data from temperature sensors and smart meters of target buildings were collected from multiple meteorological years and used as training samples for the building's inertial thermal energy storage model. Based on the training samples of the building inertial thermal energy storage model, the generalized cross-validation criterion is adopted to determine the form and corresponding smoothing parameters of the smoothing function term, category function term, and interaction function term in the building inertial thermal energy storage model by minimizing the residual sum of squares, as well as the physical parameters to be identified in the building inertial thermal energy storage model.

[0009] According to the technical solution provided by the present invention, the step of inputting the current building inertial thermal energy storage charge state into a preset building thermal dynamic model, and calculating the upward flexibility power boundary and downward flexibility power boundary of the target building in the current state in combination with user thermal comfort constraints, specifically includes: A building thermal dynamic state space model is constructed, and user thermal comfort constraints are set; wherein, the building thermal dynamic state space model is used to characterize the trajectory of indoor air temperature change under different thermal power inputs; Based on the current state of charge of the building's inertial thermal energy storage, the maximum thermal capacity of the building's inertial thermal energy storage, and the midpoint temperature of the user's thermal comfort constraints, the reference indoor temperature at the current moment is calculated. Based on the state of charge of the building's inertial thermal energy storage and the range of thermal capacity, the remaining space that can be charged and the remaining space that can be released at the current moment are calculated respectively. The upward flexibility power boundary is obtained based on the remaining heat dissipation space, the current indoor air temperature change trajectory, and the termination boundary corresponding to the user thermal comfort constraint. The downward flexibility power boundary is obtained based on the remaining rechargeable heatable space, the current indoor air temperature change trajectory, and the termination boundary corresponding to the user thermal comfort constraint.

[0010] According to the technical solution provided by the present invention, the target building is connected to an electric heat transfer device; the upward flexibility power boundary is determined in the following manner: Based on the current power consumption of the electric heat exchanger, the input heat power of the electric heat exchanger is gradually reduced. The indoor air temperature change trajectory is used until the indoor air temperature reaches the first preset condition at any predicted time. The cumulative power consumption reduction of the electric heat exchanger is then used as the upward flexibility power boundary. The downward flexibility power boundary is determined in the following manner: Based on the current power consumption of the electric heat exchanger, the power consumption of the electric heat exchanger is gradually increased. Using the indoor air temperature change trajectory, until the indoor air temperature reaches the second preset condition at any predicted time, the cumulative increase in power consumption of the electric heat exchanger is used as the downward flexibility power boundary. The first and second preset conditions are both set based on the rated power of the electric heat transfer equipment, the termination boundary of the user's thermal comfort constraint, and the remaining heat dissipation space.

[0011] According to the technical solution provided by the present invention, a building cluster flexibility curve is obtained based on the upward flexibility power boundary and downward flexibility power boundary of each of the target buildings, including: Based on the upward flexibility power boundary and downward flexibility power boundary of each target building, the upward flexibility power boundary of each target building at the same time is recorded to obtain the cluster upward flexibility power boundary at that time; the downward flexibility power boundary of each target building at the same time is recorded to obtain the cluster downward flexibility power boundary at that time. By iterating through each time point within multiple prediction steps, the upward flexibility power boundary and the downward flexibility power boundary of the cluster corresponding to each time point are obtained. Based on the upward flexibility power boundary and the downward flexibility power boundary of the cluster corresponding to each time point, the building cluster flexibility curve is formed.

[0012] According to the technical solution provided by this invention, a joint optimization model comprising a power grid and a district heating network is constructed. The objective function is to minimize operating costs, and the building cluster flexibility curve is used as a constraint. Combining photovoltaic power generation forecasts, grid time-of-use pricing, and heat load demand, the model solves for power allocation data within the predicted future timeframe. Based on this power allocation data, scheduling instructions are issued to the power-to-heat conversion equipment of each target building, including: The building thermal dynamic state space model and the nonlinear regression relationship are embedded into a multi-energy flow collaborative optimization framework at the park or power distribution network level, and the joint optimization model is constructed by adopting a rolling time-domain optimization mechanism of model predictive control. The photovoltaic power output forecast, grid time-of-use electricity price, and heat load demand are used as the input boundary conditions of the joint optimization model. The objective function is to minimize the operating cost. The joint optimization model is solved under constraints to obtain the power allocation sequence for the future forecast period. The constraints include at least the power adjustment range constraint defined by the building cluster flexibility curve. Based on the power allocation sequence, scheduling instructions are issued to the power-to-heat conversion equipment of each target building according to the power allocation data corresponding to the current time.

[0013] Secondly, the present invention provides a load control system based on building thermal dynamic characteristics, comprising: The data acquisition module is used to collect historical operational data of all target buildings. The historical operational data includes at least outdoor temperature, solar irradiance, time category variables, and week category variables. The charge state value output module is used to establish a nonlinear regression relationship between the historical operating data and the building inertial thermal energy storage charge state using a generalized additive model, and output the building inertial thermal energy storage charge state at the current moment based on the nonlinear regression relationship. The power boundary output module is used to input the current building inertial thermal energy storage charge state into the preset building thermal dynamics model, and calculate the upward flexibility power boundary and downward flexibility power boundary of the target building in the current state in combination with user thermal comfort constraints; and to obtain the building cluster flexibility curve based on the upward flexibility power boundary and downward flexibility power boundary of each target building. The dispatch instruction processing module is used to construct a joint optimization model that includes the power network and the district heating network. The objective function is to minimize the operating cost, and the building cluster flexibility curve is used as the constraint. The module combines the photovoltaic power output forecast, the grid time-of-use electricity price, and the heat load demand to solve for the power allocation data in the future forecast time. Based on the power allocation data, the module issues dispatch instructions to the power-to-heat equipment of each target building.

[0014] Thirdly, the present invention provides a terminal device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described above.

[0015] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method described above.

[0016] In summary, this technical solution specifically discloses a load control method, system, equipment, and medium based on building thermal dynamic characteristics. The method includes: collecting historical operating data of all target buildings, whereby the historical operating data includes at least outdoor temperature, solar irradiance, time-of-day category variables, and week-of-day category variables; establishing a nonlinear regression relationship between the historical operating data and the building's inertial thermal energy storage state of charge using a generalized additive model, and outputting the current building inertial thermal energy storage state of charge based on the nonlinear regression relationship; inputting the current building inertial thermal energy storage state of charge into a preset building thermal dynamic model, and combining... Considering user thermal comfort constraints, the upward and downward flexibility power boundaries of the target buildings under the current state are calculated. Based on the upward and downward flexibility power boundaries of each target building, the flexibility curve of the building cluster is obtained. A joint optimization model including the power network and the district heating network is constructed, with the minimum operating cost as the objective function and the flexibility curve of the building cluster as the constraint. Combining the photovoltaic power output forecast, the grid time-of-use electricity price, and the heat load demand, the power allocation data for the future forecast period is solved, and the dispatch instructions are issued to the power-to-heat equipment of each target building based on the power allocation data.

[0017] Beneficial Effects: This invention utilizes the nonlinear fitting capability and interaction terms of the generalized additive model to capture the coupling effect of complex operating conditions, significantly improving the estimation accuracy of the state of charge (SOC) of building inertial thermal energy storage compared to traditional fixed RC parameter models. By inputting real-time SOC values ​​into the building's thermal dynamics model and dynamically calculating the flexibility power boundary, it overcomes the problem of misjudgment of adjustment potential caused by the lag in state perception in traditional fixed boundary methods. Furthermore, the aggregated building cluster flexibility curves are embedded as key constraints into the joint optimization model of the power network and the district heating network. The optimal power allocation scheme is solved using a rolling time-domain mechanism of model predictive control, effectively mitigating renewable energy fluctuations, reducing system operating costs, and alleviating the problems of reverse power flow and voltage over-limit in the distribution network. This technical solution organically integrates data-driven state estimation, physical model constraints, and model predictive control, realizing refined perception and collaborative optimization utilization of building virtual energy storage resources. Attached Figure Description

[0018] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a schematic diagram of a load control method based on the thermal dynamic characteristics of a building.

[0019] Figure 2 This is a schematic diagram illustrating the step-by-step process of step S200 in a load control method based on building thermal dynamics.

[0020] Figure 3 This is a schematic diagram illustrating the step S300 of a load control method based on building thermal dynamics.

[0021] Figure 4 This is a schematic diagram illustrating the step-by-step process of step S500 in a load control method based on building thermal dynamics.

[0022] Figure 5 This is a schematic diagram of a load control system based on the thermal dynamic characteristics of a building.

[0023] Figure 6 This is a schematic diagram of a terminal device. Detailed Implementation

[0024] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0025] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0026] Example 1 To make the technical solutions of the embodiments of the present invention clearer and easier to understand, the application background of the embodiments of the present invention will be introduced below.

[0027] With the accelerated global energy transition, the penetration rate of renewable energy sources, such as wind and solar power, in the power system continues to increase. However, the inherent volatility and intermittency of renewable energy pose a severe challenge to the supply and demand balance of the power grid, especially at the distribution level, where large-scale integration of distributed photovoltaic power leads to serious reverse power flow and voltage over-limit issues during certain periods. To mitigate volatility and improve grid integration, traditional solutions mainly rely on physical energy storage devices such as battery energy storage (BES) or hot water tanks (HWT). However, these hardware solutions not only face high initial investment costs and operational burdens, but are also often constrained by space and security limitations in densely populated urban areas, making large-scale deployment difficult.

[0028] Against this backdrop, buildings, as the largest energy consumption terminals in cities, are considered a highly promising "virtual energy storage" (VES) resource due to their enormous thermal inertia. Existing technologies typically utilize the thermal capacity characteristics of the building envelope, combined with power-to-heat (P2H) devices, to convert excess electrical energy into heat energy temporarily stored within the building structure. This sacrifices a slight improvement in indoor temperature comfort in exchange for improved load regulation capabilities on the grid side. Currently, mainstream virtual energy storage modeling methods mostly employ low-order equivalent thermal parameter (ETP) models based on physical mechanisms, such as the classic 3-resistance-2-capacitance (3R2C) model, to describe the thermodynamic dynamic processes between the air and walls inside the building. While these methods possess a degree of physical interpretability, they exhibit significant limitations when faced with complex and ever-changing external environments. On one hand, building thermal processes are inherently highly nonlinear and involve multiple factors. External factors such as direct / scattered solar radiation, multi-directional facade effects, and thermal disturbances from occupants are often difficult to accurately characterize using linear differential equations. On the other hand, fixed resistance-capacitance (RC) parameters cannot adapt to the dynamic characteristics of buildings under different seasons and operating modes. Consequently, the estimated inertial thermal energy storage charge state based on such models will deviate from the actual stored heat capacity, thus limiting the precise quantification of dispatchability flexibility.

[0029] Meanwhile, data-driven modeling studies on building thermal characteristics have shown that the Generalized Additive Model (GAM) has unique advantages in handling nonlinear external factors. Compared to traditional multiple linear regression, GAM can flexibly fit the complex relationships between outdoor temperature, irradiance, and time-specific variables (hours, days of the week) and building thermal state using smoothing functions such as logarithms and splines. It can also achieve high-precision SoE estimation using easily measurable parameters such as ceiling surface temperature as proxy variables. However, existing GAM applications are mostly limited to load forecasting or simple condition monitoring of individual buildings, and a systematic framework for embedding it into multi-building aggregation and combining it with distribution network-level optimized scheduling has not yet been formed.

[0030] Therefore, the main problem with existing technologies lies in the lack of a method that can organically combine high-precision data-driven state estimation with multi-energy flow collaborative optimization. Traditional virtual energy storage control often leads to "over-adjustment" or "under-adjustment" due to inaccurate state perception—that is, it may exceed the user's thermal comfort boundary when pursuing the absorption of renewable energy, or fail to fully tap the building's thermal energy storage potential under conservative strategies. How to integrate the nonlinear state estimation capability of GAM with the VES multi-energy system architecture to build a closed-loop control system that can both perceive complex environmental disturbances and participate in grid optimization is a technical challenge that urgently needs to be solved.

[0031] In view of this, please refer to Figure 1 The flowchart shown in this embodiment illustrates a load control method based on building thermal dynamics. The executing entity of this embodiment can be a regional integrated energy system management platform or a building cluster collaborative optimization controller. A practical application scenario could be a park with multiple buildings, each equipped with an air-source heat pump as an electric heat conversion device, and also with indoor temperature sensors, ceiling surface temperature sensors, and smart meters, etc. No specific limitations are imposed here. The following is a further explanation of each step of this invention, which includes the following steps: S100. Collect historical operational data for all target buildings. The historical operational data should include at least outdoor temperature, solar irradiance, time-of-day category variables, and week-of-day category variables. Specifically, variables that have a significant impact on building thermal dynamics and are easily obtained through sensors or weather stations are selected as predictors. These variables include both continuous numerical variables and discrete categorical variables; among them, continuous numerical variables include: outdoor temperature at any time , Diffusion level radiation at time as well as Direct normal radiation at time These variables directly determine the heat transfer boundary conditions of the building envelope and the solar radiation intensity (solar radiation intensity is determined by the aforementioned scattered radiation). With direct normal radiation This can be obtained by combining the angle of solar incidence); discrete categorical variables include: time of day categories. and week categories These variables reflect the temporal distribution of thermal disturbances (such as heat dissipation from people and equipment) inside the building.

[0032] S200. Establish a nonlinear regression relationship between historical operating data and building inertial thermal energy storage state of charge using a generalized additive model, and output the building inertial thermal energy storage state of charge at the current moment based on the nonlinear regression relationship. The core of this step is to use a generalized additive model as a data-driven modeling tool to establish a nonlinear mapping relationship between external measurable variables and the state of charge of building inertial thermal energy storage, thereby replacing the traditional physical model based on fixed RC parameters and achieving higher accuracy in SoE estimation.

[0033] The generalized additive model (GAM) is a highly interpretable nonparametric regression method. Its basic idea is to represent the expected value of the dependent variable as a linear sum of the smoothing functions of each predictor variable. Compared to traditional multiple linear regression models, GAM's core advantage lies in the fact that it does not require prior assumptions about linear relationships between variables. Instead, it adaptively learns the influence function form of each variable through data, thus flexibly capturing nonlinear effects in building thermal dynamics.

[0034] Combination Figure 2 Specifically, S200 includes the following steps: S201. Based on the building's inertial thermal energy storage state of charge and historical operating data, establish a mapping architecture; introduce an interaction term function into the mapping architecture, which is used to capture the coupling influence between various predictive variables on the building's inertial thermal energy storage state of charge under composite conditions. In this invention, the building inertial thermal energy storage state of charge (SoE) is defined as the ratio of the thermal energy stored in the building envelope at the current moment to its maximum storable thermal energy, with a value ranging from 0 to 1. The SoE value reflects the building's energy storage level at the current moment and serves as an input parameter for subsequent flexibility boundary quantification.

[0035] The selection criteria for predictor variables are as follows: (i) Outdoor temperature is the main driver of building heat loss, and there is a significant nonlinear relationship between it and the building's inertial thermal energy storage state of charge (SoE). For example, in the lower temperature range, heat loss increases rapidly with increasing temperature difference, while in the higher temperature range, heat gain tends to saturate with increasing temperature difference.

[0036] (ii) Solar irradiance directly affects the heat gain process of a building, and its relationship with the building's inertial thermal energy storage state of charge (SoE) also exhibits nonlinear characteristics. For example, in the low irradiance range, the heat gain increases approximately linearly with irradiance; however, in the high irradiance range, due to factors such as window shading and heat capacity saturation, the increase in heat gain gradually decreases.

[0037] (iii) Time-based categorical variables are used to characterize the periodic changes in thermal disturbances within a building, including human activity, lighting heat dissipation, and equipment operation. Although these internal thermal disturbances are difficult to measure directly, they have obvious diurnal periodicity and can be indirectly modeled through time-based categorical variables.

[0038] (iv) Weekday category variables are used to characterize the systematic differences between weekdays and rest days in terms of personnel stay patterns, equipment usage intensity, etc.

[0039] The purpose of selecting the above variables is to integrate external environmental factors (temperature, radiation) and internal behavioral patterns (time of day, week) into the mapping framework through a generalized additive model, thereby achieving a comprehensive characterization of the building's thermal state.

[0040] S202. Based on the mapping architecture, the outdoor temperature and the solar irradiance are respectively subjected to nonlinear fitting using spline smoothing functions to generate smoothing function terms corresponding to the outdoor temperature and solar irradiance; the time category variable and the week category variable are respectively encoded using category functions to generate category function terms corresponding to the time category variable and the week category variable. Spline smoothing is a piecewise polynomial function. Its basic idea is to divide the range of a variable into several sub-intervals, fit a low-order polynomial (usually a cubic polynomial) to each sub-interval, and ensure the smooth continuity of the function at the connection points of the sub-intervals. Compared to a single global polynomial, spline functions have stronger local fitting capabilities and can better capture the different changing patterns of variables across different ranges. For example, the impact mechanism of a cooling process in the outdoor temperature range of -10℃ to 5℃ on the SoE (Solar Equivalent) is different from that in the temperature range of 20℃ to 35℃. Spline smoothing functions can learn suitable response curves for different intervals.

[0041] The category function encodes discrete categorical variables into numerical features for use by the model. For time-based categorical variables, it divides a 24-hour day into several time-segment categories; for weekday-based categorical variables, it divides Monday to Sunday into weekdays and weekends. The value of the category function represents the average effect of that category, i.e., the offset relative to the baseline category.

[0042] It should be noted that the reason for introducing the interaction term function is that the building thermal dynamics process is not a simple superposition of various external factors, but rather involves coupling effects. For example, the combined conditions of "high temperature at noon combined with strong sunlight" and "low temperature at noon combined with strong sunlight" have drastically different effects on SoE. The former is subjected to the dual heat gain of high temperature and strong radiation, which may lead to a rapid increase in SoE; the latter, although subject to strong radiation, has a low outdoor temperature, and some heat is lost through the building envelope, resulting in a smaller increase in SoE.

[0043] S204. Linearly sum the smoothing function term, category function term, and interaction function term to construct a building inertial thermal energy storage model, so as to characterize the nonlinear regression relationship between historical operating data and the building inertial thermal energy storage state of charge.

[0044] The linear summation of the above terms yields the complete expression for the GAM state estimator. Each function term of the GAM in this model independently reflects the marginal impact of the corresponding predictor variable on the SoE. This allows the model to provide high-accuracy predictions while maintaining physical interpretability, facilitating understanding and verification of the model output by engineers.

[0045] Specifically, the complete expression for the GAM state estimator can be found in the following formula (1): Formula (1); In the formula, For the join function, the present invention preferably uses the identity join function, i.e. ; For expectation operators; For building inertial thermal energy storage at any time The state of charge; For building inertial thermal energy storage The state of charge at any given moment; For building inertial thermal energy storage at any time The heat input to the power-to-heat conversion equipment; This is the self-discharge loss coefficient; For heat charging efficiency; This represents the time step for discretization; , , These are smoothing function terms for outdoor temperature, diffuse horizontal radiation, and direct normal radiation, respectively, used to capture the nonlinear effects of outdoor temperature and solar irradiance; and These are functions for time category and weekday category variables, respectively, used to capture pattern differences between weekdays and holidays, and between daytime and nighttime; This refers to the interaction between outdoor temperature and time category; This is the interaction term between outdoor temperature and weekday category; This is the interaction term between scattered radiation and time category; This is the interaction term between scattered radiation and weekday category; This is the interaction term between direct radiation and time category; These are the interaction terms between direct radiation and weekday categories. The above interaction terms are the interaction term functions used to capture the coupling effects under combined extreme conditions.

[0046] Furthermore, the training process of the building inertial thermal energy storage model includes the following steps: collecting historical monitoring data from temperature sensors and smart meters of the target building over multiple meteorological years as training samples for the building inertial thermal energy storage model; based on the training samples of the building inertial thermal energy storage model, using the generalized cross-validation criterion, by minimizing the sum of squared residuals, determining the form and corresponding smoothing parameters of the smoothing function term, category function term, and interaction function term in the building inertial thermal energy storage model, as well as determining the physical parameters to be identified in the building inertial thermal energy storage model.

[0047] In practical applications, the physical parameters to be identified during the training of building inertial thermal energy storage models include: thermal efficiency. and self-discharge loss coefficient Heat charging efficiency This reflects the efficiency of the building envelope in converting input electrical energy into effectively stored thermal energy; its value is affected by factors such as building heat capacity and heat transfer path; self-discharge loss coefficient. It reflects the rate at which stored heat energy is naturally dissipated over time, and its value is affected by factors such as building insulation performance and outdoor temperature.

[0048] The Generalized Cross Validation (GCV) criterion is calculated as follows: for a given combination of smoothing parameters, the average prediction error of the model under leave-one-out cross-validation is calculated, and the combination of smoothing parameters that minimizes this error is selected. The GCV criterion strikes a balance between model goodness of fit and model complexity, avoiding overfitting or underfitting. Finally, the trained building inertial thermal energy storage model will output two physical parameters to be identified: thermal efficiency and self-discharge loss coefficient, as well as a series of smoothing function terms and interaction terms. In actual operation, the building inertial thermal energy storage model can quickly calculate the SoE value at the current time based on real-time input outdoor temperature, solar irradiance, time category variables, and day-of-week category variables, without needing to solve complex physical equations or perform iterative calculations.

[0049] Through the above steps, this invention constructs a building energy storage state estimator that can accurately perceive the impact of complex environmental factors, laying a data foundation for the subsequent quantification of the flexibility of virtual energy storage.

[0050] S300: Input the current building inertial thermal energy storage charge state into the preset building thermal dynamic model, and calculate the upward flexibility power boundary and downward flexibility power boundary of the target building in the current state in combination with user thermal comfort constraints. The core of this step lies in inputting the SoE value estimated by the data-driven model in S200 into the physical mechanism-based building thermal dynamics model, thereby achieving the integration of data-driven and physical modeling. On the one hand, the GAM model provides high-precision real-time SoE estimation, solving the problem of state estimation bias caused by fixed parameters in traditional physical models; on the other hand, the thermal dynamics model provides physical constraints on temperature evolution, ensuring that the calculation of the flexibility boundary does not violate thermodynamic laws.

[0051] Combination Figure 3 Specifically, S300 includes the following steps: S301, construct a building thermal dynamic state space model and set user thermal comfort constraints; among which, the building thermal dynamic state space model is used to characterize the trajectory of indoor air temperature change under different thermal power inputs; This step employs a second-order resistive-capacitive model to describe the building's thermal dynamics, thus constructing a spatial model of the building's thermal dynamics. Specifically, the 3R2C model equates the building to a "resistor-capacitor" network, where resistance represents the thermal resistance along the heat transfer path, and capacitance represents the thermal capacity of the building components. In detail, the 3R2C model includes three resistive and two capacitive nodes, representing the heat exchange relationship between indoor air nodes and wall nodes, respectively.

[0052] Specifically, define the state vector. ,in This represents the change in indoor air temperature over time. The average temperature of the building envelope (walls / ceiling). Let be the transpose matrix. The state-space equations of the thermal dynamic state-space model are expressed as follows: Equations (2) and (3): Formula (2); Formula (3); In the formula, the output The measurable indoor air temperature; State vector Regarding time The first derivatives, matrices A, B, and C are composed of thermal resistance and heat capacity parameters, and each matrix is ​​specifically described in formulas (4), (5), and (6) below: Formula (4); Formula (5); Formula (6); In the formula, , , These are respectively the wall conduction thermal resistance, the wall-air coupling thermal resistance, and the convection / osmosis thermal resistance. , These are the wall heat capacity and the indoor air heat capacity, respectively. Input vector Includes outdoor temperature over time Variables of solar irradiance over time The variable of input heat power of electric heating equipment over time ; , , , All of these are heat flow distribution coefficients, representing the proportion of solar heat gain and heat pump heat gain distributed to wall nodes and air nodes. Their values ​​can be 0.6, 0.4, 0.5 and 0.5 respectively, without any special restrictions.

[0053] Discretizing the above continuous state-space equations yields a discrete form suitable for digital control. This discrete equation forms the basis for predicting future temperature trajectories, and its physical meaning is: given the current indoor temperature, wall temperature, and input heat power, predict the temperature value at the next moment; see the following formula (7) for the specific form: Formula (7); In the formula, I is the identity matrix. For natural index; In order to be in The state vector at any given time; In order to be in The input vector at time step; In order to be in The state vector at time t.

[0054] Next, user thermal comfort constraints are set based on the thermal comfort ranges defined in ISO 7730 or ASHRAE Standard 55, limiting the indoor temperature to an acceptable upper and lower limit for users. The upper and lower temperature limits of the thermal comfort constraints can be flexibly configured according to building type, season, and user preferences.

[0055] Specifically, in this invention, the indoor temperature is set. The hard constraint boundary that must be satisfied to ensure that the user does not accept the damage can be found in the following formula (8): Formula (8); In the formula, The lower limit temperature of thermal comfort in the termination boundary corresponding to the user's thermal comfort constraint; In order to be in The indoor temperature of the target building at all times; The upper limit temperature of thermal comfort in the termination boundary corresponding to the user thermal comfort constraint; This is the set of time steps for scheduling or control.

[0056] S302. Based on the current state of charge of the building's inertial thermal energy storage, the maximum thermal energy capacity of the building's inertial thermal energy storage, and the midpoint temperature of the user's thermal comfort constraint, calculate the reference indoor temperature at the current moment, and based on the state of charge of the building's inertial thermal energy storage and the range of thermal energy capacity, calculate the remaining chargeable space and the remaining heat release space at the current moment respectively. This step establishes a data-driven link between the SoE (Solar Equilibrium) and the physical temperature field. The reference indoor temperature represents the indoor air temperature level at the current SoE level. When the SoE is high, it indicates that the building envelope has stored a significant amount of heat, and the reference indoor temperature is close to the upper comfort limit; conversely, when the SoE is low, the reference indoor temperature is close to the lower comfort limit.

[0057] Specifically, it facilitates the real-time output of the building's inertial thermal energy storage state of charge from the building's inertial thermal energy storage model, and combines this with the midpoint temperature constrained by user thermal comfort. Maximum permissible temperature difference with the target building Calculate the current reference indoor temperature The process is shown in the following formula (9): Formula (9); In the formula, Set the midpoint temperature for comfort; For the target building in The state of charge of the building's inertial thermal energy storage at all times; The maximum allowable temperature difference for the target building; This represents the change in baseline indoor temperature over time.

[0058] Next, the remaining inductible heat space refers to the additional heat energy that the building can absorb without exceeding the upper limit of comfort temperature; the remaining inductible heat space refers to the heat energy that the building can release without falling below the lower limit of comfort temperature. In summary, these two physical quantities correspond to the thermodynamic upper limits of downward flexibility and upward flexibility, respectively, as detailed in formulas (10) and (11) below: Formula (10); Formula (11); In the formula, For the target building in The remaining chargeable space at any given time; The maximum thermal capacity within the target building's thermal capacity range; For the target building in The state of charge of the building's inertial thermal energy storage at all times; For the target building in The remaining space available for heat dissipation at any given moment; The minimum thermal capacity within the target building's thermal capacity range.

[0059] S303. Based on the remaining heat dissipation space, the current indoor air temperature change trajectory, and the termination boundary corresponding to the user thermal comfort constraint, the upward flexibility power boundary is obtained. S304. Based on the remaining rechargeable space, the current indoor air temperature change trajectory, and the termination boundary corresponding to the user's thermal comfort constraints, the downward flexibility power boundary is obtained.

[0060] The upward and downward flexibility power boundaries are core parameters for buildings to participate in grid regulation as virtual energy storage resources. Therefore, the core of this step is to determine, through simulation prediction, the maximum power consumption that can be reduced or increased by the building without violating thermal comfort constraints.

[0061] The upward flexibility power boundary is determined as follows: based on the current power consumption of the electric heating equipment, the input heat power of the electric heating equipment is gradually reduced. The indoor air temperature change trajectory is used until the indoor air temperature reaches the first preset condition at any predicted time. The cumulative power consumption value of the current electric heating equipment is then used as the upward flexibility power boundary. The first preset condition is set according to the rated power of the electric heating equipment, the termination boundary of the user's thermal comfort constraint, and the remaining heat release space. The specific setting of the first preset condition can be found in the following formulas (13), (14), and (15).

[0062] Specifically, assume the building is currently operating at a baseline power, with the indoor temperature at a stable level. When the system needs to adjust upwards (i.e., reduce electrical load), the output heat power of the electrothermal equipment gradually decreases, and the building relies on its stored thermal inertia to maintain the indoor temperature. As the heat power decreases, the indoor temperature gradually drops. When the indoor temperature reaches the user's thermal comfort lower limit at any predicted time, it indicates that the maximum reduction limit has been reached. Further reduction will cause the indoor temperature to exceed the comfort range. Therefore, the cumulative reduction power value at this point is taken as the upward flexibility power boundary. Thus, upward flexibility essentially corresponds to the "discharge" process of virtual energy storage: the building uses its stored thermal energy to maintain the indoor temperature, releasing the electrical energy originally used for heating to power other loads on the power grid. Its upper limit is limited by the current operating power, the comfort lower limit temperature, and the building's inertial thermal energy storage charge limit. The specific process can be converted into the following mathematical description to solve the optimization problem: Formula (12); Formula (13); Formula (14); Formula (15); In the formula, For upward flexibility power boundary; To obtain the maximum value; The power regulation quantity to be solved; "Subject to" is an abbreviation for "meeting the following constraints". () represents the current baseline electrical power; The power consumption of the electric heat exchange equipment; This refers to the change in power consumption of the electric heat conversion equipment. The limit of the building's inertial thermal energy storage charge state for the target building; This is the lower limit temperature for thermal comfort. The predicted indoor temperature for the next moment; The predicted value of the building's inertial thermal energy storage charge state at the next moment.

[0063] Furthermore, the downward flexibility power boundary is determined as follows: based on the current power consumption of the electric heating equipment, the power consumption of the electric heating equipment is gradually increased, and the trajectory of indoor air temperature changes is used until the indoor air temperature reaches the second preset condition at any predicted time. The cumulative increase in power consumption of the current electric heating equipment is then used as the downward flexibility power boundary. The second preset condition is also set based on the rated power of the electric heating equipment, the termination boundary of the user's thermal comfort constraint, and the remaining chargeable space. The specific setting of the second preset condition can be found in the following formulas (17), (18), and (19).

[0064] Specifically, when the system needs to adjust downwards (i.e., increase the electrical load to absorb excess renewable energy), the output heat power of the power-to-heat conversion equipment gradually increases, the building absorbs more heat energy and stores it in the building envelope, and the indoor temperature gradually rises. When the indoor temperature reaches the upper limit of user thermal comfort at any predicted time, it indicates that the maximum increase has been reached, and further increases will cause the indoor temperature to exceed the comfort range. The cumulative increase in power at this point is taken as the downward flexibility power boundary. Thus, downward flexibility essentially corresponds to the "charging" process of virtual energy storage: the building uses excess electrical energy to convert it into heat energy and stores it in the building envelope. Its upper limit is limited by the rated power of the equipment, the upper limit of comfort temperature, and the upper limit of the building's inertial thermal energy storage state of charge. The specific process can be converted into the following mathematical description to solve the optimization problem: Formula (16); Formula (17); Formula (18); Formula (19); In the formula, For downward flexibility power boundary; The power regulation quantity to be solved; Current baseline electrical power; The upper limit of the building's inertial thermal energy storage state of charge for the target building; Rated power of the equipment; This refers to the upper limit temperature for thermal comfort. The predicted indoor temperature for the next moment; The predicted value of the building's inertial thermal energy storage charge state at the next moment.

[0065] To be used by the upper-level optimizer, this embodiment not only calculates the power boundary at the current moment, but also predicts the temperature and the building inertial thermal energy storage charge state trajectory of the target building in the future multiple steps, and outputs a flexibility profile in the form of a time series and the corresponding building inertial thermal energy storage charge state change constraints.

[0066] Formula (20); Formula (21); Formula (22); In the formula, For downward flexibility power boundary; For upward flexibility power boundary; For a moment The charging power; This represents the current level of thermal energy storage. Maximum thermal storage capacity; To predict the number of time-domain steps.

[0067] S400. Based on the upward flexibility power boundary and downward flexibility power boundary of each target building, the flexibility curve of the building cluster is obtained. The core of this step lies in aggregating the individual flexibility resources of multiple buildings to form a cluster flexibility curve that can be used for distribution network-level dispatch. Since different buildings differ in terms of thermal inertia of their building envelopes, energy consumption patterns, and comfort preferences, simple addition or averaging will lead to aggregation bias. This invention achieves refined aggregation by recording the flexibility boundaries of all buildings at the same time.

[0068] Specifically, based on the individual upward and downward flexibility power boundaries of each target building, the upward flexibility power boundaries of each target building at the same time are recorded to obtain the cluster upward flexibility power boundary at that time; similarly, the downward flexibility power boundaries of each target building at the same time are recorded to obtain the cluster downward flexibility power boundary at that time. Finally, by traversing every time within multiple future prediction steps, the cluster upward and downward flexibility power boundaries corresponding to each time are obtained, and a building cluster flexibility curve is formed based on these boundaries.

[0069] Through the above aggregation method, the building cluster flexibility curve retains the differences of each building in terms of thermal inertia, energy consumption patterns, comfort preferences, etc., while providing a unified description of the power adjustment feasible domain for the upper-level optimizer.

[0070] S500: Construct a joint optimization model that includes the power grid and the district heating network. With the minimum operating cost as the objective function and the building cluster flexibility curve as the constraint, combine the photovoltaic power output forecast, the grid time-of-use electricity price and the heat load demand to solve the power allocation data for the future forecast time. Based on the power allocation data, issue scheduling instructions to the power-to-heat equipment of each target building.

[0071] The core of this step is to embed the GAM state estimator in S200 and the building thermal dynamics model in S300 into the multi-energy flow collaborative optimization framework at the distribution network level, so as to realize a complete closed loop from single building state perception to cluster collaborative control; the future prediction time here can refer to the coverage of the next 24 hours, without any special limitation.

[0072] Combination Figure 3 Specifically, the S500 includes the following steps: S501. Embed the building thermal dynamic state space model and nonlinear regression relationship into the multi-energy flow collaborative optimization framework at the park or power distribution network level, and construct a joint optimization model by adopting the rolling time-domain optimization mechanism of model predictive control. Unlike traditional open-loop optimization methods, Model Predictive Control (MPC) employs a rolling time-domain optimization strategy: at each control time step, based on the current system state and future prediction information, a control sequence within a finite time domain is solved, but only the control command for the current time step is executed; at the next time step, the solution is re-solved based on the latest state feedback, so that scheduling decisions can respond promptly to fluctuations in renewable energy output and load changes. In other words, in the joint optimization model, the nonlinear regression relationship represented by the GAM state estimator infers the real-time input data at each control time step, outputting the current state of charge (SOC) value of each target building, which serves as the initial state of the MPC prediction model; the building thermal dynamics state-space model, as the internal prediction model of MPC, iteratively calculates the indoor temperature trajectory and SOC change trajectory for future time steps based on the current state and candidate control input sequences within each prediction step. Through this approach, two data-driven / physical models are embedded into the MPC's prediction-optimization-execution loop.

[0073] Specifically, model predictive control performs the following three steps at each control time: (Step 1) Prediction: Based on the current system state With the future The system uses predictive inputs (including photovoltaic power output and heat load demand) to predict future output trajectories using a system model.

[0074] (Step 2) Optimization: Solve a finite-time optimization problem to obtain the future control sequence. .

[0075] (Step 3) Loop: Only control quantity at the current moment is used. Acting on the physical system; until the next moment Then, remeasure the state and repeat the above process.

[0076] S502. Using the photovoltaic forecast output, grid time-of-use electricity price, and heat load demand as the input boundary conditions of the joint optimization model, the objective function is to minimize the operating cost. The joint optimization model is then solved under constraints to obtain the power allocation sequence for the future forecast period. The constraints include at least the power adjustment range constraint defined by the building cluster flexibility curve. Photovoltaic power output forecasts are predicted values ​​of photovoltaic power generation over a future period, obtained by extrapolating numerical weather forecasts or historical data. Their uncertainty can be described by the forecast interval. Time-of-use pricing is a key factor influencing economic decisions, meaning prioritizing electricity purchases or increasing load during off-peak hours and prioritizing the use of local photovoltaic power or reducing load during peak hours.

[0077] The objective function is preferably to minimize the total operating cost of the system, which should at least include the cost of purchasing electricity from the grid, the revenue from selling electricity, and the operation and maintenance / fuel costs of each piece of equipment, as shown in the following formula (23): Formula (23); In the formula, To obtain the minimum value; Let be the objective function. for Real-time power grid purchase for The electricity price at any given time; for Constantly importing electrical power from the power grid; for The electrical power output to the power grid at any given time; This represents the total number of steps in the entire scheduling cycle.

[0078] also, For the first Class unit at time The operation and maintenance / fuel cost item mainly includes CHP, boiler, P2H, and conventional hot tank HWT, and its specific forms are as follows: Formula (24); Formula (25); Formula (26); Formula (27); In the formula, For CHP equipment Total operating cost at any given time; This refers to the unit price of fuel. This is the power operation and maintenance cost coefficient; For CHP equipment Thermal power output at any given moment; For CHP equipment Electrical power output at any given moment; This represents the cost coefficient for thermal operation and maintenance.

[0079] For boiler equipment Total operating cost at any given time; For boilers in Thermal power output at any given moment; For P2H equipment Total operating cost at any given time; For P2H equipment in Thermal power output at any given moment; For conventional hot tank equipment in Total operating cost at any given time; For conventional hot tank equipment in Thermal power output at any given time.

[0080] The constraints include, but are not limited to, electrical power balance constraints, thermal power balance constraints, SoE (Social Energy Expansion) state equation constraints, SoE upper and lower bound constraints, and thermal power boundary constraints. Among these, the power regulation range constraint defined by the building cluster flexibility curve is a key constraint connecting the building side and the grid side. It aggregates the single-building flexibility boundary calculated in S300 into a cluster-level power regulation feasible region, thus limiting the regulation range of the total power of the power-to-heat conversion equipment in the optimization problem.

[0081] The specific constraints are as follows: (Constraint 1) Electric power balance constraint Formula (28); (Constraint 2) Thermal power balance constraint Formula (29); In the formula, For photovoltaics The electrical power output at any given time; The battery is The electrical power output at any given time; For CHP equipment The electrical power output at any given time; The electrical power purchased from the power grid; The battery is The electrical power consumed at all times; For P2H equipment in The electrical power consumed at all times; For other charges in the building The electrical power consumed at all times; This refers to the electrical power sold from the power grid.

[0082] For photovoltaics The thermal power output at all times; For P2H equipment in The thermal power output at all times; For hot water storage tanks The heat output power at all times; For boilers in The thermal power output at all times; For hot water storage tanks The electrical power consumed at all times; For the heat load at the end of the building heating system; The static heat loss power of the target building envelope; The target building's dynamic heat loss power.

[0083] (Constraint 3) Constraints on the State Equation of Building Inertial Thermal Energy Storage (SoE) Formula (30); In the formula, For the GAM external disturbance term set, it can be directly calculated using the predicted meteorological / temporal variables in each prediction step of MPC; This is the self-discharge loss coefficient; For building inertial thermal energy storage The state of charge at any given moment; For building inertial thermal energy storage The state of charge at any given moment; For building inertial thermal energy storage The heat output of the power-to-heat conversion equipment at any given time; For heat charging efficiency.

[0084] (Constraint 4) SoE Upper and Lower Bound Constraints Formula (31); In the formula, This represents the minimum state of charge of the building's inertial thermal energy storage. For building inertial thermal energy storage The state of charge at any given moment; This represents the maximum value of the building's inertial thermal energy storage state of charge.

[0085] (Constraint 5) Boundary constraints on heat charging power Formula (32); In the formula, For building inertial thermal energy storage at any time The heat input to the power-to-heat conversion equipment; For the target building in The upward flexibility power boundary at any time, the negative sign in the constraint condition indicates the direction of power reduction, and is used to limit the lower limit of the power adjustment amount; For the target building Downward flexibility power boundary at any given moment; For the target building The upward flexibility power boundary at any given moment.

[0086] (Constraint 6) Efficiency and self-discharge parameter constraints Formula (33); Formula (34); In the formula, For heat charging efficiency; This is the self-discharge loss coefficient.

[0087] S503. Based on the power allocation sequence, issue scheduling instructions to the power conversion equipment of each target building according to the power allocation data corresponding to the current time.

[0088] The power allocation sequence obtained from the MPC solution contains scheduling instructions for the current time and future time periods. The system issues scheduling instructions to the power-to-heat conversion equipment in each target building according to the power allocation data corresponding to the current time, including the start / stop status and power setpoint of each power-to-heat conversion device. Upon receiving the instructions, the controllers of the power-to-heat conversion equipment in each building execute the corresponding power adjustment operations, completing the entire control link from global optimization to local execution. In this invention, power-to-heat conversion equipment refers to all types of equipment that convert electrical energy into heat energy, including but not limited to electric boilers, air-source heat pumps, water-source heat pumps, and electric heaters.

[0089] The above text combined Figures 1-4 The load control method based on building thermal dynamic characteristics provided in the embodiments of the present invention has been described in detail below. Figure 5 , Figure 6 The system, device and medium provided in the embodiments of the present invention will be described.

[0090] Combination Figure 6 This invention also proposes a load control system based on building thermal dynamics, the load control system 600 specifically comprising the following: The data acquisition module 601 is used to collect historical operating data of all target buildings. The historical operating data includes at least outdoor temperature, solar irradiance, time category variables, and week category variables. The charge state value output module 602 is used to establish a nonlinear regression relationship between historical operating data and building inertial thermal energy storage charge state using a generalized additive model, and output the building inertial thermal energy storage charge state at the current moment based on the nonlinear regression relationship. The power boundary output module 603 is used to input the current building inertial thermal energy storage charge state into the preset building thermal dynamic model, and calculate the upward flexibility power boundary and downward flexibility power boundary of the target building in the current state in combination with user thermal comfort constraints; and to obtain the building cluster flexibility curve based on the upward flexibility power boundary and downward flexibility power boundary of each target building. The dispatch instruction processing module 604 is used to construct a joint optimization model that includes the power network and the district heating network. The objective function is to minimize the operating cost, and the building cluster flexibility curve is used as the constraint. It combines the photovoltaic power output forecast, the grid time-of-use electricity price and the heat load demand to solve the power allocation data in the future forecast time. Based on the power allocation data, it issues dispatch instructions to the power-to-heat equipment of each target building.

[0091] In a preferred embodiment, the charge state value output module 602 is further configured to establish a mapping architecture based on the building's inertial thermal energy storage state of charge and historical operating data; an interaction term function is introduced into the mapping architecture, which is used to capture the coupling influence between various predictive variables on the building's inertial thermal energy storage state of charge under combined operating conditions; based on the mapping architecture, spline smoothing functions are used to perform nonlinear fitting on the outdoor temperature and the solar irradiance, respectively, to generate smoothing function terms corresponding to the outdoor temperature and the solar irradiance; categorical functions are used to encode the time-of-day category variable and the week-of-day category variable, respectively, to generate categorical function terms corresponding to the time-of-day category variable and the week-of-day category variable; the smoothing function terms, categorical function terms, and interaction term function are linearly summed to construct a building inertial thermal energy storage model to characterize the nonlinear regression relationship between the historical operating data and the building's inertial thermal energy storage state of charge.

[0092] In a preferred embodiment, the training process of the building inertial thermal energy storage model obtained in the charge state value output module 602 includes: collecting historical monitoring data of temperature sensors and smart meters of target buildings in multiple meteorological years; based on the historical monitoring data of temperature sensors and smart meters, using the generalized cross-validation criterion, by minimizing the sum of squared residuals, determining the form and corresponding smoothing parameters of the smoothing function term, the function term of each category, and the interaction term function in the building inertial thermal energy storage model, as well as determining the physical parameters to be identified in the building inertial thermal energy storage model.

[0093] In a preferred embodiment, the power boundary output module 603 is further used to construct a building thermal dynamic state space model and set user thermal comfort constraints; wherein, the building thermal dynamic state space model is used to characterize the trajectory of indoor air temperature change under different thermal power inputs; using the current building inertial thermal energy storage charge state, the maximum thermal energy capacity of the building inertial thermal energy storage, and the midpoint temperature of the user thermal comfort constraints, the reference indoor temperature at the current moment is calculated, and based on the building inertial thermal energy storage charge state and the maximum thermal energy capacity, the remaining chargeable heat space and the remaining heat release space at the current moment are calculated respectively; according to the remaining heat release space, the current indoor air temperature change trajectory, and the termination boundary corresponding to the user thermal comfort constraints, the upward flexibility power boundary is obtained; according to the remaining chargeable heat space, the current indoor air temperature change trajectory, and the termination boundary corresponding to the user thermal comfort constraints, the downward flexibility power boundary is obtained.

[0094] Furthermore, the target building is connected to an electric heat exchanger; the upward flexibility power boundary is determined as follows: based on the current power consumption of the electric heat exchanger, the input heat power of the electric heat exchanger is gradually reduced, and the trajectory of indoor air temperature change is used until the indoor air temperature reaches the first preset condition at any predicted time. The cumulative power consumption reduction of the current electric heat exchanger is then used as the upward flexibility power boundary. The downward flexibility power boundary is determined as follows: based on the current power consumption of the electric heat exchanger, the power consumption of the electric heat exchanger is gradually increased. The indoor air temperature change trajectory is used until the indoor air temperature reaches the second preset condition at any predicted time. The cumulative increase in power consumption of the current electric heat exchanger is then used as the downward flexibility power boundary. The first and second preset conditions are set based on the rated power of the electric heat transfer equipment, the termination boundary of the user's thermal comfort constraint, and the remaining heat release space.

[0095] In a preferred embodiment, the power boundary output module 603 is further configured to record the upward flexibility power boundary of each target building at the same time, based on the upward flexibility power boundary and downward flexibility power boundary of each target building, to obtain the cluster upward flexibility power boundary at that time; record the downward flexibility power boundary of each target building at the same time, to obtain the cluster downward flexibility power boundary at that time; traverse each time within multiple future prediction steps to obtain the cluster upward flexibility power boundary and cluster downward flexibility power boundary corresponding to each time, and form a building cluster flexibility curve based on the cluster upward flexibility power boundary and cluster downward flexibility power boundary corresponding to each time.

[0096] In a preferred embodiment, the scheduling instruction processing module 604 is further configured to embed the building thermal dynamic state space model and nonlinear regression relationship into a multi-energy flow collaborative optimization framework at the park or distribution network level, and to construct a joint optimization model using a rolling time-domain optimization mechanism based on model predictive control; using photovoltaic predicted output, grid time-of-use electricity price, and heat load demand as input boundary conditions for the joint optimization model, and using the lowest operating cost as the objective function, and solving the joint optimization model under constraints to obtain the power allocation sequence within the future prediction time; the constraints include at least the power adjustment range constraint defined by the building cluster flexibility curve; and, based on the power allocation sequence, issuing scheduling instructions to the power-to-heat conversion equipment of each target building according to the power allocation data corresponding to the current time.

[0097] The load control system 600 according to an embodiment of the present invention can correspond to performing the method described in the embodiment of the present invention, and the above and other operations and / or functions of each module of the load control system 600 are respectively for implementing Figure 1The corresponding process of the method in the illustrated embodiment will not be described in detail here for the sake of brevity.

[0098] A terminal device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a load control method based on building thermal dynamics as described in Embodiment 1.

[0099] In this embodiment, as Figure 6 As shown, the terminal device 500 includes a CPU 501 (Central Processing Unit), which can perform various appropriate actions and processes according to a program stored in ROM 502 (Read-Only Memory) or a program loaded from storage into RAM 503 (Random Access Memory). RAM 503 also stores various programs and data required for system operation. The CPU 501, ROM 502, and RAM 503 are interconnected via bus 504. An I / O interface 505 (Input / Output Interface) is also connected to bus 504.

[0100] The following components are connected to I / O interface 505: an input section 506 including a keyboard, mouse, etc.; an output section 507 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive is also connected to I / O interface 505 as needed. A removable medium 511, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 510 as needed so that computer programs read from it can be installed into storage section 508 as needed.

[0101] In particular, according to embodiments of the present invention, the above-described flowchart is as follows. Figure 1 The described process can be implemented as a computer software program. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network via a communication component, and / or installed from a removable medium. When the computer program is executed by the central processing unit (CPU 501), it performs the functions defined in the system of the present invention.

[0102] It should be noted that the computer-readable medium shown in this invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM503), read-only memory (ROM502), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0103] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0104] The units described in the embodiments of the present invention can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the specific unit itself. The described units or modules can also be located in a processor; for example, a processor can be described as including a first generation module, an acquisition module, a search module, a second generation module, and a merging module. The names of these units or modules do not necessarily limit the specific unit or module itself; for example, the acquisition module can also be described as "an acquisition module for acquiring multiple instances to be probed in the base table".

[0105] The present invention also provides a computer-readable medium, which may be included in the electronic device described in the above embodiments; or it may exist independently and not assembled into the electronic device. The computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to implement a load control method based on building thermal dynamics as described in the above embodiments.

[0106] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention is not limited to the specific combination of the above-described technical features, but also includes other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in this invention.

Claims

1. A load regulation method based on building thermal dynamic characteristics, characterized in that, Includes the following steps: Collect historical operational data for all target buildings. The historical operational data includes at least outdoor temperature, solar irradiance, time-of-day category variables, and week-of-day category variables. A nonlinear regression relationship between the historical operating data and the building's inertial thermal energy storage state of charge is established using a generalized additive model, and the current building's inertial thermal energy storage state of charge is output based on the nonlinear regression relationship. The current building inertial thermal energy storage charge state is input into the preset building thermal dynamics model, and combined with user thermal comfort constraints, the upward flexibility power boundary and downward flexibility power boundary of the target building in the current state are calculated. Based on the upward and downward flexibility power boundaries of each target building, the flexibility curve of the building cluster is obtained; A joint optimization model incorporating the power grid and the district heating network is constructed, with the minimum operating cost as the objective function and the building cluster flexibility curve as the constraint. Combining the predicted photovoltaic output, time-of-use electricity pricing, and heat load demand, the model solves for power allocation data within the predicted future timeframe. Based on this power allocation data, scheduling instructions are issued to the power-to-heat conversion equipment of each target building. The step involves inputting the current building inertial thermal energy storage charge state into a preset building thermal dynamics model, and, in conjunction with user thermal comfort constraints, calculating the upward and downward flexibility power boundaries of the target building in the current state. Specifically, this includes: A building thermal dynamic state space model is constructed, and user thermal comfort constraints are set; wherein, the building thermal dynamic state space model is used to characterize the trajectory of indoor air temperature change under different thermal power inputs; Based on the current state of charge of the building's inertial thermal energy storage, the maximum thermal capacity of the building's inertial thermal energy storage, and the midpoint temperature of the user's thermal comfort constraints, the reference indoor temperature at the current moment is calculated. Based on the state of charge of the building's inertial thermal energy storage and the range of thermal capacity, the remaining space that can be charged and the remaining space that can be released at the current moment are calculated respectively. The upward flexibility power boundary is obtained based on the remaining heat dissipation space, the current indoor air temperature change trajectory, and the termination boundary corresponding to the user thermal comfort constraint. The downward flexibility power boundary is obtained based on the remaining rechargeable heatable space, the current indoor air temperature change trajectory, and the termination boundary corresponding to the user thermal comfort constraint.

2. The method according to claim 1, characterized in that, The method of establishing a nonlinear regression relationship between the historical operating data and the building's inertial thermal energy storage state of charge using a generalized additive model includes: A mapping architecture is established based on the building's inertial thermal energy storage state of charge and historical operating data. An interaction term function is introduced into the mapping architecture to capture the coupling influence between various predictive variables on the building's inertial thermal energy storage state of charge under combined operating conditions. Based on the mapping architecture, spline smoothing functions are used to perform nonlinear fitting on the outdoor temperature and the solar irradiance to generate smoothing function terms corresponding to the outdoor temperature and the solar irradiance, respectively; and class functions are used to encode the time category variable and the week category variable to generate class function terms corresponding to the time category variable and the week category variable, respectively. The smoothing function, category function, and interaction function are linearly summed to construct a building inertial thermal energy storage model, which characterizes the nonlinear regression relationship between the historical operating data and the building inertial thermal energy storage state of charge.

3. The method according to claim 2, characterized in that, The training process of the building inertial thermal energy storage model includes: Historical monitoring data from temperature sensors and smart meters of target buildings were collected from multiple meteorological years and used as training samples for the building's inertial thermal energy storage model. Based on the training samples of the building inertial thermal energy storage model, the generalized cross-validation criterion is adopted to determine the form and corresponding smoothing parameters of the smoothing function term, category function term, and interaction function term in the building inertial thermal energy storage model by minimizing the residual sum of squares, as well as the physical parameters to be identified in the building inertial thermal energy storage model.

4. The method according to claim 1, characterized in that, The target building is connected to an electric heat transfer device; the upward flexibility power boundary is determined in the following manner: Based on the current power consumption of the electric heat exchanger, the input heat power of the electric heat exchanger is gradually reduced. The indoor air temperature change trajectory is used until the indoor air temperature reaches the first preset condition at any predicted time. The cumulative power consumption reduction of the electric heat exchanger is then used as the upward flexibility power boundary. The downward flexibility power boundary is determined in the following manner: Based on the current power consumption of the electric heat exchanger, the power consumption of the electric heat exchanger is gradually increased. Using the indoor air temperature change trajectory, until the indoor air temperature reaches the second preset condition at any predicted time, the cumulative increase in power consumption of the electric heat exchanger is used as the downward flexibility power boundary. The first and second preset conditions are both set based on the rated power of the electric heat transfer equipment, the termination boundary of the user's thermal comfort constraint, and the remaining heat dissipation space.

5. The method according to claim 1, characterized in that, Based on the upward and downward flexibility power boundaries of each target building, the flexibility curves of the building cluster are obtained, including: Based on the upward flexibility power boundary and downward flexibility power boundary of each target building, the upward flexibility power boundary of each target building at the same time is recorded to obtain the cluster upward flexibility power boundary at that time; the downward flexibility power boundary of each target building at the same time is recorded to obtain the cluster downward flexibility power boundary at that time. By iterating through each time point within multiple prediction steps, the upward flexibility power boundary and the downward flexibility power boundary of the cluster corresponding to each time point are obtained. Based on the upward flexibility power boundary and the downward flexibility power boundary of the cluster corresponding to each time point, the building cluster flexibility curve is formed.

6. The method according to claim 1, characterized in that, A joint optimization model integrating the power grid and the district heating network is constructed, with the minimum operating cost as the objective function and the building cluster flexibility curve as the constraint. Combining the predicted photovoltaic output, grid time-of-use pricing, and heat load demand, the model solves for power allocation data within the predicted future timeframe. Based on this power allocation data, scheduling instructions are issued to the power-to-heat conversion equipment of each target building, including: The building thermal dynamic state space model and the nonlinear regression relationship are embedded into a multi-energy flow collaborative optimization framework at the park or power distribution network level, and the joint optimization model is constructed by adopting a rolling time-domain optimization mechanism of model predictive control. The photovoltaic power output forecast, grid time-of-use electricity price, and heat load demand are used as the input boundary conditions of the joint optimization model. The objective function is to minimize the operating cost. The joint optimization model is solved under constraints to obtain the power allocation sequence for the future forecast period. The constraints include at least the power adjustment range constraint defined by the building cluster flexibility curve. Based on the power allocation sequence, scheduling instructions are issued to the power-to-heat conversion equipment of each target building according to the power allocation data corresponding to the current time.

7. A load control system based on building thermal dynamics, characterized in that, The system includes: The data acquisition module is used to collect historical operational data of all target buildings. The historical operational data includes at least outdoor temperature, solar irradiance, time-of-day category variables, and weekday category variables. The charge state value output module is used to establish a nonlinear regression relationship between the historical operating data and the building inertial thermal energy storage charge state using a generalized additive model, and output the building inertial thermal energy storage charge state at the current moment based on the nonlinear regression relationship. The power boundary output module is used to input the current building inertial thermal energy storage charge state into the preset building thermal dynamics model, and calculate the upward flexibility power boundary and downward flexibility power boundary of the target building in the current state in combination with user thermal comfort constraints; and to obtain the building cluster flexibility curve based on the upward flexibility power boundary and downward flexibility power boundary of each target building. The dispatch instruction processing module is used to construct a joint optimization model that includes the power network and the district heating network. The objective function is to minimize the operating cost, and the building cluster flexibility curve is used as the constraint. The module combines the photovoltaic power output forecast, the grid time-of-use electricity price and the heat load demand to solve the power allocation data for the future forecast time. The module then issues dispatch instructions to the power-to-heat equipment of each target building based on the power allocation data. The power boundary output module is also used to construct a building thermal dynamic state space model and set user thermal comfort constraints. The building thermal dynamic state space model characterizes the indoor air temperature change trajectory under different thermal power inputs. Based on the current building inertial thermal energy storage charge state, the maximum thermal energy capacity of the building inertial thermal energy storage, and the midpoint temperature of the user thermal comfort constraints, the reference indoor temperature at the current moment is calculated. Based on the building inertial thermal energy storage charge state and thermal energy capacity range, the remaining chargeable heat space and remaining heat release space at the current moment are calculated respectively. The upward flexibility power boundary is obtained based on the remaining heat release space, the current indoor air temperature change trajectory, and the termination boundary corresponding to the user thermal comfort constraints. The downward flexibility power boundary is obtained based on the remaining chargeable heat space, the current indoor air temperature change trajectory, and the termination boundary corresponding to the user thermal comfort constraints.

8. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 6.

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

Citation Information

Patent Citations

  • Building integrated energy system double-layer optimization scheduling method considering virtual energy storage

    CN117952402A

  • Park energy carbon management platform, method and device

    CN121146988A