Big data information processing method and system based on deep learning

Abstract

The invention provides a big data information processing method and system based on deep learning. According to the method, an energy distribution model is generated by combining a building information model with infrared thermal imaging data, and a dynamic thermal balance equation set is constructed according to the model so as to generate three-dimensional temperature field reconstruction data. Then, thermal resistance parameters in the energy distribution model, heat flow abnormal area coordinates in the three-dimensional temperature field reconstruction data and air conditioning unit current data are analyzed through a deep learning algorithm, and space-time coupling characteristics of equipment power consumption and thermal environment parameters are extracted; then, a priority regulation and control chain of the temperature control equipment is formulated, and multi-stage linkage rules of the fresh air exchange rate, illumination dimming and cold release of the cold storage tank are covered; finally, the chilled water pump phase difference, the air conditioner group operation mode and the emergency lighting power limiting value are dynamically adjusted, and power load stabilization is achieved. According to the technical scheme provided by the invention, the data information processing efficiency and stability of the large building can be improved.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and particularly to a big data information processing method and system based on deep learning. Background Art

[0002] With the development of modern buildings towards intelligence and energy conservation, precise control and optimized management of the internal environment of buildings have become the key to improving energy efficiency and reducing operating costs. Especially in high-energy-consuming buildings such as large commercial complexes and data centers, how to dynamically adjust the working state of temperature control equipment to adapt to changes in external meteorological conditions while ensuring indoor temperature comfort is one of the important challenges currently faced. In addition, in order to meet increasingly stringent environmental protection requirements, it is also particularly important to achieve effective management of power load and reduce unnecessary energy losses.

[0003] Currently, in response to the above technical requirements, data acquisition and analysis systems based on Internet of Things technology combined with cloud computing platforms have been widely used. This system distributes various sensors (such as temperature sensors, humidity sensors, etc.) in the building to collect various environmental data in real time and upload these data to the cloud server for centralized processing. With the help of advanced algorithm models, the system can predict the change trend of heat load in the next period of time, thereby providing a basis for the automatic adjustment of temperature control equipment.

[0004] Although the solutions based on Internet of Things and cloud computing have improved the building energy management problem to a certain extent, there are still some limitations. First, due to relying on the cloud platform for data analysis, this leads to the problem of data processing delay. Especially in the case of poor network conditions, real-time regulation may not be achieved, affecting the user experience. Second, existing systems often focus on monitoring single types of environmental parameters and lack the comprehensive analysis ability of multi-source heterogeneous data (such as electrical parameters, spatial topology, etc.), making it difficult to comprehensively capture the complex thermal environment characteristics inside the building. Finally, the scalability and flexibility of the system are insufficient. When facing different building structures or special application scenarios, a large amount of customized development work may be required, increasing the implementation difficulty and cost. Summary of the Invention

[0005] The embodiments of this application provide a big data information processing method and system based on deep learning to solve the problems of low efficiency and poor stability in the processing of building data information in the prior art.

[0006] In a first aspect, the embodiments of this application provide a big data information processing method based on deep learning, including: Obtain the big data information of a building, where the big data information includes the thermal resistance parameters of the envelope structure and the spatial topology data extracted through a building information model, and combine with the thermal distribution map collected by an infrared thermal imaging array to generate an energy distribution model with thermal property marks; Based on the heat conduction path characteristics in the energy distribution model, construct a set of dynamic heat balance equations, and generate three-dimensional temperature field reconstruction data through the difference compensation between the temperature field scan data and the simulation results of the set of dynamic heat balance equations; Input the thermal resistance parameters in the energy distribution model, the coordinates of the heat flow abnormal area in the three-dimensional temperature field reconstruction data, and the air conditioner unit current data into an event feature extraction network driven by deep learning to extract the spatio-temporal coupling characteristics of equipment power consumption and thermal environment parameters; Based on the spatio-temporal coupling characteristics and meteorological forecast data, construct a priority control chain for temperature control equipment, and generate a multi-level linkage rule including the adjustment of the fresh air unit ventilation rate, the optimization of the lighting dimming gradient, and the triggering of cold release from the cold storage tank; According to the sensitive node parameters of the priority control chain, dynamically generate a power load suppression strategy including the phase difference control of the chilled water pump, the staggered operation of the air conditioner group, and the dynamic power limit of the emergency lighting.

[0007] Optionally, the event feature extraction network driven by deep learning extracts the spatio-temporal coupling characteristics of equipment power consumption and thermal environment parameters, including: Overlay the thermal resistance parameters in the energy distribution model with the boundary data of the heat flow abnormal area in the three-dimensional temperature field reconstruction data to generate a spatial correlation map with a heat conduction lag effect; Based on the phase fluctuation characteristics of the air conditioner unit current data, mark the heat disturbance propagation path strongly related to the operation of electrical equipment in the spatial correlation map; According to the spatial coverage range of the heat disturbance propagation path, attenuate and weight the thermal resistance parameters according to the conduction path length, and overlay the air flow characteristic data in the three-dimensional temperature field reconstruction data to generate a thermal power coupling factor; Based on the three-dimensional distribution density of the thermal power coupling factor, construct a joint response sequence, and the joint response sequence generates multi-dimensional association rules by analyzing the co-occurrence probability of the current data spectrum characteristics and the geometric topology of the heat flow abnormal area; Perform cross-domain matching on the joint response sequence and the multi-dimensional association rules, and extract the spatio-temporal coupling characteristics of equipment power consumption and thermal environment parameters.

[0008] Optionally, the construction of the joint response sequence, where the joint response sequence generates multi-dimensional association rules by analyzing the co-occurrence probability of the current data spectrum characteristics and the geometric topology of the heat flow abnormal area, includes: Based on the three-dimensional distribution density of the thermoelectric coupling factor, the dynamic correlation intensity between the geometric topology of the heat flow anomaly area and the spectral characteristics of the current data is superimposed in multiple channels to construct a joint response sequence; Based on the coupling relationship between the axial gradient distribution of the joint response sequence and the frequency-domain energy attenuation rate, extract the set of boundary curvature change points of the geometric topology of the heat flow anomaly area; Match the set of boundary curvature change points with the zero-crossing distribution of the spectral characteristics of the current data to generate a dynamic co-occurrence probability field constrained by the main axis of the geometric topology of the heat flow anomaly area. The dynamic co-occurrence probability field includes the non-linear mapping relationship between the zero-crossing density and the boundary curvature change rate; Based on the synchronous change interval between the thermal fluctuation amplitude of the main axis and the frequency-domain energy attenuation rate, generate multi-dimensional correlation rules by constraining the boundary curvature change threshold and the gradient range of the zero-crossing density.

[0009] Optionally, the matching of the set of boundary curvature change points with the zero-crossing distribution of the spectral characteristics of the current data to generate a dynamic co-occurrence probability field constrained by the main axis of the geometric topology of the heat flow anomaly area includes: Perform gradient direction clustering on the set of boundary curvature change points to generate a cluster group of curvature change point sets, and use the centroid connection line of the cluster group of curvature change point sets as the initial matching reference chain; Perform circular neighborhood phase polarity analysis on the zero-crossing distribution of the spectral characteristics of the current data, screen out the zero-crossing sequence, and segment the zero-crossing sequence to generate a zero-crossing segmented trajectory chain; According to the direction of the main axis of the geometric topology of the heat flow anomaly area, dynamically adjust the local curvature weight of the initial matching reference chain and the phase polarity weight of the zero-crossing segmented trajectory chain to generate a double-chain dynamic matching probability map; According to the spatial aggregation degree of the double-chain dynamic matching probability map, perform coupling analysis on the cluster group of curvature change point sets, and generate a continuous probability field by superimposing the probability density gradients of adjacent cluster groups on the main axis; Perform direction-aware probability diffusion on the continuous probability field, apply anisotropic diffusion constraints along the main axis, and generate a dynamic co-occurrence probability field according to the distribution density of the double-chain dynamic matching probability map.

[0010] Optionally, the generation of three-dimensional temperature field reconstruction data by compensating for the difference between the temperature field scan data and the simulation results of the dynamic heat balance equation set includes: Based on the discrete sampling point distribution of the temperature field scan data, perform spatial discretization processing on the simulation results of the dynamic heat balance equation set to generate simulated temperature field grid data; By analyzing the temperature difference between the temperature field scan data and the simulated temperature field grid data, a local compensation factor is obtained, and smoothed diffusion is performed according to the spatial continuity constraint to generate continuous compensation field data; Couple and iterate the continuous compensation field data with the simulation results of the dynamic heat balance equation set, and perform reverse correction on the heat conduction term of the dynamic heat balance equation set to generate a corrected dynamic heat balance equation set; Perform backpropagation analysis on the spatial gradient distribution of the temperature field scan data according to the corrected dynamic heat balance equation set to generate compensated temperature field increment data, and generate three-dimensional temperature field reconstruction data by superimposing the simulated temperature field grid data.

[0011] Optionally, the coupling and iteration of the continuous compensation field data with the simulation results of the dynamic heat balance equation set, and the reverse correction of the heat conduction term of the dynamic heat balance equation set to generate a corrected dynamic heat balance equation set include: Based on the amplitude and direction of the compensation amount of the continuous compensation field data, a node correction coefficient is generated, and the absolute value of the node correction coefficient is positively correlated with the amplitude of the continuous compensation field data; According to the topological relationship of the heat conduction path of the dynamic heat balance equation set, define a heat conduction path weight factor, and perform conductivity weighted balancing of the node correction coefficient across nodes to generate a node correction coefficient distribution; Reverse superimpose the node correction coefficient distribution on the heat conduction term of the dynamic heat balance equation set to generate an intermediate correction equation set, and extract the heat flow residual parameters of the intermediate correction equation set and the dynamic heat balance equation set; Iteratively readjust the node correction coefficient distribution through the heat flow residual parameters until the heat flow residual parameters meet the heat flow conservation constraint conditions of the dynamic heat balance equation set to generate a corrected dynamic heat balance equation set.

[0012] Optionally, the construction of the priority control chain of the temperature control device based on the spatio-temporal coupling characteristics and meteorological forecast data includes: According to the correlation between the distributed thermal inertia parameter of the temperature control device in the spatio-temporal coupling characteristics and the future time period temperature fluctuation parameter in the meteorological forecast data, extract the spatial thermal inertia index and the time response sensitivity index of the temperature control device; Based on the correlation between the spatial thermal inertia index and the time response sensitivity index, establish the control influence weight of the temperature control device, and combine the topological relationship of the heat conduction path between the temperature control devices to generate a preliminary priority sequence; According to the spatio-temporal distribution characteristics of extreme temperature events in meteorological forecast data, dynamically adjust the preliminary priority sequence, and correct the regulation influence weight through the matching degree between the area covered by the extreme temperature event and the spatial position of the temperature control device to generate a dynamic priority sequence; By iteratively screening the temperature control device nodes in the dynamic priority sequence that violate the energy consumption threshold or the thermal stability threshold, construct a priority regulation chain for the temperature control devices.

[0013] In a second aspect, an embodiment of the present application provides a big data information processing system based on deep learning, including: An extraction module, which acquires big data information of a building, where the big data information includes the thermal resistance parameters of the building envelope and spatial topology data extracted through a building information model, and combines the thermal distribution map collected by an infrared thermal imaging array to generate an energy distribution model with thermal property marks; A scanning module, which constructs a set of dynamic heat balance equations based on the heat conduction path characteristics in the energy distribution model, and generates three-dimensional temperature field reconstruction data through the difference compensation between the temperature field scanning data and the simulation results of the set of dynamic heat balance equations; An input module, which inputs the thermal resistance parameters in the energy distribution model, the coordinates of the heat flow anomaly area in the three-dimensional temperature field reconstruction data, and the current data of the air conditioner unit into an event feature extraction network driven by deep learning to extract the spatio-temporal coupling characteristics of equipment power consumption and thermal environment parameters; A construction module, which constructs a priority regulation chain for temperature control devices based on the spatio-temporal coupling characteristics and meteorological forecast data, and generates a multi-level linkage rule including the adjustment of the fresh air unit ventilation rate, the optimization of the lighting dimming gradient, and the triggering of cold release from the cold storage tank; A generation module, which dynamically generates a power load suppression strategy including the phase difference control of the chilled water pump, the staggered operation of the air conditioner group, and the dynamic power limit of the emergency lighting according to the sensitive node parameters of the priority regulation chain.

[0014] In a third aspect, an embodiment of the present application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a big data information processing method based on deep learning as described in the first aspect above.

[0015] In a fourth aspect, an embodiment of the present application provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it implements a big data information processing method based on deep learning as described in the first aspect.

[0016] In the embodiments of the present application, big data information of a building is obtained. The big data information includes the thermal resistance parameters of the envelope structure and spatial topology data extracted through a building information model, and combines with the thermal distribution map collected by an infrared thermal imaging array to generate an energy distribution model with thermal property marks; based on the heat conduction path characteristics in the energy distribution model, a set of dynamic heat balance equations is constructed, and through the differential compensation between the temperature field scan data and the simulation results of the set of dynamic heat balance equations, three-dimensional temperature field reconstruction data is generated; the thermal resistance parameters in the energy distribution model, the coordinates of the heat flow anomaly area in the three-dimensional temperature field reconstruction data, and the current data of the air conditioning unit are input into an event feature extraction network driven by deep learning to extract the spatio-temporal coupling characteristics of equipment power consumption and thermal environment parameters; based on the spatio-temporal coupling characteristics and meteorological forecast data, a priority control chain of temperature control equipment is constructed, and a multi-level linkage rule including the adjustment of the ventilation rate of the fresh air unit, the optimization of the lighting dimming gradient, and the triggering of cold release from the cold storage tank is generated; according to the sensitive node parameters of the priority control chain, a power load suppression strategy including the phase difference control of the chilled water pump, the staggered operation of the air conditioning group, and the dynamic power limit of the emergency lighting is dynamically generated.

[0017] The technical solution of the present application has the following beneficial effects: The present application obtains the big data information of the building. The big data information includes the thermal resistance parameters of the envelope structure and spatial topology data extracted through a building information model, and combines with the thermal distribution map collected by an infrared thermal imaging array, which can accurately describe the thermal environment characteristics inside the building. This process provides detailed basic data for subsequent analysis and improves the understanding accuracy of the heat distribution inside the building. A set of dynamic heat balance equations is constructed based on the heat conduction path characteristics in the energy distribution model, and differential compensation is performed using the temperature field scan data to generate three-dimensional temperature field reconstruction data. This method can more accurately simulate the actual temperature field inside the building, improving the accuracy and response speed of temperature control. A priority control chain of temperature control equipment is constructed based on the spatio-temporal coupling characteristics and meteorological forecast data, and a multi-level linkage rule including the adjustment of the ventilation rate of the fresh air unit, the optimization of the lighting dimming gradient, and the triggering of cold release from the cold storage tank is formulated. This method can achieve intelligent adjustment of temperature control equipment, improving energy efficiency while ensuring indoor comfort. According to the sensitive node parameters of the priority control chain, a power load suppression strategy including the phase difference control of the chilled water pump, the staggered operation of the air conditioning group, and the dynamic power limit of the emergency lighting is dynamically generated. This not only helps to reduce the peak power consumption but also ensures the stable operation of the system.

[0018] Furthermore, by superimposing the thermal resistance parameters in the energy distribution model on the boundary data of the heat flux anomaly region in the reconstructed data of the three-dimensional temperature field, a spatial correlation map with a heat conduction lag effect is generated. Based on the current data of the air conditioner unit, the heat disturbance propagation path is marked. The thermal-electric coupling factor is calculated by superimposing the weighted decay of the thermal resistance parameters on the air flow characteristic data. Finally, the multi-dimensional association rules are generated by jointly analyzing the spectral characteristics of the current data and the geometric topology co-occurrence probability of the heat flux anomaly region, so as to extract the spatio-temporal coupling characteristics of the device power consumption and the thermal environment parameters. This method greatly enhances the understanding of the operating state of electrical equipment in a complex thermal environment and its impact on the thermal environment, enabling the system to more accurately identify the key regions and factors causing energy loss. By precisely locating and analyzing these key points, not only can the temperature control equipment be more finely regulated, but also the energy consumption can be effectively reduced while ensuring comfort, and at the same time, the intelligent level and response efficiency of the entire building energy management system can be improved.

[0019] These aspects or other aspects of the present application will be more clearly understood in the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0021] Figure 1 The flowchart of a big data information processing method based on deep learning provided by the present application is shown; Figure 2 The structural schematic diagram of a big data information processing system based on deep learning provided by the present application is shown; Figure 3 The structural schematic diagram of a computing device provided by the present application is shown. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] In order to enable those skilled in the art to better understand the solution of the present application, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application.

[0023] In some of the processes described in the specification, claims, and the above-mentioned drawings of this application, a number of operations appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order in which they appear herein or may be executed in parallel. The operation numbers, such as 101, 102, etc., are only used to distinguish different operations, and the numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions such as "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent a sequence, and do not limit that "first" and "second" are different types.

[0024] This application aims to generate an energy distribution model with thermal property markings by using building information models and infrared thermal imaging technology. Subsequently, a set of dynamic thermal balance equations is constructed based on the heat conduction path characteristics in the model, and differential compensation is performed through temperature field scan data to generate three-dimensional temperature field reconstruction data. Further, by combining the energy distribution model, the three-dimensional temperature field reconstruction data, and the air conditioner unit current data, the spatio-temporal coupling characteristics of equipment power consumption and thermal environment parameters are extracted by inputting them into a deep learning network. Finally, based on these characteristics and meteorological forecast data, a temperature control device regulation strategy is formulated to optimize the control of the building interior environment.

[0025] Next, the technical solutions in the embodiments of this application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative efforts fall within the scope of protection of this application.

[0026] Figure 1 The flowchart of a big data information processing method based on deep learning provided for the embodiments of this application is as Figure 1 shown, and the method includes: 101. Obtain the big data information of the building, where the big data information includes the thermal resistance parameters of the envelope structure and the spatial topology data extracted through the building information model, and combine the thermal distribution map collected by the infrared thermal imaging array to generate an energy distribution model with thermal property markings; In this step, the building information model is a digital tool for creating and managing building project information, which contains all the physical and functional characteristic data of the building.

[0027] The thermal resistance parameter of the envelope structure refers to the resistance value of the building envelope such as the exterior wall and roof to heat conduction, and is an important indicator to measure the thermal insulation performance of building materials.

[0028] Spatial topological data describes the spatial layout inside a building and its interrelationships, such as the location, size, and connection methods of rooms.

[0029] An infrared thermal imaging array is a device that uses infrared technology to capture the temperature distribution on the surface of an object. The generated thermal distribution map can visually display the temperature differences in different areas inside and outside the building.

[0030] A thermal distribution map is an image generated by infrared thermal imaging technology, showing the temperature distribution on the surface of an object. In the field of architecture, it can be used to identify abnormal heat flow areas in a building, such as cold bridges around windows or insulation failure points in walls.

[0031] An energy distribution model is a comprehensive digital model that integrates all the above information and attaches thermal property tags to show the energy flow status inside the building.

[0032] In the embodiments of this application, first, the thermal resistance parameters and spatial topological data of the building envelope are extracted through a building information model. This process involves using specific software to parse the building design documents to obtain relevant parameters. Then, the building is scanned using an infrared thermal imaging array to collect its thermal distribution map. When processing these data, image processing algorithms are used to identify and mark abnormal heat flow areas, and this information is integrated into the energy distribution model. Finally, combining all the collected data, an energy distribution model containing detailed thermal property information is generated. This model accurately depicts the heat distribution inside the building and reflects its response characteristics to changes in the external environment.

[0033] In an office building project, first, building information model software was used to collect the thermal resistance parameters and spatial topological data of the entire building, including the material properties and relative positions of each floor slab and wall. Subsequently, under the cold weather conditions in winter, a high-resolution infrared thermal imaging device was used to conduct a comprehensive scan of the building, obtaining a detailed thermal distribution map and discovering several obvious cold bridge effects around the windows. Based on these data, an energy distribution model that accurately reflects the actual heat flow was constructed, providing a basis for subsequent optimization of the air conditioning system operation.

[0034] 102. Based on the heat conduction path characteristics in the energy distribution model, construct a set of dynamic thermal balance equations. Through the difference compensation between the temperature field scan data and the simulation results of the set of dynamic thermal balance equations, generate three-dimensional temperature field reconstruction data; In this step, the energy distribution model is a comprehensive digital model that integrates all the above information and attaches thermal property tags to show the energy flow status inside the building.

[0035] The dynamic heat balance equation set is a group of mathematical expressions used to simulate the heat transfer process in a building, taking into account factors such as the thermal conductivity of building materials and boundary conditions.

[0036] The temperature field scan data is the temperature values at various positions in the building collected in real time by a sensor network.

[0037] The three-dimensional temperature field reconstruction data is a three-dimensional temperature distribution model after compensation, which more accurately reflects the actual temperature distribution inside the building.

[0038] In the embodiment of the present application, first, based on the heat conduction path characteristics in the energy distribution model, a dynamic heat balance equation set is established. Then, a distributed temperature sensor network deployed inside the building is used to collect temperature field scan data. By comparing the temperature field scan data with the simulation results of the dynamic heat balance equation set, the difference points are found and adjusted through a compensation algorithm, and finally, more realistic three-dimensional temperature field reconstruction data is generated. This step ensures that the temperature field model can truly reflect the actual temperature distribution inside the building.

[0039] Continuing with the data from the previous step, a dynamic heat balance equation set is established based on the energy distribution model, and a set of distributed temperature sensor networks is deployed to monitor the temperature changes inside the office building in real time. By analyzing the collected data, some local hot spots are found. The compensation algorithm is used to finely adjust these areas, and finally, a three-dimensional temperature field reconstruction model that accurately reflects the actual situation is generated, providing a scientific basis for optimizing the operation of the air conditioning system.

[0040] 103. Input the thermal resistance parameters in the energy distribution model, the coordinates of the heat flow anomaly areas in the three-dimensional temperature field reconstruction data, and the current data of the air conditioning unit into the event feature extraction network driven by deep learning to extract the spatio-temporal coupling features of equipment power consumption and thermal environment parameters; In this step, the thermal resistance parameters are data describing the heat conduction resistance of the building envelope (such as walls, windows, etc.). It reflects the ability of the material to prevent heat transfer and is crucial for evaluating the insulation performance of the building.

[0041] The coordinates of the heat flow anomaly areas refer to the position coordinates marked in the three-dimensional temperature field reconstruction data where the temperature is significantly higher or lower than the surrounding areas.

[0042] The current data of the air conditioning unit records the current consumption during the operation of the air conditioning system and is one of the important indicators for evaluating equipment power consumption.

[0043] The spatio-temporal coupling features refer to the relationship patterns of these variables changing with time and space, revealing how equipment power consumption responds to different thermal environment conditions.

[0044] In the embodiments of the present application, first, the thermal resistance parameters, the coordinates of the heat flux anomaly region, and the air conditioner unit current data in the energy distribution model are input into an event feature extraction network driven by deep learning. The network learns the spatio-temporal coupling features between the device power consumption and the thermal environment parameters through training. This process involves a large amount of data preprocessing, feature engineering, and model training work, and the final output results help to understand the device operating state and its impact on the environment.

[0045] Based on the data obtained in the previous two steps, a deep learning model is constructed to analyze the energy consumption in different areas of the office building. By continuously monitoring the current consumption of the air conditioning system and the corresponding temperature changes, it is found that the energy consumption increases significantly during certain specific time periods, and the specific reasons for this situation are determined, such as sudden changes in external air temperature or increased personnel activities.

[0046] 104. Based on the spatio-temporal coupling features and the meteorological forecast data, construct a priority control chain for the temperature control equipment, and generate multi-level linkage rules including the adjustment of the fresh air unit ventilation rate, the optimization of the lighting dimming gradient, and the triggering of cold release from the cold storage tank; In this step, the spatio-temporal coupling features refer to the relationship patterns of these variables changing with time and space, which reveal how the device power consumption responds to different thermal environment conditions.

[0047] The meteorological forecast data contains weather prediction information for a period of time in the future, including the change trends of elements such as temperature and humidity.

[0048] The multi-level linkage rules refer to a series of operation instructions set according to different conditions, covering multiple aspects such as the adjustment of the fresh air unit ventilation rate and the optimization of the lighting dimming gradient, aiming to achieve the energy-saving goal while ensuring comfort.

[0049] The priority control chain is a series of control strategies formulated according to the spatio-temporal coupling features and the meteorological forecast data to guide the effective operation of the temperature control equipment.

[0050] In the embodiments of the present application, combining the spatio-temporal coupling features extracted in the above steps and the meteorological forecast data, a priority control chain for the temperature control equipment is formulated. This chain contains multiple levels of operation instructions, and automatically adjusts the device operating state according to different climate conditions and indoor demands. For example, the cooling equipment is preferentially started in high-temperature weather, and the ventilation volume is appropriately reduced in low temperature. This not only saves energy but also improves the user experience.

[0051] During the summer high - temperature warning period, based on meteorological forecast data, the energy distribution model of the office building, and the reconstructed data of the three - dimensional temperature field, a priority control chain for temperature - control equipment was formulated. First, the ventilation rate of the fresh - air unit was adjusted, and night ventilation was increased to reduce the indoor temperature baseline. At the same time, the dimming gradient of the lighting system was optimized, and the light intensity was reduced during non - working hours. These measures not only reduced the starting pressure of the air - conditioning system but also prepared for the upcoming high - temperature weather.

[0052] 105. According to the sensitive node parameters of the priority control chain, a power load smoothing strategy including chilled - water pump phase - difference control, staggered operation of air - conditioning groups, and dynamic power limit of emergency lighting is dynamically generated.

[0053] In this step, the sensitive node parameters include, but are not limited to, information such as the maximum power, starting time, and operating mode of the equipment. By monitoring and analyzing these parameters, effective load management strategies can be formulated.

[0054] Chilled - water pump phase - difference control refers to adjusting the working phase of the chilled - water pump motor to optimize its working state and avoid grid overload caused by starting multiple high - power devices simultaneously.

[0055] Staggered operation of air - conditioning groups is a load management technique that arranges different air - conditioning units to operate in different time periods to avoid peak loads caused by all devices starting simultaneously.

[0056] The dynamic power limit of emergency lighting refers to the maximum power limit of the temporary lighting system set to cope with emergencies.

[0057] The power load smoothing strategy is a series of technical means and operating procedures aimed at balancing the relationship between power demand and supply.

[0058] In the embodiment of the present application, based on the priority control chain of the above - mentioned steps, the power load management strategy is further refined. Through real - time monitoring of sensitive node parameters, the working phase - difference of the chilled - water pump is dynamically adjusted, the staggered operation plan of the air - conditioning system is implemented, and the power upper limit of emergency lighting is set. These measures work together to effectively reduce the peak power load and improve the stability of the power grid.

[0059] Based on the priority control chain, phase - difference control is further implemented for the chilled - water pumps in the office building, and the staggered operation of air - conditioning groups is arranged to avoid grid overload caused by all devices starting simultaneously. In addition, the dynamic power limit of the emergency lighting system is set to ensure that the basic lighting needs can be maintained during peak power consumption periods without affecting normal office activities. Through these refined management measures, the peak power load of the office building has been successfully reduced, significantly improving energy - use efficiency and ensuring power supply stability. These two steps work together to ensure that the office building can still operate efficiently and stably under extreme weather conditions.

[0060] In summary, steps 101 to 105 achieve the refined management and high-efficiency operation of the internal environment of the building. From the initial data collection to the final strategy execution, each step makes full use of modern information technology means to ensure that the building can achieve the best energy-saving effect while meeting the needs of users. This comprehensive solution not only improves the energy utilization efficiency but also enhances the building's ability to adapt to changes in the external environment.

[0061] To further improve the understanding of the spatio-temporal coupling characteristics between equipment power consumption and thermal environment parameters, a spatial correlation map is generated by superimposing the thermal resistance parameters in the energy distribution model and the boundary data of the heat flux anomaly area in the reconstructed three-dimensional temperature field data. According to the spatial coverage range of the heat disturbance propagation path, the co-occurrence probability of the spectral characteristics of the current data and the geometric topology of the heat flux anomaly area is analyzed, and finally, the accurate extraction of the spatio-temporal coupling characteristics between equipment power consumption and thermal environment parameters is realized, providing a basis for subsequent intelligent regulation. In some embodiments, the spatio-temporal coupling characteristics of equipment power consumption and thermal environment parameters are extracted through the event feature extraction network driven by deep learning in step 103, including: 201. Superimpose the thermal resistance parameters in the energy distribution model and the boundary data of the heat flux anomaly area in the reconstructed three-dimensional temperature field data to generate a spatial correlation map with a heat conduction lag effect; In step 201, the thermal resistance parameters describe the data of the resistance of the building envelope to heat transfer, reflecting the ability of different materials to prevent heat from transferring from one side to the other. The boundary data of the heat flux anomaly area are the position coordinates marked in the reconstructed three-dimensional temperature field data where the temperature is significantly different from the surrounding area. The spatial correlation map is a chart generated by combining these two types of data, used to display the lag effect and its influence range in the heat transfer process. The heat conduction lag effect refers to the time delay phenomenon caused by material properties and distance in the heat transfer process.

[0062] In the embodiment of the present application, first, the thermal resistance parameters in the energy distribution model and the boundary data of the heat flux anomaly area in the reconstructed three-dimensional temperature field data are superimposed and processed, and an image fusion technology is used to generate a spatial correlation map with a heat conduction lag effect. This process not only reveals the basic mode of heat flow inside the building but also considers the influence of material properties on the heat transfer speed. The final result is a spatial correlation map that can accurately reflect the heat flow situation inside the building.

[0063] 202. Based on the phase fluctuation characteristics of the current data of the air conditioning unit, mark the heat disturbance propagation path strongly related to the operation of electrical equipment in the spatial correlation map; In step 202, the phase fluctuation characteristics of the air conditioner unit current data refer to the data recording the current change situation when the air conditioning system is operating, and these data can reflect the working state and load level of the equipment. The heat disturbance propagation path is the propagation path of the heat fluctuation caused by the operation of electrical equipment in the building, identifying which areas are most affected by the operation of electrical equipment.

[0064] In the embodiment of the present application, based on the phase fluctuation characteristics of the air conditioner unit current data, signal processing algorithms are used to analyze the current data, identify the phase fluctuation characteristic points, and map them to the corresponding heat disturbance propagation paths in the spatial correlation map, thereby identifying the key heat transfer paths. By marking the heat disturbance propagation paths strongly correlated with the operation of electrical equipment in the spatial correlation map, this step helps to accurately identify the hot spots of heat fluctuation caused by the operation of electrical equipment.

[0065] 203. According to the spatial coverage range of the heat disturbance propagation path, attenuate and weight the thermal resistance parameters according to the conduction path length, and superimpose the air flow characteristic data in the reconstructed three-dimensional temperature field data to generate a thermoelectric power coupling factor; In step 203, the thermoelectric power coupling factor is an index combining the thermal resistance parameters and the air flow characteristic data, used to quantify the interaction intensity between heat and electric energy. The attenuation weighting of the conduction path length is a method of adjusting the influence degree of heat according to the distance of heat transmission along different paths. The air flow characteristic data describes the direction and speed of air flow inside the building, which is crucial for understanding how heat diffuses in the building.

[0066] In the embodiment of the present application, first, analyze the spatial coverage range of the heat disturbance propagation path and calculate the length of each conduction path. Then, perform attenuation weighting processing on the thermal resistance parameters based on these lengths. Next, superimpose the air flow characteristic data in the reconstructed three-dimensional temperature field data to generate a thermoelectric power coupling factor. Specifically, use the finite element analysis method to process complex multi-dimensional data sets to ensure that the generated factor can accurately reflect the actual situation. Finally, integrate all the information to form a comprehensive thermoelectric power coupling factor.

[0067] 204. Based on the three-dimensional distribution density of the thermoelectric power coupling factor, construct a joint response sequence, and the joint response sequence generates multi-dimensional association rules by analyzing the co-occurrence probability of the current data spectrum characteristics and the geometric topology of the heat flow abnormal area; In step 204, the combined response sequence is a set of data sequences obtained by analyzing the co-occurrence probability of the spectral characteristics of current data and the geometric topology of the heat flux anomaly region. The multi-dimensional association rules are extracted from these sequences to describe the interaction patterns between different variables. The spectral characteristics of current data are the results obtained from the frequency-domain analysis of current data, while the geometric topology of the heat flux anomaly region refers to the spatial form and positional relationship of the heat flux anomaly region.

[0068] In the embodiments of the present application, the fast Fourier transform is used to analyze the spectral characteristics of current data, and a geometric modeling tool is combined to analyze the geometric topology of the heat flux anomaly region to find the co-occurrence pattern between the two. Then, multi-dimensional association rules are generated. This step uses statistical methods, such as Bayesian networks or decision trees, to establish the association rules between different variables. Finally, all the information is integrated to form a comprehensive multi-dimensional association rule for guiding subsequent optimization strategies.

[0069] 205. Perform cross-domain matching on the combined response sequence and the multi-dimensional association rules, and extract the spatio-temporal coupling characteristics of the device power consumption and the thermal environment parameters.

[0070] In step 205, cross-domain matching refers to comparing the combined response sequence with the multi-dimensional association rules to find the corresponding relationship between the two. The spatio-temporal coupling characteristics are extracted from this matching to describe the relationship pattern of the device power consumption and the thermal environment parameters changing with time and space. The cross-domain matching technology aims to discover the potential connections between different data sources to provide a more comprehensive perspective to understand the dynamic behavior of the system.

[0071] In the embodiments of the present application, a time series analysis method, such as the autoregressive integrated moving average model, is used in combination with geographic information system data to analyze the changing laws of the device power consumption and the thermal environment parameters with time and space. Then, based on the above analysis results, spatio-temporal coupling characteristics are generated. This step uses data mining techniques, such as clustering analysis and principal component analysis, to extract the key characteristics. Finally, all the information is integrated to form a comprehensive spatio-temporal coupling characteristic for guiding subsequent optimization strategies.

[0072] The following is a specific example: In an office building project, first, the building information model system was used to extract the thermal resistance parameters of the building envelope and the spatial topology data of the whole building, and an infrared thermal imaging device was used to obtain a detailed thermal distribution map. Then, the temperature field scan data was collected through a distributed temperature sensor network to generate three-dimensional temperature field reconstruction data. On this basis, the technical personnel superimposed the thermal resistance parameters and the boundary data of the heat flow anomaly area to generate a spatial correlation map. Then, based on the phase fluctuation characteristics of the current data of the air conditioning unit, the main heat disturbance propagation paths were marked on the spatial correlation map. Next, according to the spatial coverage of these paths, the thermoelectric coupling factor was calculated, and a joint response sequence was further constructed. Finally, through cross-domain matching technology, the spatio-temporal coupling characteristics of the equipment power consumption and the thermal environment parameters were extracted, providing a scientific basis for optimizing the overall temperature control strategy of the office building.

[0073] In summary, through steps 201 to 205, through a series of in-depth data analysis and technical means, an accurate understanding and description of the complex relationship between the internal equipment power consumption and the thermal environment parameters of the building have been achieved. This method not only improves the efficiency of energy management but also provides a solid foundation for formulating more intelligent and personalized energy-saving strategies. Through the detailed analysis of the interaction between heat and electricity inside the building, unnecessary energy losses can be effectively reduced, and the overall energy efficiency performance of the building can be improved. At the same time, this method enhances the ability to cope with extreme weather conditions and ensures the comfort and stability of the indoor environment.

[0074] To further improve the understanding of the co-occurrence probability between the spectral characteristics of the current data and the geometric topology of the heat flow anomaly area, the solution superimposes the three-dimensional distribution density of the thermoelectric coupling factor through multiple channels, extracts the set of boundary curvature change points of the geometric topology of the heat flow anomaly area, and generates a dynamic co-occurrence probability field. Based on the synchronous change interval of the thermal fluctuation amplitude and the frequency domain energy attenuation rate of the main axis, multi-dimensional association rules are generated, improving the understanding accuracy and response speed of the system to complex thermal environment changes. In some embodiments, in step 204, the construction of the joint response sequence, the joint response sequence generates multi-dimensional association rules by analyzing the co-occurrence probability of the spectral characteristics of the current data and the geometric topology of the heat flow anomaly area, including: 301. Based on the three-dimensional distribution density of the thermoelectric coupling factor, superimpose the dynamic association intensity of the geometric topology of the heat flow anomaly area and the spectral characteristics of the current data in the spatial coordinates through multiple channels to construct a joint response sequence; In step 301, the thermoelectric coupling factor is an index that combines thermal resistance parameters and air flow characteristic data, and is used to quantify the interaction strength between heat and electricity. The three-dimensional distribution density refers to the spatial distribution of the thermoelectric coupling factor within the entire building. The geometric topology of the heat flow anomaly region describes the spatial form and positional relationship of the heat flow anomaly region. The spectral characteristics of the current data are the results obtained after performing frequency-domain analysis on the current data. The joint response sequence is a set of data sequences obtained by analyzing these data, and is used to reveal the dynamic correlation strength between different variables.

[0075] In the embodiment of the present application, first, based on the three-dimensional distribution density of the thermoelectric coupling factor, the geometric topology of the heat flow anomaly region and the spectral characteristics of the current data are superimposed in multiple channels under spatial coordinates by using image processing technology to generate a joint response sequence. This process involves complex mathematical modeling and signal processing algorithms, such as the fast Fourier transform, and the final result is a joint response sequence that can accurately reflect the dynamic correlation strength between different variables.

[0076] 302. Extract a set of boundary curvature change points of the geometric topology of the heat flow anomaly region based on the coupling relationship between the axial gradient distribution of the joint response sequence and the frequency-domain energy attenuation rate; In step 302, the axial gradient distribution of the joint response sequence refers to the change trend of the sequence in a certain direction. The frequency-domain energy attenuation rate refers to the speed at which energy decreases with frequency in the frequency domain. The set of boundary curvature change points is extracted from the joint response sequence and represents the set of positions where the boundary curvature changes in the geometric topology of the heat flow anomaly region.

[0077] In the embodiment of the present application, based on the coupling relationship between the axial gradient distribution of the joint response sequence and the frequency-domain energy attenuation rate, a mathematical analysis method is used to extract a set of boundary curvature change points of the geometric topology of the heat flow anomaly region. Specifically, differential geometry techniques are used to calculate the change points of the boundary curvature and mark them to form a set containing all key points. This step helps to accurately identify the important change regions in the heat flow anomaly region.

[0078] 303. Match the set of boundary curvature change points with the zero-crossing distribution of the spectral characteristics of the current data to generate a dynamic co-occurrence probability field constrained by the main axis of the geometric topology of the heat flow anomaly region, and the dynamic co-occurrence probability field includes a non-linear mapping relationship between the zero-crossing density and the boundary curvature change rate; In step 303, the set of boundary curvature change points is a set composed of the positions where the boundary curvature changes in the geometric topology of the heat flow anomaly region. The zero-crossing distribution of the spectral characteristics of the current data refers to the position distribution where the signal crosses zero in the current data spectrogram. The dynamic co-occurrence probability field is a probability distribution map generated by combining the above two types of data, used to display the co-occurrence probability relationship between the two. The main axis refers to the main extension direction of the geometric topology of the heat flow anomaly region.

[0079] In the embodiment of the present application, the set of boundary curvature change points is matched with the zero-crossing distribution of the spectral characteristics of the current data, and a dynamic co-occurrence probability field constrained by the main axis of the geometric topology of the heat flow anomaly region is generated using pattern recognition technology. In this process, by comparing the set of boundary curvature change points with the zero-crossing distribution, the co-occurrence probability between the two is calculated, and a dynamic co-occurrence probability field including the non-linear mapping relationship between the zero-crossing density and the boundary curvature change rate is generated. This step helps to understand the complex relationship between the current data and the heat flow anomaly region.

[0080] 304. Based on the synchronous change interval of the thermal fluctuation amplitude of the main axis and the frequency domain energy attenuation rate, generate multi-dimensional association rules by constraining the boundary curvature change threshold and the gradient range of the zero-crossing density.

[0081] In step 304, the thermal fluctuation amplitude of the main axis refers to the temperature fluctuation amplitude of the heat flow anomaly region along its main extension direction. The frequency domain energy attenuation rate is the speed at which the energy in the frequency domain decreases with the change of frequency. The multi-dimensional association rules are extracted from the dynamic co-occurrence probability field and used to describe the interaction patterns between different variables. The boundary curvature change threshold is a numerical range used to limit the boundary curvature change rate. The zero-crossing density refers to the number of times the signal crosses zero in the current data spectrogram.

[0082] In the embodiment of the present application, first, obtain the thermal fluctuation amplitude of the main axis and the frequency domain energy attenuation rate, and determine the synchronous change interval. This step uses sensor data and spectral analysis techniques, such as fast Fourier transform, to extract key features. Then, set the boundary curvature change threshold and calculate the gradient range of the zero-crossing density. Specifically, use mathematical modeling tools, such as finite element analysis, to set reasonable boundary conditions and use statistical methods to calculate the zero-crossing density and its gradient range. Then, based on the above processing results, generate multi-dimensional association rules. This step uses machine learning algorithms, such as decision trees or neural networks, to establish the association rules between different variables. Finally, integrate all the information to form a comprehensive multi-dimensional association rule for guiding subsequent optimization strategies.

[0083] The following is a specific example: In an office building project, technicians first use image processing technology to perform multi-channel superposition of the geometric topology of the heat flow anomaly area and the spectral characteristics of current data in the spatial coordinates to generate a joint response sequence. Then, differential geometry technology is used to extract the set of boundary curvature change points of the geometric topology of the heat flow anomaly area. Next, pattern recognition technology is used to generate a dynamic co-occurrence probability field constrained by the main axis of the geometric topology of the heat flow anomaly area. Finally, by constraining the gradient range of the boundary curvature change threshold and the zero-crossing density, multi-dimensional association rules are generated. These rules help optimize the operation strategy of the air conditioning system in the office building, improve the energy use efficiency, and ensure the comfort of the indoor environment at the same time.

[0084] In summary, through steps 301 to 304, through a series of in-depth data analysis and technical means, an accurate understanding and description of the complex relationship between the spectral characteristics of current data and the geometric topology of the heat flow anomaly area are achieved. This method not only improves the efficiency of energy management, but also provides a solid foundation for formulating more intelligent and personalized energy-saving strategies. By carefully analyzing the interaction between heat and electricity inside the building, the overall energy efficiency performance of the building is improved. At the same time, this method enhances the ability to cope with extreme weather conditions, ensuring the comfort and stability of the indoor environment. This method also improves the response speed and accuracy of the system, making the energy management system more efficient and intelligent.

[0085] To further improve the understanding of the matching between the set of boundary curvature change points and the zero-crossing distribution of the spectral characteristics of current data, the solution generates clusters by clustering the set of boundary curvature change points to form a zero-crossing segmented trajectory chain, and further analyzes the spatial aggregation degree to superimpose the probability density gradient of adjacent clusters to generate a continuous probability field, obtaining a dynamic co-occurrence probability field. In some embodiments, the step 303 of matching the set of boundary curvature change points with the zero-crossing distribution of the spectral characteristics of current data to generate a dynamic co-occurrence probability field constrained by the main axis of the geometric topology of the heat flow anomaly area includes: 401. Perform gradient direction clustering on the set of boundary curvature change points to generate a cluster of curvature change point sets, and use the line connecting the centroids of the cluster of curvature change point sets as the initial matching reference chain; In step 401, the set of boundary curvature change points is a set composed of the positions where the boundary curvature changes in the geometric topology of the heat flow anomaly area. Gradient direction clustering is a clustering method based on the similarity of the gradient directions of points in the point set, used to divide the set of boundary curvature change points into several clusters. The line connecting the centroids is the line segment formed by connecting the centroids of all points in each cluster, used as the initial matching reference chain. The initial matching reference chain is the basic reference line for the subsequent matching process.

[0086] In the embodiment of the present application, first, the boundary information in the design drawings of the outer wall of the office building is collected, and the curvature change at each boundary point is calculated. Then, the k-means clustering algorithm is used to classify these curvature change points according to their gradient directions, thereby forming multiple clusters with similar characteristics. Next, the centroid position is determined for each cluster, and these centroids are connected in sequence to form an initial matching reference chain. This chain not only summarizes the design characteristics of the outer wall of the building but also provides a basis for further analysis.

[0087] 402. Perform circular neighborhood phase polarity analysis on the zero-crossing distribution of the current data spectrum characteristics, filter out the zero-crossing sequence, and segment the zero-crossing sequence to generate a zero-crossing segmented trajectory chain. In step 402, the zero-crossing distribution of the current data spectrum characteristics refers to the position distribution where the signal in the current data spectrogram crosses zero. The circular neighborhood phase polarity analysis is a method for local area analysis of the current data spectrum characteristics, used to filter out the zero-crossing sequence with specific phase and polarity. The zero-crossing sequence is an ordered set where the signal in the current data spectrum characteristics crosses zero after filtering. The zero-crossing segmented trajectory chain is a trajectory chain obtained by segmenting the zero-crossing sequence.

[0088] In the embodiment of the present application, first, the current data spectrum characteristics of the internal circuit of the office building are obtained, and all zero-crossing points are identified. Then, circular neighborhood phase polarity analysis is performed on each zero-crossing point to filter out the zero-crossing sequence that meets specific conditions. Next, these zero-crossing points are segmented according to the current fluctuation pattern, and each segment represents a continuous current characteristic segment. Finally, a zero-crossing segmented trajectory chain is formed by concatenating these segments, which helps with subsequent fault detection and location work.

[0089] 403. Dynamically adjust the local curvature weight of the initial matching reference chain and the phase polarity weight of the zero-crossing segmented trajectory chain according to the direction of the main axis of the geometric topology of the heat flow anomaly area to generate a double-chain dynamic matching probability map. In step 403, the main axis refers to the main extension direction of the geometric topology of the heat flow anomaly area. The centroid connection line generated by the initial matching reference chain. The local curvature weight is different weight values assigned to the curvature change degree of each point on the initial matching reference chain. The segmented trajectory chain of the current data spectrum characteristics generated by the zero-crossing segmented trajectory chain. The phase polarity weight is different weight values assigned to the phase and polarity of each point on the zero-crossing segmented trajectory chain. The double-chain dynamic matching probability map is a probability distribution map generated by combining the above two types of data, used to display the matching relationship between the two.

[0090] In the embodiments of the present application, first, based on the geometric topology of the heat flow anomaly region, the direction of the main axis is determined, which is crucial for dynamically adjusting the local curvature weight of the initial matching reference chain. Then, in combination with the adjusted weight parameters, the matching degree between the initial matching reference chain and the zero-crossing segmented trajectory chain is calculated. Next, based on the above matching results, a double-chain dynamic matching probability map is generated, where the high-probability region indicates a high degree of consistency between the two chains. This map shows the matching possibilities of the two chains at different positions, laying a foundation for the next analysis.

[0091] 404. Perform a coupling analysis on the cluster group of curvature change points according to the spatial aggregation degree of the double-chain dynamic matching probability map. By superimposing the probability density gradients of adjacent cluster groups on the main axis, a continuous probability field is generated. In step 404, the spatial aggregation degree refers to the density of data points in this probability map. The probability density gradient refers to the rate of change of the probability density between adjacent cluster groups on the main axis. The continuous probability field is a probability distribution field generated by superimposing the probability density gradients of adjacent cluster groups.

[0092] In the embodiments of the present application, first, the spatial aggregation degree of the cluster group of curvature change points is calculated to quantify the density of the cluster group of curvature change points in space. Then, a coupling analysis is performed to evaluate the influence between adjacent cluster groups. Next, the probability density gradients of these cluster groups are superimposed on the main axis to form a smooth and continuous probability field. This probability field reveals potential patterns and trends within the entire research area, providing a scientific basis for optimizing building design.

[0093] 405. Perform direction-aware probability diffusion on the continuous probability field, apply anisotropic diffusion constraints along the main axis, and generate a dynamic co-occurrence probability field according to the distribution density of the double-chain dynamic matching probability map.

[0094] In step 405, direction-aware probability diffusion refers to a method of probability diffusion along the main axis, considering the influence of direction factors. Anisotropic diffusion constraints refer to the diffusion limitations in different directions applied during the diffusion process. The dynamic co-occurrence probability field is the finally generated probability distribution map, showing the co-occurrence probability relationship between the geometric topology of the heat flow anomaly region and the spectral characteristics of current data.

[0095] In the embodiments of the present application, first, the direction-aware probability diffusion algorithm is applied along the main axis to make the probability information propagate in a specific direction, while considering the influence of direction factors. Then, anisotropic diffusion constraints are introduced to ensure that this diffusion process is more in line with the actual situation. Next, according to the distribution density of the double-chain dynamic matching probability map, a dynamic co-occurrence probability field is generated. Finally, this probability field comprehensively reflects the probability distribution under the combined action of various factors, helping to predict possible future structural problems or energy consumption patterns.

[0096] The following is a specific example: In an office building project, technicians began to process the design drawings of the building's exterior wall, extract boundary information from them, and calculate the set of curvature change points. These points were successfully classified into several clusters, and the centroids of each cluster were connected to form an initial matching reference chain. Subsequently, the technicians analyzed the current data of the internal circuits in the office building to form a zero-crossing segmentation trajectory chain. Based on these two chains, the technicians further analyzed the temperature distribution inside the building, determined the main axis of the heat flow anomaly area, and generated a double-chain dynamic matching probability map. Next, by calculating the spatial aggregation degree and performing coupling analysis, the technicians created a continuous probability field that comprehensively reflects the structural characteristics of the office building. Finally, the technicians applied the direction-aware probability diffusion algorithm along the main axis to generate a dynamic co-occurrence probability field. These steps helped optimize the operation strategy of the office building's air conditioning system, improve energy use efficiency, and ensure the comfort of the indoor environment.

[0097] In summary, through steps 401 to 405, through a series of in-depth data analysis and technical means, an accurate understanding and description of the complex relationship between the set of boundary curvature change points and the spectral characteristics of current data have been achieved. This method not only improves the efficiency of energy management but also provides a solid foundation for formulating more intelligent and personalized energy-saving strategies. By carefully analyzing the interaction between heat and electricity inside the building, unnecessary energy losses can be effectively reduced, and the overall energy efficiency performance of the building can be improved. At the same time, this method enhances the ability to cope with extreme weather conditions, ensuring the comfort and stability of the indoor environment. In addition, this method also improves the response speed and accuracy of the system, making the energy management system more efficient and intelligent.

[0098] To further improve the matching accuracy between the temperature field scan data and the simulation results of the dynamic heat balance equation set, first, based on the spatial discretization of the temperature field scan data, simulated temperature field grid data is generated. Then, the temperature difference between the actual and simulated data is analyzed to generate a local compensation factor, and continuous compensation field data is generated through smooth diffusion. The compensation field data is coupled with the simulation results to iteratively correct the heat conduction term, generating a corrected dynamic heat balance equation set. Finally, by performing backpropagation analysis on the spatial gradient distribution of the temperature field scan data, compensated temperature field increment data is generated, and the simulated data is superimposed to generate the final three-dimensional temperature field reconstruction data. In some embodiments, step 102 of generating three-dimensional temperature field reconstruction data by compensating for the difference between the temperature field scan data and the simulation results of the dynamic heat balance equation set includes: 501. Based on the discrete sampling point distribution of the temperature field scan data, the simulation results of the dynamic heat balance equation set are spatially discretized to generate simulated temperature field grid data; In step 501, the discrete sampling point distribution of the temperature field scan data refers to the actual temperature values at various positions inside the building collected from the actual environment. These data provide the true situation of the temperature field. The simulated temperature field grid data is generated by spatially discretizing the results of the dynamic heat balance equation set and is used to represent the temperature distribution at each point inside the building under simulated conditions. The simulated temperature field grid data provides a basic reference for subsequent steps.

[0099] In the embodiments of the present application, first, the temperature field scan data from the temperature sensor network is collected and combined with the known dynamic heat balance equation set. Then, numerical simulation techniques such as the finite element method are used to solve the dynamic heat balance equation to obtain a preliminary temperature distribution prediction. Then, this prediction result is discretized according to the spatial coordinates to form a temperature field grid data composed of multiple small grid cells. Finally, by integrating the information of these grid cells, a simulated temperature field grid data that comprehensively reflects the environmental temperature distribution is generated.

[0100] 502. By analyzing the temperature difference between the temperature field scan data and the simulated temperature field grid data, a local compensation factor is obtained, and it is smoothly diffused according to the spatial continuity constraint to generate continuous compensation field data; In step 502, the local compensation factor is calculated by comparing the temperature difference between the temperature field scan data and the simulated temperature field grid data and is used to correct the errors in the simulation results. The spatial continuity constraint means that when generating the continuous compensation field data, it is required that the compensation factor maintains a smooth transition throughout the space and avoids sudden changes. The continuous compensation field data is a distribution map of the compensation factor after smooth diffusion processing and is used to adjust the simulated temperature field grid data.

[0101] In the embodiments of the present application, first, the temperature difference in each grid cell is calculated to identify the areas that need to be corrected. Then, using the interpolation algorithm and the smoothing filter, the local compensation factor is smoothly diffused based on the spatial continuity constraint. Then, these compensation factors are applied to the original simulated temperature field grid data to generate a more accurate continuous compensation field data. Finally, this data can effectively improve the errors in the initial simulation results.

[0102] 503. Couple and iterate the continuous compensation field data with the simulation results of the dynamic heat balance equation set, and perform reverse correction on the heat conduction term of the dynamic heat balance equation set to generate a corrected dynamic heat balance equation set; In step 503, coupled iteration refers to the process of combining continuous compensation field data with the simulation results of a set of dynamic heat balance equations and performing multiple iterative optimizations. Reverse correction of the heat conduction term means adjusting the parameters in the heat conduction equation according to the compensation field data to improve the accuracy of the model. The corrected set of dynamic heat balance equations is a new set of equations generated after multiple iterative optimizations and is used to more accurately simulate the temperature changes inside the building.

[0103] In the embodiments of the present application, first, the continuous compensation field data is coupled and iteratively calculated with the simulation results of the original set of dynamic heat balance equations to gradually adjust the matching degree between the two. Then, based on the error patterns found during the iteration process, the heat conduction term in the set of dynamic heat balance equations is reversely corrected. Then, this process is repeated until the predetermined accuracy requirement is met, thereby generating a corrected set of dynamic heat balance equations. This process improves the adaptability and accuracy of the model to the actual situation.

[0104] 504. Perform backpropagation analysis on the spatial gradient distribution of the temperature field scan data according to the corrected set of dynamic heat balance equations to generate compensated temperature field increment data, and generate three-dimensional temperature field reconstruction data by superimposing the simulated temperature field grid data.

[0105] In step 504, the compensated temperature field increment data is data generated by performing backpropagation analysis on the spatial gradient distribution of the temperature field scan data and is used to correct the errors in the simulated temperature field grid data. The three-dimensional temperature field reconstruction data is the final temperature distribution map generated by superimposing the compensated temperature field increment data and the simulated temperature field grid data and can more accurately reflect the actual temperature distribution inside the building.

[0106] In the embodiments of the present application, first, according to the corrected set of dynamic heat balance equations, perform backpropagation analysis on the spatial gradient distribution of the temperature field scan data to determine the temperature field increments that need to be supplemented. Then, superimpose these increment data on the original simulated temperature field grid data to construct a more accurate three-dimensional temperature field reconstruction data. Finally, using this improved data, the true temperature distribution of the environment can be more accurately reflected. This process not only improves the accuracy of the simulation results but also ensures that the temperature field reconstruction data can truly reflect the actual temperature distribution inside the building.

[0107] The following is a specific example: In a large commercial building project, technicians first deployed a network of temperature sensors throughout the building to collect temperature field scan data. Subsequently, the finite element method was used to solve the dynamic heat balance equation to generate simulated temperature field grid data. Then, by analyzing the differences between the actual data and the simulated data, the local compensation factor was calculated and applied to create continuous compensation field data. On this basis, the technicians further optimized the set of dynamic heat balance equations by repeatedly adjusting the model parameters through iteration until a satisfactory revised version was obtained. Finally, the original data was enhanced using the revised model to generate three-dimensional temperature field reconstruction data with more details. This not only helps improve the efficiency of the building's internal environment control system but also provides a scientific basis for energy-saving renovation.

[0108] In summary, through steps 501 to 504, through a series of in-depth data analysis and technical means, an accurate understanding and description of the internal temperature field of the building have been achieved. This method not only improves the accuracy of temperature field simulation but also provides a solid foundation for formulating more intelligent and personalized temperature control strategies. By carefully analyzing the heat transfer process inside the building, the overall energy efficiency performance of the building is improved. At the same time, this method enhances the ability to cope with extreme weather conditions, ensuring the comfort and stability of the indoor environment. In addition, this method also improves the response speed and accuracy of the system, making the temperature control management system more efficient and intelligent. Through accurate temperature field reconstruction, the heat flow distribution inside the building can be better understood, thereby optimizing energy use and improving the comfort of the occupants.

[0109] To further improve the matching accuracy between the continuous compensation field data and the simulation results of the set of dynamic heat balance equations, a method based on node correction coefficients was designed. First, the cross-node conductivity was weighted and balanced using the heat conduction path weight factor to generate the distribution of node correction coefficients. Then, this distribution was inversely superimposed with the heat conduction term to generate an intermediate correction equation set, resulting in a revised set of dynamic heat balance equations. This method improves the accuracy and adaptability of the model. In some embodiments, step 503, which couples and iterates the continuous compensation field data with the simulation results of the set of dynamic heat balance equations and inversely corrects the heat conduction term of the set of dynamic heat balance equations to generate a revised set of dynamic heat balance equations, includes: 601. Generate node correction coefficients based on the amplitude and direction of the compensation amount of the continuous compensation field data, where the absolute value of the node correction coefficient is positively correlated with the amplitude of the continuous compensation field data; In step 601, the continuous compensation field data is a distribution map of compensation factors after smoothing and diffusion processing. The magnitude of the compensation amount refers to the size of the compensation factor, and the direction refers to the direction of the temperature difference. The node correction coefficient is a coefficient generated based on the magnitude and direction of the compensation amount for adjusting the parameters of the heat conduction equation, and its absolute value is proportional to the magnitude of the compensation amount. The node correction coefficient is used to correct the errors in the simulation results to make the simulation closer to the actual situation.

[0110] In the embodiment of the present application, first, the compensation amount and its direction required for each node are determined based on the continuous compensation field data. Then, according to the magnitude and direction of these compensation amounts, the corresponding correction coefficients are calculated for each node. Next, these correction coefficients are applied to the corresponding nodes to ensure that they can accurately reflect the actual temperature change. Finally, by integrating the correction coefficients of all nodes, a distribution map of node correction coefficients that comprehensively reflects the temperature field correction requirements is generated.

[0111] 602. Define a heat conduction path weight factor according to the heat conduction path topological relationship of the dynamic heat balance equation set, and perform conductivity weighted equilibrium across nodes on the node correction coefficient to generate a node correction coefficient distribution; In step 602, the heat conduction path topological relationship describes the spatial structure of the heat transfer path inside the building. The heat conduction path weight factor is different weight values assigned to different paths to reflect the importance of the paths. The node correction coefficient distribution is a distribution map generated by performing conductivity weighted equilibrium across nodes on the node correction coefficient. This distribution map considers the mutual influence between different paths and provides a basis for subsequent correction.

[0112] In the embodiment of the present application, first, analyze the heat conduction path topological relationship in the dynamic heat balance equation set to identify the main heat conduction paths. Then, assign corresponding weight factors according to the importance of these paths. Next, use these weight factors to recalculate the correction coefficients of each node to ensure that heat can be evenly distributed throughout the system. Finally, through this process, a distribution map of node correction coefficients after conductivity weighted equilibrium is generated, improving the overall stability of the system.

[0113] 603. Reverse superimpose the node correction coefficient distribution and the heat conduction term of the dynamic heat balance equation set to generate an intermediate correction equation set, and extract the heat flow residual parameters of the intermediate correction equation set and the dynamic heat balance equation set; In step 603, the reverse superposition of the heat conduction term means combining the node correction coefficient distribution and the heat conduction term of the dynamic heat balance equation set to generate a new equation set. The intermediate correction equation set is the new equation set generated through reverse superposition. The heat flow residual parameter is extracted from the intermediate correction equation set and is an index used to measure the difference between the equation set and the actual temperature field scan data.

[0114] In the embodiments of the present application, first, the node correction coefficient distribution is combined with the heat conduction term of the dynamic heat balance equation set to form an intermediate correction equation set. Then, the heat flux residual parameters are extracted from this new equation set to evaluate the correction effect. Next, these residual parameters are used to guide the subsequent adjustment work to ensure that the final result is as close as possible to the actual situation. Finally, an intermediate correction equation set that can more accurately describe the system behavior is generated through the above process.

[0115] 604. The node correction coefficient distribution is iteratively readjusted through the heat flux residual parameters until the heat flux residual parameters satisfy the heat flux conservation constraint condition of the dynamic heat balance equation set, and a corrected dynamic heat balance equation set is generated.

[0116] In step 604, the heat flux residual parameter is an index extracted to measure the difference between the equation set and the actual data. The iterative readjustment means that the node correction coefficient distribution is adjusted multiple times according to the heat flux residual parameter until the heat flux conservation constraint condition is satisfied. The corrected dynamic heat balance equation set is a new equation set generated after multiple iterations of optimization and is used to more accurately simulate the temperature change inside the building.

[0117] In the embodiments of the present application, first, the node correction coefficient distribution is initially adjusted according to the heat flux residual parameter. Then, this process is repeated multiple times, and each adjustment is based on the latest residual parameter to gradually narrow the error range. Next, it is checked whether the heat flux conservation constraint condition is satisfied during each iteration process. Finally, when the predetermined accuracy requirement is met, a corrected dynamic heat balance equation set is generated, and this equation set can more accurately simulate the heat conduction behavior of the actual system.

[0118] The following is a specific example: In a large data center project, technicians first generate continuous compensation field data based on the data collected by the temperature sensor network and calculate the correction coefficients of each cooling equipment node accordingly. Then, the topological relationship of the heat conduction path inside the data center is analyzed, and appropriate heat conduction path weight factors are defined to perform conductivity weighted balancing processing on the node correction coefficients. Subsequently, the adjusted node correction coefficient distribution is combined with the heat conduction term of the dynamic heat balance equation set to generate an intermediate correction equation set, and the heat flux residual parameters are extracted from it. Finally, the node correction coefficient distribution is adjusted multiple times until the heat flux residual parameters satisfy the heat flux conservation constraint condition, effectively improving the efficiency and stability of the data center cooling system.

[0119] In summary, the accurate understanding and description of the internal temperature field of a building are achieved through steps 601 to 604. This method not only improves the accuracy of temperature field simulation but also provides a solid foundation for formulating more intelligent and personalized temperature control strategies. By carefully analyzing the heat transfer process inside the building, the overall energy efficiency performance of the building is enhanced. At the same time, this method improves the ability to cope with extreme weather conditions, ensuring the comfort and stability of the indoor environment. In addition, this method also improves the response speed and accuracy of the system, making the temperature control management system more efficient and intelligent. Through precise temperature field reconstruction and correction of the dynamic heat balance equation set, the heat flow distribution inside the building can be better understood, thereby optimizing energy use and improving the comfort of the occupants.

[0120] To further improve the intelligence and personalization of the temperature control device regulation strategy, the solution constructs a priority regulation chain for the temperature control device, using a method that combines spatio-temporal coupling features and meteorological forecast data. First, the spatial thermal inertia index and the time response sensitivity index of the temperature control device are extracted, and a preliminary priority sequence is generated by combining the topological relationship of the heat conduction path. Then, the regulation weight is corrected by the matching degree between the coverage area and the device location to generate a dynamic priority sequence. Finally, the device nodes that violate the energy consumption or stability threshold are iteratively screened to construct a priority regulation chain. This method improves the response speed and accuracy of the system. In some embodiments, step 104 of constructing a priority regulation chain for the temperature control device based on the spatio-temporal coupling features and meteorological forecast data includes: 701. Extract the spatial thermal inertia index and the time response sensitivity index of the temperature control device according to the correlation between the distributed thermal inertia parameter of the temperature control device in the spatio-temporal coupling feature and the temperature fluctuation parameter in the future time period of the meteorological forecast data; In step 701, the spatio-temporal coupling feature describes the relationship pattern of the device power consumption and the thermal environment parameters changing with time and space. The distributed thermal inertia parameter is an index that measures the reaction speed of different positions in the building to temperature changes. The temperature fluctuation parameter in the future time period of the meteorological forecast data provides the temperature prediction information for a period of time in the future. The spatial thermal inertia index reflects the resistance of each point inside the building to temperature changes, and the time response sensitivity index represents the response speed of the temperature control device to temperature changes.

[0121] In the embodiments of the present application, first, the spatial distribution information and its thermal inertia parameters of the temperature control device are collected, and the meteorological forecast data for the future time period is obtained. Then, through data analysis algorithms (such as regression analysis), the spatial thermal inertia index and the time response sensitivity index of each temperature control device are extracted. Then, based on these indexes, the adaptability of each device to future temperature fluctuations is evaluated. Finally, all the information is integrated to form a data set that comprehensively reflects the adaptability and response speed of the temperature control device.

[0122] 702. Based on the correlation between the spatial thermal inertia index and the time response sensitivity index, establish the regulatory influence weight of the temperature control device, and combine the topological relationship of the heat conduction path among the temperature control devices to generate a preliminary priority sequence. In step 702, the regulatory influence weight is established based on these two indices and is used to evaluate the influence of each temperature control device during the regulation process. The topological relationship of the heat conduction path describes the heat transfer path among the temperature control devices. The preliminary priority sequence is generated according to the above parameters and relationships to guide the priority ranking of the temperature control devices.

[0123] In the embodiment of the present application, first calculate the regulatory influence weight according to the spatial thermal inertia index and the time response sensitivity index. Then, analyze the topological relationship of the heat conduction path among the temperature control devices to identify the key nodes. Next, use graph theory algorithms to assign priorities to each device based on the regulatory influence weight. Finally, generate a preliminary priority sequence that takes into account the interactions among the devices and their importance to the system temperature control.

[0124] 703. According to the spatio-temporal distribution characteristics of extreme temperature events in the meteorological forecast data, dynamically adjust the preliminary priority sequence, and correct the regulatory influence weight through the matching degree between the area covered by the extreme temperature events and the spatial positions of the temperature control devices to generate a dynamic priority sequence. In step 703, extreme temperature events refer to significant high or low temperature weather that may occur in a certain future period. The spatio-temporal distribution characteristics describe the time and location of these extreme events. The regulatory influence weight is further adjusted based on this, and a dynamic priority sequence is generated by considering the matching degree between the area covered by the extreme temperature events and the spatial positions of the temperature control devices. This sequence is more in line with the actual needs and improves the ability to cope with extreme weather.

[0125] In the embodiment of the present application, first analyze the meteorological forecast data to identify possible future extreme temperature events and their specific locations and times. Then, compare the coverage of these events with the positions of the temperature control devices to evaluate the affected degree. Next, adjust the regulatory influence weight of each device according to the evaluation results and re-rank to generate a dynamic priority sequence. Finally, ensure that this sequence can effectively cope with the upcoming extreme temperature challenges. This process helps to identify the temperature control devices that most need to be regulated, enhances the system's ability to cope with extreme weather, and ensures the efficient operation of the system.

[0126] 704. By iteratively screening the temperature control device nodes in the dynamic priority sequence that violate the energy consumption threshold or the thermal stability threshold, construct a priority regulation chain for the temperature control devices.

[0127] In step 704, the dynamic priority sequence is the adjusted device priority sorting generated in step 703. The energy consumption threshold and the thermal stability threshold are respectively the set upper limit of energy consumption and the minimum temperature requirement for maintaining indoor comfort. By iteratively screening the temperature control device nodes in the dynamic priority sequence that violate the above thresholds, a priority regulation chain for temperature control devices is constructed. This regulation chain ensures the efficient operation of the system while meeting the user's comfort requirements.

[0128] In the embodiments of the present application, first, the energy consumption and thermal stability thresholds are set as the screening criteria. Then, each device node in the dynamic priority sequence is checked one by one to see if it meets these conditions. Then, the nodes that do not meet the conditions are marked and removed from the sequence or their priorities are rearranged. Finally, a priority regulation chain that meets all the criteria is constructed to ensure the efficient and stable operation of the entire system. The priority regulation chain provides a guarantee for the optimized operation of the temperature control system.

[0129] The following is a specific example: In an office building project, the technical staff first extracted the spatial thermal inertia index and the time response sensitivity index of each device according to the spatial distribution and thermal inertia parameters of the temperature control devices, combined with the meteorological forecast data for the future period. Then, based on these indexes, a regulation influence weight was established, and considering the topological relationship of the heat conduction paths between the temperature control devices, a preliminary priority sequence was generated. Subsequently, the extreme temperature events in the meteorological forecast data were analyzed to dynamically adjust the preliminary priority sequence to better cope with extreme weather conditions. Finally, through an iterative screening process, the device nodes that violate the energy consumption or thermal stability thresholds were removed, and an efficient priority regulation chain was constructed, significantly improving the response speed and energy utilization efficiency of the temperature control system in the building.

[0130] In summary, through steps 701 to 704, through a series of in-depth data analyses and technical means, an accurate understanding and optimization of the regulation strategy for temperature control devices are achieved. This method not only improves the response speed and accuracy of the temperature control system, but also provides a solid foundation for formulating more intelligent and personalized energy-saving strategies. By carefully analyzing the internal heat transfer process and external meteorological conditions of the building, unnecessary energy losses can be effectively reduced, and the overall energy efficiency performance of the building can be improved. At the same time, this method enhances the ability to cope with extreme weather conditions, ensuring the comfort and stability of the indoor environment. In addition, this method also improves the response speed and accuracy of the system, making the temperature control management system more efficient and intelligent. Through an accurate priority regulation chain, energy consumption can be better managed, and the satisfaction of the occupants can be improved.

[0131] Figure 2 The structural schematic diagram of a big data information processing system based on deep learning is provided for the embodiments of the present application, as Figure 2As shown, the system includes: An acquisition module 21 that acquires big data information of a building, where the big data information includes envelope thermal resistance parameters and spatial topology data extracted through a building information model, and combines a thermal distribution map collected by an infrared thermal imaging array to generate an energy distribution model with thermal property markings; A generation module 22 that constructs a set of dynamic heat balance equations based on the heat conduction path characteristics in the energy distribution model, and generates three-dimensional temperature field reconstruction data through differential compensation between temperature field scan data and the simulation results of the set of dynamic heat balance equations; An extraction module 23 that inputs the thermal resistance parameters in the energy distribution model, the coordinates of the heat flow anomaly area in the three-dimensional temperature field reconstruction data, and the current data of the air conditioning unit into an event feature extraction network driven by deep learning to extract the spatio-temporal coupling characteristics of equipment power consumption and thermal environment parameters; The generation module is further configured to construct a priority control chain for temperature control equipment based on the spatio-temporal coupling characteristics and meteorological forecast data, and generate a multi-level linkage rule including fresh air unit ventilation rate adjustment, lighting dimming gradient optimization, and chilled water storage tank cold release trigger; and dynamically generate a power load suppression strategy for the building based on the sensitive node parameters of the priority control chain.

[0132] Figure 2 The described big data information processing system based on deep learning can execute Figure 1 The described big data information processing method based on deep learning in the illustrated embodiment, and its implementation principle and technical effects will not be elaborated further. For the big data information processing system based on deep learning in the above embodiment, the specific ways in which each module and unit perform operations have been described in detail in the embodiment related to the method, and will not be elaborated here.

[0133] In a possible design, Figure 2 The big data information processing system based on deep learning in the illustrated embodiment can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32; The storage component 31 stores one or more computer instructions, where the one or more computer instructions are called and executed by the processing component 32.

[0134] The processing component 32 is used for the Figure 1 Described big data information processing method in the above embodiment.

[0135] Among them, the processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above methods. Of course, the processing component may also be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components for executing the above methods.

[0136] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0137] Of course, the computing device may also necessarily include other components, such as input / output interfaces, display components, communication components, etc.

[0138] The input / output interface provides an interface between the processing component and the peripheral interface module, and the above peripheral interface module may be an output device, an input device, etc.

[0139] The communication component is configured to facilitate communication between the computing device and other devices in a wired or wireless manner, etc.

[0140] Among them, the computing device may be a physical device or an elastic computing host provided by a cloud computing platform, etc. At this time, the computing device may refer to a cloud server, and the above processing component, storage component, etc. may be basic server resources leased or purchased from a cloud computing platform.

[0141] The embodiment of the present application also provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it can implement the above Figure 1 shown embodiment of a big data information processing method based on deep learning.

[0142] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0143] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative work.

[0144] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0145] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present application.

Claims

1. A big data information processing method based on deep learning, characterized in that: include: Obtaining big data information of buildings, including thermal resistance parameters and spatial topology data of enclosure structures extracted through building information models, combined with thermal distribution maps collected by infrared thermal imaging arrays, to generate an energy distribution model with thermal attribute markers; Based on the heat conduction path characteristics in the energy distribution model, a dynamic heat balance equation set is constructed, and three-dimensional temperature field reconstruction data is generated by compensating the difference between the temperature field scanning data and the simulation results of the dynamic heat balance equation set; Input the thermal resistance parameters in the energy distribution model, the coordinates of the heat flow abnormal area in the three-dimensional temperature field reconstruction data, and the current data of the air-conditioning unit into the event feature extraction network driven by deep learning to extract the spatiotemporal coupling characteristics of equipment power consumption and thermal environment parameters; Based on the spatiotemporal coupling characteristics and meteorological forecast data, a priority control chain of temperature control equipment is constructed to generate multi-level linkage rules including ventilation rate adjustment of fresh air units, lighting dimming gradient optimization, and cold storage tank release triggering; According to the sensitive node parameters of the priority regulation chain, a power load balancing strategy for the building is dynamically generated.

2. The method according to claim 1, characterized in that The event feature extraction network driven by deep learning extracts the spatiotemporal coupling features of device power consumption and thermal environment parameters, including: The thermal resistance parameters in the energy distribution model are superimposed with the boundary data of the heat flow anomaly area in the three-dimensional temperature field reconstruction data to generate a spatial correlation map with the heat conduction hysteresis effect; Based on the phase fluctuation characteristics of the current data of the air-conditioning unit, the thermal disturbance propagation path that is strongly related to the operation of the electrical equipment is marked in the spatial correlation map; According to the spatial coverage of the thermal disturbance propagation path, the thermal resistance parameter is attenuated and weighted according to the conduction path length, and the air flow characteristic data in the three-dimensional temperature field reconstruction data is superimposed to generate a thermal-electric coupling factor; Based on the three-dimensional distribution density of the thermal-electric coupling factor, a joint response sequence is constructed, wherein the joint response sequence generates multi-dimensional association rules by analyzing the co-occurrence probability of the current data spectrum characteristics and the geometric topology of the thermal flow anomaly area; The joint response sequence is cross-domain matched with the multi-dimensional association rules, and the spatiotemporal coupling characteristics of the device power consumption and thermal environment parameters are extracted.

3. The method according to claim 2, characterized in that The joint response sequence is constructed, and the joint response sequence generates multi-dimensional association rules by analyzing the co-occurrence probability of the current data spectrum characteristics and the geometric topology of the heat flow abnormal area, including: Based on the three-dimensional distribution density of the thermal-electric coupling factor, the dynamic correlation intensity of the geometric topology of the heat flow anomaly area and the spectrum characteristics of the current data in the spatial coordinates is superimposed in multiple channels to construct a joint response sequence; Based on the coupling relationship between the axial gradient distribution of the joint response sequence and the frequency domain energy attenuation rate, a set of boundary curvature change points of the geometric topology of the heat flow anomaly area is extracted; Matching the boundary curvature change point set with the zero-crossing point distribution of the current data spectrum feature to generate a dynamic co-occurrence probability field constrained by the main axis of the geometric topology of the heat flow anomaly area, wherein the dynamic co-occurrence probability field includes a nonlinear mapping relationship between the zero-crossing point density and the boundary curvature change rate; Based on the synchronous change interval of the thermal fluctuation amplitude of the main axis and the frequency domain energy attenuation rate, a multi-dimensional association rule is generated by constraining the boundary curvature change threshold and the gradient range of the zero-crossing point density.

4. The method according to claim 3, characterized in that The step of matching the boundary curvature change point set with the zero-crossing point distribution of the current data spectrum feature to generate a dynamic co-occurrence probability field constrained by the main axis of the geometric topology of the heat flow anomaly area includes: Performing gradient direction clustering on the boundary curvature change point set to generate a curvature change point cluster group, and using the centroid connection line of the curvature change point cluster group as an initial matching reference chain; Performing annular neighborhood phase polarity analysis on the zero-crossing point distribution of the current data spectrum characteristics, screening out a zero-crossing point sequence, and segmenting the zero-crossing point sequence to generate a zero-crossing point segmented trajectory chain; According to the direction of the main axis of the geometric topology of the heat flow anomaly area, dynamically adjust the local curvature weight of the initial matching reference chain and the phase polarity weight of the zero-crossing segmented trajectory chain to generate a double-chain dynamic matching probability map; According to the spatial aggregation degree of the double-chain dynamic matching probability map, coupling analysis is performed on the clusters of curvature change points, and a continuous probability field is generated by superimposing the probability density gradients of adjacent clusters on the main axis; Direction-aware probability diffusion is performed on the continuous probability field, anisotropic diffusion constraints are applied along the main axis, and a dynamic co-occurrence probability field is generated according to the distribution density of the double-chain dynamic matching probability map.

5. The method according to claim 1, characterized in that The method of compensating the difference between the temperature field scanning data and the simulation results of the dynamic heat balance equation set to generate three-dimensional temperature field reconstruction data includes: Based on the discrete sampling point distribution of the temperature field scanning data, the simulation results of the dynamic heat balance equation set are spatially discretized to generate simulated temperature field grid data; The local compensation factor is obtained by analyzing the temperature difference between the temperature field scanning data and the simulated temperature field grid data, and the continuous compensation field data is generated by smooth diffusion according to the spatial continuity constraint; Iteratively coupling the continuous compensation field data with the simulation results of the dynamic heat balance equation set, and reversely correcting the heat conduction term of the dynamic heat balance equation set to generate a corrected dynamic heat balance equation set; The spatial gradient distribution of the temperature field scanning data is subjected to back propagation analysis according to the modified dynamic heat balance equation set to generate compensated temperature field increment data, and three-dimensional temperature field reconstruction data is generated by superimposing the simulated temperature field grid data.

6. The method according to claim 5, characterized in that The coupling and iteration of the continuous compensation field data and the simulation results of the dynamic heat balance equation set, and reverse correction of the heat conduction term of the dynamic heat balance equation set to generate a corrected dynamic heat balance equation set, includes: Generate a node correction coefficient based on the compensation amount amplitude and direction of the continuous compensation field data, wherein the absolute value of the node correction coefficient is positively correlated with the amplitude of the continuous compensation field data; According to the heat conduction path topological relationship of the dynamic heat balance equation set, a heat conduction path weight factor is defined, and the node correction coefficient is weightedly balanced for the conductivity across nodes to generate a node correction coefficient distribution; Reversely superimpose the node correction coefficient distribution and the heat conduction term of the dynamic heat balance equation set to generate an intermediate correction equation set, and extract heat flow residual parameters of the intermediate correction equation set and the dynamic heat balance equation set; The node correction coefficient distribution is iteratively readjusted by using the heat flow residual parameter until the heat flow residual parameter satisfies the heat flow conservation constraint condition of the dynamic heat balance equation set, thereby generating a corrected dynamic heat balance equation set.

7. The method according to claim 1, characterized in that The priority control chain of the temperature control equipment is constructed based on the spatiotemporal coupling characteristics and the meteorological forecast data, including: Extracting the spatial thermal inertia index and the time response sensitivity index of the temperature control device according to the correlation between the distributed thermal inertia parameters of the temperature control device in the spatiotemporal coupling characteristics and the temperature fluctuation parameters of the future period in the meteorological forecast data; Based on the correlation between the spatial thermal inertia index and the time response sensitivity index, the control influence weight of the temperature control device is established, and the preliminary priority sequence is generated in combination with the topological relationship of the heat conduction path between the temperature control devices; According to the spatiotemporal distribution characteristics of extreme temperature events in the meteorological forecast data, the preliminary priority sequence is dynamically adjusted, and the control influence weight is corrected by the matching degree between the extreme temperature event coverage area and the spatial position of the temperature control equipment to generate a dynamic priority sequence; By iteratively screening the temperature control device nodes that violate the energy consumption threshold or the thermal stability threshold in the dynamic priority sequence, a priority regulation chain of the temperature control device is constructed.

8. A big data information processing system based on deep learning, characterized in that: include: An acquisition module is used to acquire big data information of buildings, wherein the big data information includes thermal resistance parameters and spatial topology data of enclosure structures extracted through building information models, combined with thermal distribution maps collected by infrared thermal imaging arrays, to generate an energy distribution model with thermal attribute markers; A generation module, for constructing a set of dynamic heat balance equations based on the heat conduction path characteristics in the energy distribution model, and generating three-dimensional temperature field reconstruction data by compensating for the difference between the temperature field scanning data and the simulation results of the set of dynamic heat balance equations; An extraction module, used to input the thermal resistance parameters in the energy distribution model, the coordinates of the heat flow abnormal area in the three-dimensional temperature field reconstruction data, and the current data of the air-conditioning unit into the event feature extraction network driven by deep learning, and extract the spatiotemporal coupling characteristics of the equipment power consumption and thermal environment parameters; The generation module is also used to build a priority control chain of temperature control equipment based on the spatiotemporal coupling characteristics and meteorological forecast data, and generate multi-level linkage rules including ventilation rate adjustment of fresh air units, lighting dimming gradient optimization and cold storage tank release triggering; according to the sensitive node parameters of the priority control chain, the power load balancing strategy of the building is dynamically generated.

9. A computing device, characterized in that It comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a big data information processing method based on deep learning as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a computer, a big data information processing method based on deep learning as described in any one of claims 1 to 7 is implemented.

Images