Temperature control equipment control method and system based on solar-storage direct-flexible building

Abstract

The present invention discloses a method and system for controlling temperature control equipment in a solar-powered, direct-flexible building. First, a building thermodynamic model and a multi-objective optimization function are constructed based on environmental data, static parameters, and state data of the solar-powered, direct-flexible building. The temperature adjustment timing of the temperature control equipment is determined based on the building thermodynamic model. The multi-objective optimization function is solved according to preset constraints to obtain the temperature adjustment value of the temperature control equipment. Finally, the temperature control equipment parameters are adjusted based on the temperature adjustment timing and the temperature adjustment value. This invention overcomes the limitations of traditional mechanical threshold control, improves the intelligence level and operation and maintenance efficiency of solar-powered, direct-flexible buildings, and makes the temperature control of the temperature control equipment more precise, reducing energy consumption.

Description

Technical Field

[0001] The present invention relates to the field of new energy buildings, and in particular to a temperature control device control method and system based on a solar-storage direct-flexible building. Background Art

[0002] As an innovative building energy solution, the PV-storage, direct current, and flexible building integrates photovoltaic power generation, energy storage, DC power distribution, and flexible power consumption technologies to achieve self-sufficiency and efficient building energy utilization while ensuring harmonious interaction with the power grid. Environmental monitoring plays a crucial role in PV-storage, direct current, and flexible buildings. It not only provides key data support for building energy management but also helps improve energy efficiency and occupant comfort. However, in PV-storage, direct current, and flexible buildings, temperature control equipment is the primary energy consumer, and its set temperature has a significant impact on energy consumption and comfort. Current mainstream temperature control equipment still relies on traditional threshold trigger mechanisms, mechanically regulating based solely on ambient temperature and preset parameters. This makes it difficult to adapt to the complex and changing energy supply and demand scenarios of PV-storage, direct current, and flexible buildings, exposing problems such as insufficient intelligence and delayed dynamic response. This extensive control model not only limits the potential for collaborative optimization of building energy systems but also results in inaccurate temperature control by temperature control equipment, increasing energy consumption. Summary of the Invention

[0003] The purpose of the embodiments of the present invention is to provide a temperature control equipment control method and system based on a photovoltaic storage direct-flexible building, which breaks through the limitations of traditional mechanical threshold control, improves the intelligence level and operation and maintenance efficiency of the photovoltaic storage direct-flexible building, and makes the temperature control of the temperature control equipment more precise and reduces energy consumption.

[0004] To achieve the above objectives, an embodiment of the present invention provides a method for controlling temperature control equipment based on a solar-storage direct-flexible building, comprising:

[0005] Obtain environmental data, static parameters of the PV-storage-direct-flexible building, and status data of the energy system operation in the PV-storage-direct-flexible building;

[0006] constructing a building thermodynamics model according to the environmental data, the static parameters and the state data, and constructing a multi-objective optimization function according to the state data;

[0007] Predicting temperature change information of the indoor temperature within a preset time period in the future based on the building thermodynamic model, and determining the temperature adjustment timing of the temperature control device based on the temperature change information;

[0008] Solving the multi-objective optimization function according to preset constraints to obtain a temperature adjustment value of the temperature control device;

[0009] Parameters of the temperature control device are adjusted according to the temperature adjustment timing and the temperature adjustment value.

[0010] As an improvement to the above solution, the environmental data includes indoor temperature, indoor humidity, indoor carbon dioxide concentration, outdoor temperature, outdoor humidity and solar radiation intensity; the static parameters include building heat capacity, building surface area and building foundation heat transfer coefficient; and the status data includes the real-time power of the temperature control equipment, the remaining power percentage of the energy storage system, the grid input power and the transmission cost.

[0011] As an improvement to the above solution, constructing a building thermodynamic model based on the environmental data, the static parameters and the state data includes:

[0012] A building thermodynamic model is constructed with the building heat capacity and temperature change rate as output variables, and the building surface area, the building foundation heat transfer coefficient, the indoor temperature, the outdoor temperature and the real-time power of the temperature control device as input variables.

[0013] As an improvement to the above solution, the method of predicting temperature change information of the indoor temperature within a preset time period in the future based on the building thermodynamic model and determining the temperature adjustment timing of the temperature control device based on the temperature change information includes:

[0014] According to the building thermodynamic model, a preset iterative algorithm is used to numerically solve the evolution process of the indoor temperature within a preset time period in the future to obtain a number of intermediate variables;

[0015] Calculate the temperature change information of the indoor temperature over time based on several intermediate variables, the current indoor temperature and the preset time step;

[0016] Comparing each indoor temperature with the temperature threshold in ascending order of time, to select the first target indoor temperature greater than the temperature threshold from the temperature change information;

[0017] The temperature adjustment timing of the temperature control device is determined according to the critical time point corresponding to the target indoor temperature, the preset minimum time interval and the building thermal response time constant.

[0018] As an improvement to the above solution, the step of constructing a multi-objective optimization function based on the state data includes:

[0019] The multi-objective optimization function is to minimize the weighted sum of energy consumption cost and comfort loss value;

[0020] Among them, the weighted value corresponding to the energy consumption cost is a first weighting factor, and the energy consumption cost is calculated based on the grid input power and the transmission cost; the weighted value corresponding to the comfort loss value is a second weighting factor, and the comfort loss value is the predicted average discomfort integral value within a preset time period in the future.

[0021] As an improvement to the above solution, the process of obtaining the first weight factor and the second weight factor includes:

[0022] Acquire a comfort parameter that is strongly correlated with comfort from an environmental factor correlation matrix; wherein the environmental factor correlation matrix is ​​obtained by processing the environmental data;

[0023] Calculating a comfort sensitivity coefficient according to the comfort parameter and the corresponding proportional coefficient;

[0024] determining a first weight base value and a second weight base value based on the comfort sensitivity coefficient and the power transmission cost;

[0025] A first weighting factor is calculated according to the first weighting base value and a preset correction coefficient, and a second weighting factor is calculated according to the second weighting base value and the correction coefficient.

[0026] As an improvement to the above solution, the process of constructing the environmental factor association matrix includes:

[0027] Obtain reference environmental data corresponding to historical typical working conditions;

[0028] performing data processing on the reference environment data to obtain feature data;

[0029] Using the feature data and the XGBoost algorithm to train an equipment load forecasting model, and extracting feature importance weight values ​​from the trained equipment load forecasting model;

[0030] The feature importance weight values ​​are normalized to obtain an environmental factor association matrix containing several feature importance weight values.

[0031] As an improvement to the above solution, the constraint condition includes at least one of the following:

[0032] The set temperature of the temperature control device is between the preset minimum temperature and the preset maximum temperature;

[0033] The temperature change of the temperature control equipment per hour does not exceed the preset maximum change rate;

[0034] The remaining power percentage of the energy storage system is not less than the preset safety power percentage.

[0035] As an improvement to the above solution, the method further includes:

[0036] When it is detected that the indoor carbon dioxide concentration is greater than a preset concentration threshold, the opening of the fresh air valve of the temperature control device is adjusted to the maximum opening.

[0037] To achieve the above objectives, an embodiment of the present invention further provides a temperature control system for a solar-storage direct-flexible building, comprising:

[0038] The data acquisition module is used to obtain environmental data, static parameters of the PV-storage direct-flexible building, and status data of the energy system operation in the PV-storage direct-flexible building;

[0039] A building thermodynamics model construction module, configured to construct a building thermodynamics model according to the environmental data, the static parameters and the state data;

[0040] A multi-objective optimization function construction module, used for constructing a multi-objective optimization function according to the state data;

[0041] a temperature adjustment timing determination module, configured to predict temperature change information of the indoor temperature within a preset time period in the future based on the building thermodynamics model, and determine the temperature adjustment timing of the temperature control device based on the temperature change information;

[0042] a temperature adjustment value determination module, configured to solve the multi-objective optimization function according to preset constraints to obtain a temperature adjustment value for the temperature control device;

[0043] A parameter adjustment module is used to adjust the parameters of the temperature control device according to the temperature adjustment timing and the temperature adjustment value.

[0044] Compared with the prior art, the temperature control device control method and system based on the photovoltaic storage direct flexible building disclosed in the present invention first constructs a building thermodynamic model and a multi-objective optimization function based on environmental data, static parameters and state data of the photovoltaic storage direct flexible building, then determines the temperature adjustment timing of the temperature control device according to the building thermodynamic model, and solves the multi-objective optimization function according to preset constraints to obtain the temperature adjustment value of the temperature control device, and finally adjusts the parameters of the temperature control device according to the temperature adjustment timing and the temperature adjustment value. The present invention deeply integrates environmental, energy, and building information data to construct a decision-making system for intelligent learning, precise prediction, and multi-objective optimization, and implements closed-loop feedback control, breaking through the limitations of traditional mechanical threshold control, and achieving multiple goals of accurately predicting environmental changes, dynamically optimizing energy scheduling, reducing operating costs, improving user comfort, ensuring system safety and reliability, and deeply adapting to the characteristics of photovoltaic storage direct flexible buildings. It not only improves the intelligence level and operation and maintenance efficiency of photovoltaic storage direct flexible buildings, but also makes the temperature control of temperature control equipment more precise and reduces energy consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a flow chart of a temperature control device control method based on a solar-storage direct-flexible building provided by an embodiment of the present invention;

[0046] Figure 2is a flow chart of generating weight factors provided by an embodiment of the present invention;

[0047] Figure 3 is a flow chart of generating an environmental factor association matrix provided by an embodiment of the present invention;

[0048] Figure 4 This is a flow chart of determining the timing of temperature adjustment provided by an embodiment of the present invention;

[0049] Figure 5 This is a structural block diagram of a temperature control device control system based on a solar-storage direct-flexible building provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0051] Through energy interconnection and data interaction, PV-storage direct-flexible buildings form a synergistic smart energy ecosystem with photovoltaic (PV), energy storage systems, and the grid (collectively referred to as the energy system). BIPV (Building Integrated Photovoltaic) panels directly generate electricity to power its own electrical devices (such as temperature control and lighting), reducing reliance on the grid. PV power is prioritized to meet building load demands, with excess power stored in the energy storage system or fed back into the grid (requiring bidirectional metering). The energy storage system stores excess PV power or off-peak grid electricity, releasing it during periods of insufficient PV demand, peak demand, or high grid prices to smooth fluctuations in building electricity consumption (for example, using stored power for air conditioning at night). As an energy storage application, PV-storage direct-flexible buildings achieve peak load shifting and valley filling through energy storage regulation, alleviating grid pressure. When PV+storage cannot meet demand, the building purchases electricity from the grid to maintain normal operation. BIPV direct-flexible buildings, through flexible loads such as temperature control devices, respond to grid dispatch (for example, reducing power consumption during peak load periods), becoming flexible regulatory elements of the grid. The temperature control device described in the embodiment of the present invention can be an air-conditioning system, such as a central air-conditioning system, a household split air-conditioning system, a terminal heat dissipation / refrigeration device, etc.

[0052] See also Figure 1 , Figure 1 This is a flow chart of a method for controlling temperature control equipment based on a solar-storage direct-flexible building, provided by an embodiment of the present invention. The method includes:

[0053] S1. Obtain environmental data, static parameters of the PV-storage-direct-flexible building, and status data of the energy system operation in the PV-storage-direct-flexible building;

[0054] S2. constructing a building thermodynamics model according to the environmental data, the static parameters and the state data, and constructing a multi-objective optimization function according to the state data;

[0055] S3. Predicting temperature change information of the indoor temperature within a preset time period in the future based on the building thermodynamic model, and determining a temperature adjustment timing of the temperature control device based on the temperature change information;

[0056] S4. Solving the multi-objective optimization function according to preset constraints to obtain a temperature adjustment value of the temperature control device;

[0057] S5. Adjust parameters of the temperature control device according to the temperature adjustment timing and the temperature adjustment value.

[0058] In an embodiment of the present invention, by deeply integrating environmental, energy, and building information data, a decision-making system with intelligent learning, precise prediction, and multi-objective optimization is constructed, and closed-loop feedback control is implemented, which breaks through the limitations of traditional mechanical threshold control and achieves multiple goals of accurately predicting environmental changes, dynamically optimizing energy scheduling, reducing operating costs, improving user comfort, ensuring system safety and reliability, and deeply adapting to the characteristics of photovoltaic, storage, direct and flexible systems. It not only improves the intelligence level and operation and maintenance efficiency of photovoltaic, storage, direct and flexible buildings, but also makes the temperature control of temperature control equipment more precise and reduces energy consumption.

[0059] Specifically, in step S1, the environmental data includes indoor temperature, indoor humidity, indoor carbon dioxide concentration, outdoor temperature, outdoor humidity and solar radiation intensity; the static parameters include building heat capacity, building surface area and building foundation heat transfer coefficient; the status data includes the real-time power of the temperature control equipment, the remaining power percentage of the energy storage system, the grid input power and the transmission cost.

[0060] For example, the indoor and outdoor temperatures can be collected using temperature sensors deployed indoors and outdoors, respectively. The indoor and outdoor humidity can also be collected using humidity sensors deployed indoors and outdoors, respectively. The temperature and humidity sensors can be integrated or separate. Indoor carbon dioxide concentration is also collected using a carbon dioxide sensor, such as an infrared absorption CO2 sensor. Solar radiation intensity can be calculated using a photoresistor and a calibration algorithm. The calculation process can be referenced in the prior art and will not be detailed here.

[0061] For example, the building heat capacity refers to the amount of heat required for each set temperature increase or decrease (e.g., 1°C) when the building envelope and internal components absorb or release heat. This reflects the building's ability to store heat and is directly related to the specific heat capacity and volume of the materials. This can be determined through on-site testing, such as by measuring the building's thermal time constant using the step response method and then inferring the heat capacity. The building surface area refers to the total area of ​​the building's envelope (e.g., exterior walls, roof, windows, floors, etc.) that is in direct contact with the external environment. The larger the surface area, the greater the heat exchange area between the building and the outside world, and the greater the amount of heat transferred through the envelope (e.g., by conduction, radiation, and convection), directly affecting the building's heating and cooling loads. This can be directly extracted from BIM models or architectural drawings. The building's basic heat transfer coefficient refers to the amount of heat transferred through a unit area of ​​the building envelope per unit time under a unit temperature difference (e.g., 1°C). It reflects the building's thermal insulation performance. A smaller value for the basic heat transfer coefficient indicates a greater thermal resistance, weaker heat transfer capacity, and better insulation. This value can be determined through a lookup table (structural material thermal resistance) or on-site heat flow meter measurements.

[0062] For example, the power of the temperature control device can be directly measured. The remaining charge percentage of the energy storage system can be calculated in real time using the current integration method or provided directly by the BMS. Grid input power is directly measured by grid-side smart meters. Transmission costs can be real-time electricity prices, obtained online, or input by staff.

[0063] Specifically, in step S2, the building thermodynamic model is constructed based on the environmental data, the static parameters and the state data, including: taking the building heat capacity and temperature change rate as output variables, and the building surface area, the building foundation heat transfer coefficient, the indoor temperature, the outdoor temperature and the real-time power of the temperature control device as input variables to construct the building thermodynamic model.

[0064] Exemplarily, the building thermodynamic model satisfies the following formula:

[0065] (1);

[0066] in, is the building heat capacity; Indicates the rate of temperature change, which is the temperature About time The first derivative of represents the instantaneous slope of temperature change, which is physically understood as "the trend of temperature change at this moment", such as When , it means that under the current trend, the indoor temperature will rise by 0.5℃ per minute; is the heat transfer coefficient of the building foundation; is the building surface area; is the outdoor temperature; is the indoor temperature; The real-time power of the temperature control device.

[0067] Specifically, in step S2, a multi-objective optimization function is constructed based on the state data, including: minimizing the weighted sum of energy consumption cost and comfort loss value as the multi-objective optimization function; wherein the weighted value corresponding to the energy consumption cost is a first weight factor, and the energy consumption cost is calculated based on the grid input power and the transmission cost; the weighted value corresponding to the comfort loss value is a second weight factor, and the comfort loss value is the predicted average discomfort integral value within a preset time period in the future.

[0068] Exemplarily, the multi-objective optimization function satisfies the following formula:

[0069] (2);

[0070] in, is the first weight factor; is the energy cost, ,in, is the grid input power, is the transmission cost; is the second weight factor; is the comfort loss value, ,in, It is the average discomfort voting value predicted within a preset future time period, which can be calculated based on the standard thermal comfort model recorded in the existing technology (such as the ISO7730 standard).

[0071] See also Figure 2 , Figure 2 is a flow chart of generating a weight factor according to an embodiment of the present invention, wherein the first weight factor and the second weighting factor The acquisition process includes:

[0072] S21. Obtaining a comfort parameter that is strongly correlated with comfort from an environmental factor correlation matrix; wherein the environmental factor correlation matrix is ​​obtained by processing the environmental data;

[0073] S22, calculating a comfort sensitivity coefficient according to the comfort parameter and a corresponding proportional coefficient;

[0074] S23. Determine a first weight base value and a second weight base value based on the comfort sensitivity coefficient and the power transmission cost;

[0075] S24: Calculate a first weight factor according to the first weight base value and a preset correction coefficient, and calculate a second weight factor according to the second weight base value and the correction coefficient.

[0076] Exemplarily, in step S21 , an element that is strongly correlated with comfort is selected from the environmental factor association matrix M, where the environmental factor association matrix M includes several feature importance weight values.

[0077] For example, in step S22, a comfort sensitivity coefficient is constructed based on these weight values ​​and corresponding proportional coefficients, satisfying the following formula:

[0078] (3);

[0079] in, 、 、 It is a preset proportional coefficient used to characterize the differential effects of indoor humidity, indoor carbon dioxide concentration, outdoor temperature and outdoor humidity on comfort; 、 、 and is the comfort parameter (i.e., the feature importance weight value), where represents the feature importance weight value corresponding to indoor humidity; represents the feature importance weight value corresponding to the indoor carbon dioxide concentration; represents the feature importance weight value corresponding to the outdoor temperature; represents the feature importance weight value corresponding to outdoor humidity; Normalized to the interval [0,1]: .

[0080] For example, in step S23, based on the comfort sensitivity coefficient and transmission costs , determine the first weight base value and the second weighted base value , satisfying the following formula:

[0081] (4);

[0082] (5);

[0083] in, and The lowest and highest values ​​of the day's electricity price. When the lowest and highest values ​​are equal, ; It is the electricity price influencing factor, which can be set according to the empirical value.

[0084] For example, in steps S24-S25, according to the first weight base value and preset correction factors Calculate the first weight factor, that is, satisfy ; and according to the second weight base value and the correction factor Calculate the second weight factor, that is, satisfy .

[0085] For example, the correction factor is given by The remaining power percentage of the energy storage system (represented by SOC) is determined and is inversely proportional to the remaining power percentage of the energy storage system. For example, when SOC≤20%, =1.2; when 20%< When ≤80%, =1.0; when SOC>80%, =0.8. Furthermore, when the SOC is too low (≤20%), the first weight factor is increased. , prioritize power conservation; when SOC is sufficient (>80%), increase the second weight factor , giving priority to improving comfort.

[0086] In an embodiment of the present invention, comfort parameters such as humidity and carbon dioxide concentration are extracted from the environmental factor association matrix, and the weight of the influence of each parameter on thermal comfort is quantified by the proportional coefficient. Compared with the traditional "temperature control with only a single parameter", it can accurately adapt to the environmental comfort requirements and enhance the physical experience. The weight factor is determined based on the transmission cost and the comfort sensitivity coefficient, and the electricity price and energy storage status are linked to optimize energy economy. In addition, the correction coefficient is dynamically adjusted with the energy storage SOC. When the energy storage is insufficient, the energy consumption weight is forced to increase, and unnecessary temperature control loads are reduced to ensure that the energy storage can supply power to key loads such as lighting; when the energy storage is sufficient, the comfort weight is increased, and the system can adjust the temperature more flexibly, and even sacrifice some energy consumption in exchange for a better comfort experience.

[0087] Specifically, in step S21, see Figure 3 , Figure 3 : is a flow chart of generating an environmental factor association matrix provided by an embodiment of the present invention. The process of constructing the environmental factor association matrix includes:

[0088] S211. Obtain reference environmental data corresponding to a historical typical working condition day;

[0089] S212, processing the reference environment data to obtain feature data;

[0090] S213, using the feature data and the XGBoost algorithm to train an equipment load prediction model, and extracting feature importance weight values ​​from the trained equipment load prediction model;

[0091] S214 , normalizing the feature importance weight values ​​to obtain an environmental factor association matrix containing a plurality of feature importance weight values.

[0092] For example, in step S211, the method for screening historical typical working condition days includes the following steps:

[0093] a. Filter by season:

[0094] Typical summer day: outdoor temperature greater than 30°C and solar radiation intensity greater than 800W / m²;

[0095] Typical winter day: outdoor temperature is less than 5°C and solar radiation intensity is less than 300W / m²;

[0096] Typical days in the transition season: outdoor temperature is between 10-25°C and solar radiation intensity is between 400-600W / m²;

[0097] It should be noted that the above typical day values ​​are only examples and can be adjusted as needed in actual applications.

[0098] b. Calculate the daily load similarity index , the calculation formula is:

[0099] (6);

[0100] in, The highest outdoor temperature of the day, is the historical average high temperature value for the same season. is the maximum radiation intensity of the day, is the historical average radiation value for the same season.

[0101] c. Screening conditions:

[0102] Select >0.85 and load fluctuation standard deviation The date with a load fluctuation of less than 15% is regarded as a historical typical working condition day, and the load fluctuation standard deviation is The calculation formula satisfies:

[0103] (7);

[0104] in, =288 (number of sampling points per day, 5 minutes / point), For the The load value of each sampling point, is the daily average load value.

[0105] It should be noted that step a is a preliminary classification by season, and step c is a further screening within each season. Step a and step c are in a progressive relationship. Finally, each season will screen out historical typical working condition days. After step c screening, each season may have multiple typical days, and these typical days will be used as training data.

[0106] For example, in step S212, the training data is first cleaned (missing value processing, outlier filtering, etc.) and feature constructed. During the feature construction process, the following two types of feature data can be extracted:

[0107] 1) Original features, which are the environmental data corresponding to historical typical working conditions, including indoor temperature , indoor humidity , indoor carbon dioxide concentration , outdoor temperature , outdoor humidity , solar radiation intensity .

[0108] 2) Derived features, including:

[0109] ① Time characteristics: including hours, day type (weekday / weekend), and season code;

[0110] ② Hysteresis characteristics: temperature control equipment load value in the previous hour and the previous 24 hours;

[0111] ③Combined characteristics: solar radiation × outdoor temperature (representing solar heat gain).

[0112] Then perform data standardization on the above features, such as for continuous features ( 、 ) for Z-score standardization and One-Hot encoding of categorical features (such as day type).

[0113] For example, in step S213, XGBoost, an optimized version of the gradient boosting tree, is highly efficient and interpretable, making it ideal for multivariate, nonlinear problems such as temperature control equipment load forecasting. The standardized dataset is divided into a training set and a test set in a ratio (e.g., 7:3). The training set is used for model learning, and the test set is used to evaluate model performance. Core XGBoost parameters, such as the number of decision trees (n_estimators), maximum tree depth (max_depth), learning rate (learning_rate), and regularization coefficients (e.g., gamma and lambda), are set. Initially, the default values ​​can be used. Parameters can then be optimized through tuning (e.g., grid search) to improve model accuracy. Using the training set data, the XGBoost algorithm iteratively constructs multiple decision trees. Each tree is trained based on the prediction error (residual) of the previous tree. A loss function (e.g., squared loss) is minimized using gradient descent, gradually optimizing the model's ability to predict equipment load. Model performance is verified using the test set data, and prediction accuracy is assessed using the mean absolute error (MAE) metric. If accuracy is insufficient, parameters are adjusted and retrained until the model meets the prediction requirements.

[0114] After the equipment load forecasting model is trained, the influence of each input variable (feature) on the predicted result (equipment load)—that is, the feature importance weight—is automatically calculated. This process calculates the feature importance score by counting the number of times each feature serves as a split node in all decision trees and the amount of loss reduction resulting from such splits. A higher score indicates a greater impact on equipment load. Corresponding weights are then assigned to each of the six input variables.

[0115] For example, in step S214, the feature importance weights are normalized to generate an environmental factor correlation matrix , whose matrix elements satisfy:

[0116] (8);

[0117] in, represents the feature importance weight value corresponding to indoor temperature; Indicates the feature importance weight value corresponding to the solar radiation intensity; The sum of all matrix elements in it is 1.

[0118] In an embodiment of the present invention, by screening historical typical working days, it is ensured that the data used for training is a valid sample that highly matches the actual equipment load characteristics, avoiding the interference of atypical data (such as extreme abnormal weather and sudden load fluctuation days) on the model, and improving data quality. Indoor and outdoor environmental parameters are used as input features to fully capture the environmental factors that affect the equipment load, so that the model can learn association rules that are more in line with actual scenarios. The advantages of the XGBoost algorithm are that it can effectively handle nonlinear relationships and feature interactions (such as the impact of high outdoor temperature and solar radiation superposition on equipment load) by integrating multiple decision trees, iterative optimization based on residuals, and adding regularization mechanisms. Compared with a single model (such as linear regression), it has higher prediction accuracy.

[0119] Furthermore, after the equipment load forecasting model is trained, the environmental factor correlation matrix is ​​optimized through a dynamic update mechanism. The optimization process includes: calculating the average absolute error (MAE) between the equipment load forecast value and the measured value every day, and starting the environmental factor correlation matrix when the average absolute error for consecutive preset days is greater than the error threshold. In the update process, the incremental learning algorithm is used to load the data within the preset time period and update the equipment load forecasting model parameters. After the model is updated, the accuracy of the new model is tested on the validation set. When the relative improvement in accuracy is greater than the preset percentage, a new environmental factor association matrix is ​​generated and deployed based on the feature importance weight value of the new model (i.e., re-extraction). , otherwise it falls back to the old matrix corresponding to the original prediction model.

[0120] For example, the mean absolute error (MAE) is calculated as:

[0121] (9);

[0122] in, =288, which is the number of data points per day (24 hours × 12 points / hour = 288 points). is the equipment load prediction value, which is generated by the equipment load prediction model based on real-time environmental data. ,in, is the model function of the trained equipment load prediction model, The actual value of the equipment load is calculated based on the real-time power of the temperature control equipment. For example, the actual value of the equipment load is = real-time power × conversion coefficient. The conversion coefficient can range from 0.9 to 0.95, with a typical value of 0.92.

[0123] For example, when using an incremental learning algorithm to update the parameters of a device load forecasting model, environmental data from the last 7 or 10 days (which can be set based on scenario requirements) can be collected. After obtaining environmental feature data using the same data processing method, the currently used device load forecasting model is called from the system and all its historical parameters, including the decision tree structure, node splitting threshold, leaf node weights, and regularization coefficients, are loaded as the initial parameters for incremental learning. The incremental learning training strategy is then configured: ① Lower the learning rate to a lower value than that used in the initial training phase (e.g., if the original learning rate is 0.1, set it to 0.01-0.05 during incremental learning) to ensure smoother parameter adjustments due to new data and reduce the impact on historically valid parameters; ② Control the number of training rounds, using only a small number of rounds (e.g., 1-3 rounds) with new data to prevent the model from overfitting to short-term fluctuations in the new data; ③ Introduce regularization protection by adding a parameter stability regularization term to the incremental training objective function. This penalizes important parameters in the original model (e.g., node parameters corresponding to environmental factors with high historical feature importance weights), limiting their update range and ensuring that core correlation patterns are not disrupted.

[0124] For example, preprocessed environmental feature data is fed into the model, and incremental training is initiated based on the aforementioned initial parameters and training strategy. For tree models, incremental learning builds on the existing decision tree and, based on the error in the new data (the deviation between predicted and measured load), selectively adds a small number of subtrees or adjusts the node splitting logic of existing trees, rather than rebuilding the entire tree. During training, the model prioritizes learning new associations reflected in the new data, such as the impact of a recent surge in outdoor humidity on load. At the same time, regularization constraints are used to maintain stable associations verified in historical data, such as the strong correlation between indoor temperature and load. Ultimately, model parameters (such as tree structure, node weights, and feature splitting gains) are gradually updated, preserving existing knowledge while incorporating new data. After incremental training completes, the updated model parameters are saved and used as the basis for the next round of prediction or further incremental learning. At this point, the model is adaptable to the associations between environmental factors and load over a preset time period and can extract new feature importance weights.

[0125] In an embodiment of the present invention, by dynamically updating the model, the model can be continuously optimized in a dynamically changing environment, which not only avoids the high cost of full retraining, but also maintains the accuracy of equipment load prediction through progressive learning, providing reliable model support for the dynamic update of the environmental factor correlation matrix.

[0126] Specifically, in step S3, see Figure 4 , Figure 4 This is a flow chart of determining the temperature adjustment timing provided by an embodiment of the present invention, wherein step S3 specifically includes:

[0127] S31. According to the building thermodynamics model, a preset iterative algorithm is used to numerically solve the evolution process of the indoor temperature within a preset time period in the future to obtain a plurality of intermediate variables;

[0128] S32, performing calculations based on a number of intermediate variables, the current indoor temperature, and a preset time step to obtain temperature change information of the indoor temperature over time;

[0129] S33, comparing each indoor temperature with the temperature threshold in ascending order of time, to select the first target indoor temperature greater than the temperature threshold from the temperature change information;

[0130] S34. Determine a temperature adjustment timing of the temperature control device according to a critical time point corresponding to the target indoor temperature, a preset minimum time interval, and a building thermal response time constant.

[0131] For example, in step S31, according to the building thermodynamic model recorded in the above formula (1), the evolution process of the indoor temperature within the future preset time period is numerically solved by an iterative algorithm (such as the Runge-Kutta method) to obtain several intermediate variables.

[0132] First, initialize some data, such as initializing the current indoor temperature (t0), current outdoor temperature (t0), current air conditioning power (t0); and set the time step (For example, 1 minute); then, based on the building thermodynamics model, the Runge-Kutta method is used to iteratively calculate the evolution of the indoor temperature within the preset future time period to numerically solve several intermediate variables that satisfy the following formula:

[0133] (10);

[0134] (11);

[0135] (12);

[0136] (13);

[0137] in, ~ is an intermediate variable.

[0138] It should be noted that the core of the Runge-Kutta method (RK method) is to break down the continuous differential equation into discrete calculations step by step, and use known quantities to deduce the temperature at future moments. For formula (1), the processing logic of the RK method is: is the instantaneous rate of change, but the RK method uses the known quantity at the current moment ( 、 、 etc.), calculate the time step The average rate of change within the time period is used to iterate the temperature at the next moment. . Calculate first ~ These four intermediate quantities are essentially different and Combination, simulate instantaneous rate of change exist Different trends over time.

[0139] For example, in step S32, calculations are performed based on several intermediate variables, the current indoor temperature, and a preset time step to obtain temperature change information of the indoor temperature over time, which satisfies the following formula:

[0140] (14).

[0141] It should be noted that formula (14) uses ~ Weighted average, we get The average temperature change in the room is actually a discrete calculation instead of a continuous one. Steps S31 to S32 are repeated continuously until the preset time is covered, and the temperature change information of the indoor temperature over time is obtained. At this time, multiple , arranged in ascending order of time, and integrated to obtain the temperature change set , a total of The indoor temperature corresponding to the future time period.

[0142] For example, in step S33, each indoor temperature is compared with the temperature threshold in ascending order of time, that is, from Filter out the first target indoor temperature that is greater than the temperature threshold, and then select the critical time point corresponding to the target indoor temperature. This process satisfies the following formula:

[0143] (15);

[0144] in, is the critical time point, the first point to satisfy Time point , that is, the time point corresponding to the target indoor temperature; Set the upper limit of temperature for users (default is 26℃). is the temperature value at each time point within the future preset time period calculated by step S32. The value is taken from the current moment, with a time step Increment until the preset time in the future ends.

[0145] For example, in step S34, the temperature adjustment timing of the temperature control device is determined according to the critical time point corresponding to the target indoor temperature, the preset minimum time interval and the building thermal response time constant. , this process satisfies the following formula:

[0146] (16);

[0147] in, is the building thermal response time constant, when there is no temperature control equipment power (i.e. =0) and the outdoor temperature is constant ( is a constant), the solution of this equation is in the form of exponential decay, and the time constant , but in practical applications, is not 0, and changes over time, so It only represents the thermal inertia time constant of the building itself and is used to estimate how quickly the building responds to temperature changes. is the preset coefficient; represents the required lead time for adjustment based on the building’s thermal inertia; is the preset minimum time interval; It refers to taking and The maximum value in .

[0148] In an embodiment of the present invention, a building thermodynamic model combined with an iterative algorithm numerically solves the evolution of indoor temperature over a preset future timeframe, generating intermediate variables reflecting temperature trends and detailed temperature change information. This process, based on the physical laws of building heat transfer, is more scientific than simple empirical predictions and allows for a more accurate understanding of the dynamic changes in indoor temperature over time, providing a precise basis for subsequent temperature control. Furthermore, by determining the timing of temperature adjustments based on the target indoor temperature, a preset minimum time interval, and the building's thermal response time constant, the adjustment timing is fully considered, taking into account the building's thermal inertia (i.e., the temperature change delay characteristic reflected by the thermal response time constant) and the minimum interval requirements for device operation. This avoids energy waste caused by frequent startup and shutdown of temperature control devices and prevents excessive deviation of the indoor temperature from the comfort range due to untimely adjustments. This allows the device to intervene and control at the most appropriate time, improving energy efficiency.

[0149] Specifically, in step S4, the multi-objective optimization function is solved according to preset constraints to obtain the temperature adjustment value of the temperature control device; wherein the constraints include at least one of the following:

[0150] 1) The set temperature of the temperature control equipment is between the preset minimum and maximum temperatures;

[0151] 2) The temperature change of the temperature control equipment per hour does not exceed the preset maximum change rate;

[0152] 3) The remaining power percentage of the energy storage system is not less than the preset safety power percentage.

[0153] Exemplarily, the process of solving the multi-objective optimization function (i.e., the above formula (2)) according to the constraints includes:

[0154] a. Define optimization variables ,in To set the adjustment value, that is, how much the temperature of the temperature control device should be adjusted, The real-time power of the equipment in each time period within a preset time period in the future (discrete representation);

[0155] b. Multi-objective optimization function Transformed into a quadratic programming sub-problem, in each iteration, the multi-objective function is transformed into a quadratic programming sub-problem at the current point It is approximately a quadratic function, and the constraints are approximately linear; since the multi-objective optimization function It is nonlinear and very complex to solve directly. Therefore, in each iteration, the objective function is approximated as a quadratic function (parabolic shape) and the constraints are approximated as straight lines. This allows the mature quadratic programming algorithm to be used to solve it.

[0156] c. Solve the quadratic programming subproblem and obtain the search direction ; Use quadratic programming algorithms (such as interior point method, active set method) to calculate the search direction under "approximate quadratic function + linear constraints" This step is a mathematical derivation to get the goal of "in which direction should the temperature and power be adjusted to make the objective function smaller";

[0157] d. Determine step size through line search ,renew After determining the adjustment direction, we need to determine the step size and find the optimal step size through line search (such as the golden section method or the Armijo criterion) to ensure that the objective function is indeed reduced after adjustment.

[0158] e. Repeat steps bd until the convergence condition is met, such as the gradient is less than the threshold or the maximum number of iterations is reached, indicating that the optimal solution has been approached;

[0159] f. After the iteration, optimize the variables The first element of It is the optimal temperature adjustment value. At this time, the optimal temperature adjustment value is output. .

[0160] In the embodiment of the present invention, through constraints and multi-objective optimization process, the temperature control strategy is upgraded from "temperature adjustment based on experience" to "data-driven precise control", achieving the triple balance of energy saving, comfort and safety. This is not only the technical implementation of flexible electricity use in solar-storage-direct-flexible buildings, but also a key step for smart buildings to move from automation to intelligence.

[0161] Specifically, in step S5, parameters of the temperature control device are adjusted according to the temperature adjustment timing and the temperature adjustment value.

[0162] For example, when the temperature adjustment timing is obtained and temperature adjustment value Afterwards, used in Send the information to the device system at any time Instructions for adjusting the set temperature.

[0163] Furthermore, the method further includes: when it is detected that the indoor carbon dioxide concentration is greater than a preset concentration threshold, adjusting the opening of the fresh air valve of the temperature control device to a maximum opening.

[0164] For example, when it is detected that the carbon dioxide concentration is greater than a preset concentration threshold, the prediction process is skipped and the opening of the fresh air valve in the fresh air system is directly adjusted to 100% to ensure indoor air quality and prevent health risks.

[0165] Furthermore, when the execution of the instruction fails, a multi-level response mechanism is activated, which includes one or more of instruction retransmission, switching of communication channels and degradation control mode. The degradation control mode is: a preset fixed temperature setting value is used to replace the output of the optimization decision module. When the optimization process fails (that is, the calculation results of the above steps S1 to S5 cannot be applied), the preset fixed temperature value is used to maintain basic operation to prevent system paralysis from causing environmental out of control and achieve fault mitigation.

[0166] Compared with the existing technology, the temperature control equipment control method based on photovoltaic storage direct and flexible buildings disclosed in the present invention deeply integrates environmental, energy, and building information data to construct a decision-making system for intelligent learning, precise prediction, and multi-objective optimization, and implements closed-loop feedback control, breaking through the limitations of traditional mechanical threshold control and achieving multiple goals of accurately predicting environmental changes, dynamically optimizing energy scheduling, reducing operating costs, improving user comfort, ensuring system safety and reliability, and deeply adapting to photovoltaic storage direct and flexible characteristics. It not only improves the intelligence level and operation and maintenance efficiency of photovoltaic storage direct and flexible buildings, but also makes the temperature control of temperature control equipment more precise and reduces energy consumption.

[0167] See also Figure 5 , Figure 51 is a structural block diagram of a temperature control device control system 100 based on a solar-storage, direct-flexible building provided by an embodiment of the present invention. The temperature control device control system 100 based on a solar-storage, direct-flexible building includes:

[0168] The data acquisition module 11 is used to obtain environmental data, static parameters of the PV-storage direct-flexible building, and status data of the energy system operation in the PV-storage direct-flexible building;

[0169] A building thermodynamics model construction module 12, configured to construct a building thermodynamics model according to the environmental data, the static parameters and the state data;

[0170] A multi-objective optimization function construction module 13 is used to construct a multi-objective optimization function according to the state data;

[0171] a temperature adjustment timing determination module 14 for predicting temperature change information of the indoor temperature within a preset time period in the future based on the building thermodynamics model, and determining a temperature adjustment timing of the temperature control device based on the temperature change information;

[0172] a temperature adjustment value determination module 15, configured to solve the multi-objective optimization function according to preset constraints to obtain a temperature adjustment value for the temperature control device;

[0173] The parameter adjustment module 16 is configured to adjust the parameters of the temperature control device according to the temperature adjustment timing and the temperature adjustment value.

[0174] Specifically, the parameter adjustment module 16 is further configured to adjust the opening of the fresh air valve of the temperature control device to a maximum opening when it is detected that the indoor carbon dioxide concentration is greater than a preset concentration threshold.

[0175] It is worth noting that the working process of each module in the above-mentioned temperature control equipment control system 100 based on photovoltaic storage direct flexible building can refer to the working process of the temperature control equipment control method based on photovoltaic storage direct flexible building described in the above embodiment, and will not be repeated here.

[0176] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A temperature control method for a solar-storage direct-flexible building, characterized in that: include: Acquire environmental data, static parameters of the PV-storage-direct-flexible building, and status data of the energy system operation in the PV-storage-direct-flexible building; wherein the status data includes the real-time power of the temperature control equipment, the remaining power percentage of the energy storage system, the grid input power, and the transmission cost; constructing a building thermodynamics model according to the environmental data, the static parameters and the state data, and constructing a multi-objective optimization function according to the state data; Predicting temperature change information of the indoor temperature within a preset time period in the future based on the building thermodynamic model, and determining the temperature adjustment timing of the temperature control device based on the temperature change information; Solving the multi-objective optimization function according to preset constraints to obtain a temperature adjustment value of the temperature control device; Adjusting parameters of the temperature control device according to the temperature adjustment timing and the temperature adjustment value; Wherein, constructing a multi-objective optimization function according to the state data includes: The multi-objective optimization function is to minimize the weighted sum of energy consumption cost and comfort loss value; The weighted value corresponding to the energy consumption cost is a first weighting factor, and the energy consumption cost is calculated based on the grid input power and the transmission cost; the weighted value corresponding to the comfort loss value is a second weighting factor, and the comfort loss value is the average discomfort integral value predicted within a preset time period in the future; The process of obtaining the first weight factor and the second weight factor includes: Acquire a comfort parameter that is strongly correlated with comfort from an environmental factor correlation matrix; wherein the environmental factor correlation matrix is ​​obtained by processing the environmental data; Calculating a comfort sensitivity coefficient according to the comfort parameter and the corresponding proportional coefficient; determining a first weight base value and a second weight base value based on the comfort sensitivity coefficient and the power transmission cost; A first weighting factor is calculated according to the first weighting base value and a preset correction coefficient, and a second weighting factor is calculated according to the second weighting base value and the correction coefficient.

2. The temperature control device control method based on the solar-storage direct-flexible building according to claim 1 is characterized in that: The environmental data include indoor temperature, indoor humidity, indoor carbon dioxide concentration, outdoor temperature, outdoor humidity and solar radiation intensity; the static parameters include building heat capacity, building surface area and building foundation heat transfer coefficient.

3. The temperature control device control method based on the solar-storage direct-flexible building according to claim 2 is characterized in that: The step of constructing a building thermodynamics model according to the environmental data, the static parameters, and the state data includes: A building thermodynamic model is constructed with the building heat capacity and temperature change rate as output variables, and the building surface area, the building foundation heat transfer coefficient, the indoor temperature, the outdoor temperature and the real-time power of the temperature control device as input variables.

4. The temperature control device control method based on the solar-storage direct-flexible building according to claim 3 is characterized in that: The method of predicting temperature change information of the indoor temperature within a preset time period in the future according to the building thermodynamics model, and determining the temperature adjustment timing of the temperature control device according to the temperature change information, includes: According to the building thermodynamic model, a preset iterative algorithm is used to numerically solve the evolution process of the indoor temperature within a preset time period in the future to obtain a number of intermediate variables; Calculate the temperature change information of the indoor temperature over time based on several intermediate variables, the current indoor temperature and the preset time step; Comparing each indoor temperature with the temperature threshold in ascending order of time, to select the first target indoor temperature greater than the temperature threshold from the temperature change information; The temperature adjustment timing of the temperature control device is determined according to the critical time point corresponding to the target indoor temperature, the preset minimum time interval and the building thermal response time constant.

5. The temperature control device control method based on the solar-storage direct-flexible building according to claim 1 is characterized in that: The process of constructing the environmental factor association matrix includes: Obtain reference environmental data corresponding to historical typical working conditions; performing data processing on the reference environment data to obtain feature data; Using the feature data and the XGBoost algorithm to train an equipment load forecasting model, and extracting feature importance weight values ​​from the trained equipment load forecasting model; The feature importance weight values ​​are normalized to obtain an environmental factor association matrix containing several comfort parameters.

6. The temperature control device control method based on the solar-storage direct-flexible building according to claim 2 is characterized in that: The constraints include at least one of the following: The set temperature of the temperature control device is between the preset minimum temperature and the preset maximum temperature; The temperature change of the temperature control equipment per hour does not exceed the preset maximum change rate; The remaining power percentage of the energy storage system is not less than the preset safety power percentage.

7. The temperature control device control method based on the solar-storage direct-flexible building according to claim 2 is characterized in that: The method further comprises: When it is detected that the indoor carbon dioxide concentration is greater than a preset concentration threshold, the opening of the fresh air valve of the temperature control device is adjusted to the maximum opening.

8. A temperature control system based on a solar-storage direct-flexible building, characterized in that: include: A data acquisition module is used to obtain environmental data, static parameters of the PV-storage-direct-flexible building, and status data of the energy system operation in the PV-storage-direct-flexible building; wherein the status data includes the real-time power of the temperature control device, the remaining power percentage of the energy storage system, the grid input power, and the transmission cost; A building thermodynamics model construction module, configured to construct a building thermodynamics model according to the environmental data, the static parameters and the state data; A multi-objective optimization function construction module, used for constructing a multi-objective optimization function according to the state data; a temperature adjustment timing determination module, configured to predict temperature change information of the indoor temperature within a preset time period in the future based on the building thermodynamics model, and determine the temperature adjustment timing of the temperature control device based on the temperature change information; a temperature adjustment value determination module, configured to solve the multi-objective optimization function according to preset constraints to obtain a temperature adjustment value for the temperature control device; a parameter adjustment module, configured to adjust parameters of the temperature control device according to the temperature adjustment timing and the temperature adjustment value; The multi-objective optimization function construction module is specifically configured to: minimize the weighted sum of energy consumption cost and comfort loss value as the multi-objective optimization function; the weighted value corresponding to the energy consumption cost is a first weighting factor, and the energy consumption cost is calculated based on the grid input power and the transmission cost; the weighted value corresponding to the comfort loss value is a second weighting factor, and the comfort loss value is a predicted average discomfort integral value within a preset time period in the future; The process of obtaining the first weight factor and the second weight factor includes: Acquire a comfort parameter that is strongly correlated with comfort from an environmental factor correlation matrix; wherein the environmental factor correlation matrix is ​​obtained by processing the environmental data; Calculating a comfort sensitivity coefficient according to the comfort parameter and the corresponding proportional coefficient; determining a first weight base value and a second weight base value based on the comfort sensitivity coefficient and the power transmission cost; A first weighting factor is calculated according to the first weighting base value and a preset correction coefficient, and a second weighting factor is calculated according to the second weighting base value and the correction coefficient.

Images