Building energy efficiency management platform based on AIoT digital twinning

Abstract

The application relates to the technical field of intelligent building energy consumption optimization control, and discloses a building energy efficiency management platform based on AIoT digital twinning, which comprises a heterogeneous data acquisition module, a building model construction module, a reinforcement learning optimization module, a fault prediction management module and a safety control execution module. Compared with the energy consumption management mode in the prior art which mainly relies on static rules or scattered subsystem control, especially in the scene where the building structure is complex, the equipment is highly heterogeneous, and the data is highly inconsistent, the technical problem that global energy consumption optimization and precise control cannot be realized is solved. Since the application constructs a multi-source fusion data processing mechanism, a high-precision digital twinning model and a self-learning reinforcement control strategy, dynamic energy efficiency collaborative scheduling among building multiple systems is realized, thereby avoiding the problem that energy is wasted due to control imbalance and equipment fault response lag in the traditional system, and the intelligent level, system stability and energy saving efficiency of energy consumption management are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent building energy consumption optimization and control technology, and in particular to a building energy efficiency management platform based on AIoT digital twin. Background Technology

[0002] Currently, building energy management primarily relies on traditional Building Automation Systems (BAS) or Building Energy Management Systems (BEMS). These systems typically use zoned sensor networks and rule-based control logic as their core to monitor and control the operational status of subsystems such as HVAC, lighting, and elevators. However, these systems still have shortcomings in real-world environments with complex building structures, diverse equipment, and heterogeneous data sources. For example, inconsistent communication protocols between different equipment manufacturers make data sharing difficult; energy consumption analysis methods often rely on experience-based threshold setting rather than dynamic modeling, failing to achieve global optimization control based on current status and predicted trends. Furthermore, traditional systems lack real-time assessment and fault prediction capabilities for equipment health, often resulting in delayed responses to equipment anomalies, leading to energy waste and decreased comfort. Existing technologies cannot fully meet the comprehensive energy efficiency management needs of new green buildings and intelligent buildings, especially in situations with widely distributed equipment, dynamically changing operational status, and numerous uncertainties. Traditional methods struggle to accurately control building energy consumption or provide timely warnings of system risks. Therefore, there is an urgent need for a new energy efficiency management platform that integrates AIoT sensing, digital twin modeling, reinforcement learning optimization, and predictive maintenance technologies. This platform should be able to dynamically model the building's operating status and optimize control strategies even in a multi-source heterogeneous data environment, thereby improving the building's overall energy efficiency, operational safety, and intelligent control level. Summary of the Invention

[0003] To address the aforementioned technical shortcomings, the present invention aims to propose a building energy efficiency management platform and method based on AIoT digital twins. This aims to solve the technical problem that existing energy consumption management methods, which mainly rely on static rules or decentralized subsystem control, cannot achieve global energy consumption optimization and precise control, especially in scenarios with complex building structures, high equipment heterogeneity, and strong data inconsistency.

[0004] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a building energy efficiency management platform based on AIoT digital twins.

[0005] The AIoT-based digital twin-based building energy efficiency management platform includes:

[0006] The heterogeneous data acquisition module is used to collect operational data of AIoT devices in the building, including time series data, spatial structure data and text description data. Based on the weighted minimum variance estimation method, the operational data of the devices is standardized and fused to obtain a multi-source data fusion vector.

[0007] The building model building module is used to build a digital twin model of a building based on the first-order differential modeling method of heat conduction, and takes the multi-source data fusion vector as the input of the digital twin model of the building and outputs the current energy consumption state vector.

[0008] The reinforcement learning optimization module is used to input the current energy consumption state vector into a preset Q-learning optimization control model and output an optimized action strategy. It also generates a set of control instructions and outputs them to the AIoT device control interface;

[0009] The fault prediction and management module uses a Long Short-Term Memory (LSTM) network to model the multi-source data fusion vector, predict the multi-source data fusion vector for the next day's time step, and calculate the residual based on the multi-source data fusion vectors from the previous two days. When the residuals are satisfied at k consecutive time points Greater than the preset residual threshold If the fault is detected, a fault warning state is determined, and a fault adjustment factor is generated. This feedback is then fed back to the Q-learning optimization control model's action space in the reinforcement learning optimization module to shield fault channels;

[0010] The safety control execution module is used to optimize action strategies. and fault adjustment factor The closed-loop control command is constructed using the national cryptographic SM2 and TLS protocols, and the closed-loop control command is sent to each AIoT device node.

[0011] Preferably, in the heterogeneous data acquisition module, the time series data includes energy consumption time series and equipment operating parameter time series; the spatial structure data comes from the preset BIM model; and the text description data includes equipment manuals and maintenance records.

[0012] Preferably, in the safety control execution module, based on the optimized action strategy... and fault adjustment factor In the steps of constructing closed-loop control instructions using the national cryptographic SM2 and TLS protocols, device authentication in the encryption process of the national cryptographic SM2 and TLS protocols uses hash signatures. Tamper-proof processing is performed, and all data is written to the blockchain distributed ledger.

[0013] Preferably, in the building model construction module, the step of constructing a digital twin model of the building based on the first-order differential modeling method of heat conduction uses the following formula: ,in, Let the air heat capacity be the i-th partition. Let be the indoor temperature at time t in the i-th partition; The temperature of other partitions j adjacent to partition i; Outdoor temperature; This is the heating / cooling input for the HVAC system in zone i; Input for heat sources within the partition; For the thermal resistance of the interval; To reduce thermal resistance to the outside.

[0014] Preferably, in the reinforcement learning optimization module, the current energy consumption state vector is input into a preset Q-learning optimization control model, and the optimized action strategy is output. The steps specifically include: constructing the state space input vector of the Q-learning optimization control model based on the current energy consumption state vector output by the building digital twin model. ; Obtain the controllable parameters of the AIoT device and construct an action space set A, where each action in action space set A contains a set of control instructions, including zone temperature control instructions, fan speed control instructions, and lighting brightness level control instructions; Set an instant reward function. Instant reward function This is used to measure the energy efficiency of each action in the action space set A; in the Q-learning optimization control model, the Bellman optimality principle is adopted based on the state space input vector. and the action space set A and the instant reward function Perform iterative updates and output optimized action strategies. .

[0015] Preferably, in the fault prediction management module, the residual is calculated based on the fusion vector of multi-source data from the two days prior and subsequent days. The steps involved a method combining mean squared error analysis with a sliding window.

[0016] Preferably, in the heterogeneous data acquisition module, the standardization and fusion processing includes data cleaning, unit normalization, categorical variable encoding and embedding, and weighted minimum variance estimation fusion processing.

[0017] This invention also provides a building energy efficiency management method based on AIoT digital twins, comprising:

[0018] Step S10: Collect AIoT device operation data in the building, including time series data, spatial structure data and text description data, and perform standardization and fusion processing on the device operation data based on the weighted minimum variance estimation method to obtain a multi-source data fusion vector;

[0019] Step S20: Construct a digital twin model of the building based on the first-order differential modeling method of heat conduction, and use the multi-source data fusion vector as the input of the digital twin model of the building, and output the current energy consumption state vector;

[0020] Step S30: Input the current energy consumption state vector into the preset Q-learning optimization control model and output the optimized action strategy. It also generates a set of control instructions and outputs them to the AIoT device control interface;

[0021] Step S40: Model the multi-source data fusion vector using a Long Short-Term Memory (LSTM) network, predict the multi-source data fusion vector for the next day's time step, and calculate the residual based on the multi-source data fusion vectors from the previous two days. When the residuals are satisfied at k consecutive time points Greater than the preset residual threshold If the fault is detected, a fault warning state is determined, and a fault adjustment factor is generated. This feedback is then fed back to the Q-learning optimization control model's action space in the reinforcement learning optimization module to shield fault channels;

[0022] Step S50: Based on the optimized action strategy and fault adjustment factor The closed-loop control command is constructed using the national cryptographic SM2 and TLS protocols, and the closed-loop control command is sent to each AIoT device node.

[0023] The present invention also provides a computer program product, including a building energy efficiency management program based on AIoT digital twin, wherein the building energy efficiency management program based on AIoT digital twin implements the building energy efficiency management method based on AIoT digital twin when executed by a processor.

[0024] The beneficial effects of this invention are as follows: Compared with the energy management methods in the prior art that mainly rely on static rules or decentralized subsystem control, especially in scenarios with complex building structures, high equipment heterogeneity, and strong data inconsistency, which cannot achieve global energy optimization and precise control, this application realizes dynamic energy efficiency collaborative scheduling among multiple building systems by constructing a multi-source fusion data processing mechanism, a high-precision digital twin model, and a self-learning enhanced control strategy. This avoids the problem of energy waste caused by control imbalance and equipment failure response lag in traditional systems, and improves the intelligence level, system stability, and energy-saving efficiency of energy management. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a schematic diagram of the platform of the first embodiment of the building energy efficiency management platform based on AIoT digital twin of the present invention.

[0027] Figure 2 This is a schematic diagram of a building energy efficiency management platform based on AIoT digital twin according to the present invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] Example 1: As Figure 1 The diagram shown is a flowchart of the first embodiment of the building energy efficiency management platform based on AIoT digital twin of the present invention, which presents the first embodiment of the building energy efficiency management platform based on AIoT digital twin of the present invention.

[0030] In the first embodiment, the AIoT digital twin-based building energy efficiency management platform includes:

[0031] The heterogeneous data acquisition module is used to collect operational data of AIoT devices in the building, including time series data, spatial structure data and text description data. Based on the weighted minimum variance estimation method, the operational data of the devices is standardized and fused to obtain a multi-source data fusion vector.

[0032] It should be noted that in the heterogeneous data acquisition module, time series data includes energy consumption time series and equipment operating parameter time series; spatial structure data comes from a preset BIM model; and text description data includes equipment manuals and maintenance records. Standardization and fusion processing includes data cleaning, unit normalization, categorical variable encoding and embedding, and weighted minimum variance estimation fusion processing. Specifically, standardization and fusion processing include: data cleaning: used to remove sensor drift values, missing values, and duplicate values; unit normalization: normalizing data from different dimensions to the [0,1] interval; categorical variable encoding and embedding: using one-hot encoding or word embedding models for text-type status fields; and weighted minimum variance estimation fusion processing: in multi-sensor redundancy scenarios, assigning different confidence weights to each channel based on historical fluctuation variance, and performing fusion estimation.

[0033] It should be understood that the aforementioned standardization and fusion processes not only improve the consistency and comparability of the original data, but also reduce the risk of data deviation caused by differences in equipment accuracy, communication delays, or sensor anomalies. This provides a highly robust and uniformly formatted input data foundation for subsequent building digital twin modeling and dynamic prediction of energy consumption status, enhancing the platform's adaptability to complex multi-source environments. The heterogeneous data acquisition module, by introducing a fusion mechanism for structured and unstructured data and a confidence-weighted strategy, not only achieves multi-dimensional perception of building internal operating data, but also significantly reduces the burden of feature alignment and anomaly handling in the subsequent modeling layer by moving data fusion forward (i.e., completing standardization and estimation at the acquisition layer).

[0034] The building model building module is used to build a digital twin model of a building based on the first-order differential modeling method of heat conduction, and takes the multi-source data fusion vector as the input of the digital twin model of the building and outputs the current energy consumption state vector.

[0035] It should be noted that in the building model building module, the formula used for constructing the building digital twin model based on the first-order differential modeling method of heat conduction is as follows: ,in, Let the air heat capacity be the i-th partition. Let be the indoor temperature at time t in the i-th partition; The temperature of other partitions j adjacent to partition i; Outdoor temperature; This is the heating / cooling input for the HVAC system in zone i; Input for heat sources within the partition; For the thermal resistance of the interval; To reduce thermal resistance to the outside.

[0036] Understandably, the above-mentioned first-order differential modeling method for heat conduction divides the building into multiple thermodynamic nodes. By establishing a state differential equation with heat capacity and thermal resistance as the core, it can simulate the dynamic process of heat exchange between different spaces and between the building and the external environment with high fidelity. The input parameters of the model all come from the multi-source data vector fused in step S10. Therefore, it can reflect the current thermal state and energy load distribution of the building in real time and accurately, providing physical modeling support for subsequent energy consumption prediction and strategy optimization.

[0037] It should be understood that the building model building module introduces a dynamic modeling mechanism based on the differential equation of heat conduction, which enables the digital twin model of a building to be no longer just a static geometric simulation or a regular mapping model, but to have the dynamic evolution capability of state differentiability, input controllability, and output predictability in continuous time.

[0038] The reinforcement learning optimization module is used to input the current energy consumption state vector into a preset Q-learning optimization control model and output an optimized action strategy. It also generates a set of control instructions and outputs them to the AIoT device control interface;

[0039] It should be noted that in the reinforcement learning optimization module, the current energy consumption state vector is input into the preset Q-learning optimization control model, and the optimized action strategy is output. The steps specifically include: constructing the state space input vector of the Q-learning optimization control model based on the current energy consumption state vector output by the building digital twin model. ; Obtain the controllable parameters of the AIoT device and construct an action space set A, where each action in action space set A contains a set of control instructions, including zone temperature control instructions, fan speed control instructions, and lighting brightness level control instructions; Set an instant reward function. Instant reward function This is used to measure the energy efficiency of each action in the action space set A; in the Q-learning optimization control model, the Bellman optimality principle is adopted based on the state space input vector. and the action space set A and the instant reward function Perform iterative updates and output optimized action strategies. .

[0040] Understandably, by constructing a Q-learning optimized control model, the building's energy consumption response characteristics and user comfort sensitivity can be learned through continuous interaction, achieving a shift from "static rule control" to "dynamic strategy generation." This module can intelligently adjust control variables such as air conditioning temperature, lighting intensity, and equipment activation strategies under different operating conditions, thereby minimizing energy consumption while meeting indoor comfort requirements. For example, during nighttime hours when there are few people and the environment is mild, it will learn strategies such as turning off some lighting areas and raising the air conditioning temperature setpoint to reduce energy consumption; while during peak hours, it prioritizes comfort and achieves refined control through zoned adjustments.

[0041] The fault prediction and management module uses a Long Short-Term Memory (LSTM) network to model the multi-source data fusion vector, predict the multi-source data fusion vector for the next day's time step, and calculate the residual based on the multi-source data fusion vectors from the previous two days. When the residuals are satisfied at k consecutive time points Greater than the preset residual threshold If the fault is detected, a fault warning state is determined, and a fault adjustment factor is generated. This feedback is then fed back to the Q-learning optimization control model's action space in the reinforcement learning optimization module to shield fault channels;

[0042] Understandably, this module, by integrating the temporal learning capabilities of LSTM with residual-driven anomaly detection logic, can proactively identify deviations in equipment performance based on "prediction-observation" trends before significant performance degradation or abnormal operating conditions occur. Compared to traditional alarm methods that rely on static threshold judgments, this module possesses dynamic, continuous, and intelligent predictive and early warning capabilities, effectively identifying potential fault trends and preventing sudden increases in energy consumption or control failures caused by abnormal equipment operation.

[0043] The safety control execution module is used to optimize action strategies. and fault adjustment factor The closed-loop control command is constructed using the national cryptographic SM2 and TLS protocols, and the closed-loop control command is sent to each AIoT device node.

[0044] It should be noted that in the safety control execution module, based on the optimized action strategy... and fault adjustment factor In the steps of constructing closed-loop control instructions using the national cryptographic SM2 and TLS protocols, device authentication in the encryption process of the national cryptographic SM2 and TLS protocols uses hash signatures. Tamper-proof processing is implemented, and all data is written to a blockchain distributed ledger. The ledger supports timestamps, tamper-proof mechanisms, and consensus mechanisms to ensure that instructions and operations are traceable and cannot be forged.

[0045] Understandably, this module, by applying the national cryptographic algorithm (SM2) and the Transport Layer Security (TLS) protocol in tandem to the building AIoT control link, achieves authentication, encryption, and integrity verification of commands throughout the entire link from generation to transmission to execution, effectively preventing commands from being illegally tampered with or forged during transmission. Simultaneously, leveraging hash signature mechanisms and blockchain ledger storage, the system constructs a decentralized, trusted control trajectory recording mechanism, enabling real-time monitoring and compliance auditing of critical equipment operations, enhancing the system's anti-attack capabilities and security controllability.

[0046] Example 2: Furthermore, the present invention provides a building energy efficiency management method based on AIoT digital twins, employing a building energy efficiency management platform based on AIoT digital twins as described in the above embodiments, which can solve a technical problem related to building energy efficiency management based on AIoT digital twins. Compared with the prior art, the beneficial effects of the building energy efficiency management method based on AIoT digital twins provided by the present invention are the same as those of the building energy efficiency management platform based on AIoT digital twins provided in the above embodiments, and other technical features in the building energy efficiency management method based on AIoT digital twins are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0047] Example 3: This invention provides a building energy efficiency management device based on AIoT digital twins. Please refer to... Figure 2A building energy efficiency management device based on AIoT digital twins includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to perform a building energy efficiency management method based on AIoT digital twins as described in Embodiment 1 above. The building energy efficiency management device based on AIoT digital twins in this embodiment of the invention may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. This building energy efficiency management device based on AIoT digital twins is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the invention. A building energy efficiency management device based on AIoT digital twin may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 1002 or a program loaded from storage device 1003 into random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the AIoT digital twin-based building energy efficiency management device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following platforms can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard drives, etc.; and communication devices 1009. Communication device 1009 allows an AIoT digital twin-based building energy management device to wirelessly or wiredly communicate with other devices to exchange data. Although the figure shows an AIoT digital twin-based building energy management device with various platforms, it should be understood that it is not required to implement or have all of the platforms shown. More or fewer platforms can be implemented alternatively.

[0048] Example 4: This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described AIoT digital twin-based building energy efficiency management method. The computer program product provided by this invention can solve a technical problem related to AIoT digital twin-based building energy efficiency management. Compared with the prior art, the beneficial effects of the computer program product provided by this invention are the same as those of the AIoT digital twin-based building energy efficiency management method provided in the above embodiments, and will not be repeated here.

[0049] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this invention.

[0050] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0051] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. An AIoT digital twin-based building energy efficiency management system, characterized in that, The system comprises: a heterogeneous data acquisition module configured to acquire AIoT device operation data in a building, including time series data, spatial structure data, and text description data, and to perform standardization processing and fusion processing on the device operation data based on a weighted least variance estimation method to obtain a multi-source data fusion vector; a building model construction module configured to construct a building digital twin model based on a heat conduction first-order differential modeling method, and to input the multi-source data fusion vector into the building digital twin model to output a current energy consumption state vector; The reinforcement learning optimization module is configured to input the current energy consumption state vector into a preset Q-learning optimization control model, and output an optimized action strategy , and generate a control instruction set and output the control instruction set to an AIoT device control interface; wherein the current energy consumption state vector is input into a preset Q-learning optimization control model, and an optimized action strategy is output , and the step of constructing a state space input vector of the Q-learning optimization control model based on the current energy consumption state vector output by the building digital twin model ; acquiring controllable parameters of the AIoT device, and constructing an action space set A, wherein each action in the action space set A contains a set of control instructions, including partition temperature control instructions, fan speed gear control instructions, and lighting brightness level control instructions; setting an immediate reward function , the immediate reward function is used to measure the energy saving performance of each action in the action space set A; in the Q-learning optimization control model, the Bellman optimality principle is used to perform iterative updates according to the state space input vector , the action space set A, and the immediate reward function , and output an optimized action strategy ; The fault prediction management module is configured to model the multi-source data fusion vector by using a long short-term memory network (LSTM), predict the multi-source data fusion vector of the next day time step, and calculate a residual error according to the multi-source data fusion vectors of the previous day and the next day When the residual error of the continuous k time points satisfies the residual error threshold , the system is determined to be in a fault warning state, a fault adjustment factor is generated, and the fault adjustment factor is fed back to the action space of the Q-learning optimization control model in the reinforcement learning optimization module to shield the fault channel. ​​ A security control execution module is configured to execute the optimized action strategy and a fault adjustment factor A closed-loop control instruction is constructed by adopting the national secret SM2 and the TLS protocol, and the closed-loop control instruction is issued to each AloT device node.

2. The building energy efficiency management system based on AIoT digital twinning of claim 1, wherein, In the heterogeneous data acquisition module, the time series data includes energy consumption time series and device operation parameter time series; the spatial structure data is derived from a preset BIM model; and the text description data includes device manuals and maintenance records. 3.The building energy efficiency management system based on AIoT digital twinning of claim 1, wherein, In the safety control execution module, the optimized action strategy is executed according to the fault adjustment factor And fault adjustment factor In the step of constructing closed-loop control instructions by adopting the national secret SM2 and the TLS protocol, the device authentication of the encryption process of the national secret SM2 and the TLS protocol uses a hash signature Tamper-proofing is performed, and all data is written into a blockchain distributed ledger. 4.The building energy efficiency management system based on AIoT digital twinning of claim 1, wherein, In the building model construction module, the formula used in the step of constructing the building digital twin model based on the heat conduction first-order differential modeling method is: wherein, is the air heat capacity of the ith partition; is the indoor temperature of the ith partition at time t; is the temperature of other partitions j adjacent to the partition i; is the outdoor temperature; is the heating / cooling input of the HVAC system of the ith partition; is the internal heat source input of the partition; is the inter-partition thermal resistance; is the thermal resistance to the outdoor. 5.The building energy efficiency management system based on AIoT digital twinning of claim 1, wherein, In the fault prediction management module, the residual is calculated according to the multi-source data fusion vectors of the previous day and the next day The method adopted is the sliding window mean square difference analysis method. 6.The building energy efficiency management system based on AIoT digital twinning of claim 1, wherein, In the heterogeneous data acquisition module, the standardization processing and fusion processing include data cleaning processing, unit normalization processing, category variable encoding and embedding processing, and weighted least variance estimation fusion processing.

7. A building energy efficiency management method based on AIoT digital twinning, applied to the building energy efficiency management system based on AIoT digital twinning in any one of claims 1 to 6, characterized in that, The method comprises: Step S10: acquiring AIoT device operation data in a building, including time series data, spatial structure data, and text description data, and performing standardization processing and fusion processing on the device operation data based on a weighted least variance estimation method to obtain a multi-source data fusion vector; Step S20: constructing a building digital twin model based on a heat conduction first-order differential modeling method, and inputting the multi-source data fusion vector into the building digital twin model to output a current energy consumption state vector; Step S30: input the current energy consumption state vector into the preset Q-learning optimization control model, and output an optimized action strategy , and generate a control instruction set and output to an AIoT device control interface; wherein the current energy consumption state vector is input into the preset Q-learning optimization control model, and an optimized action strategy is output , specifically comprising: constructing a state space input vector of the Q-learning optimization control model based on the current energy consumption state vector output by the building digital twin model ; obtaining controllable parameters of the AIoT device to construct an action space set A, wherein each action in the action space set A contains a set of control instructions, including partition temperature control instructions, fan speed gear control instructions, and lighting brightness level control instructions; setting an immediate reward function , the immediate reward function is used to measure the energy saving performance of each action in the action space set A; in the Q-learning optimization control model, the Bellman optimality principle is used to perform iterative updates according to the state space input vector , the action space set A, and the immediate reward function , and output an optimized action strategy ; Step S40: modeling the multi-source data fusion vector by using a long short-term memory network LSTM, predicting the next day time step multi-source data fusion vector, and calculating the residual according to the multi-source data fusion vectors of the previous and next days When the residual is greater than a preset residual threshold , it is determined that a fault early warning state is present, a fault adjustment factor is generated, and is fed back to the action space of the Q-learning optimization control model in the reinforcement learning optimization module to shield the fault channel. Step S50: According to the optimization action policy And fault adjustment factor Adopt the national secret SM2 and the TLS protocol to jointly construct the closed loop control instruction, and the closed loop control instruction is issued to each AloT device node.

8. An AIoT digital twin-based building energy efficiency management device, characterized in that, The AIoT digital twin-based building energy efficiency management device comprises a memory, a processor, and an AIoT digital twin-based building energy efficiency management program stored in the memory and executable on the processor, and the AIoT digital twin-based building energy efficiency management program, when executed by the processor, implements the AIoT digital twin-based building energy efficiency management system of any one of claims 1 to 6.

9. A computer program product, characterised in that, The computer program product comprises an AIoT digital twin-based building energy efficiency management program, and the AIoT digital twin-based building energy efficiency management program, when executed by a processor, implements the AIoT digital twin-based building energy efficiency management system of any one of claims 1 to 6.

Images