An AI-based green building operation diagnosis and energy-saving regulation method and system

Abstract

The application provides an AI-based green building operation diagnosis and energy-saving regulation method and system, relates to the technical field of green building operation diagnosis and optimization, and comprises the following steps: collecting green building multi-source data, and preprocessing the green building multi-source data to obtain a multi-source fusion data set; constructing a green building operation diagnosis model, taking the multi-source fusion data set as input, detecting the green building operation state, and obtaining an operation state detection result; constructing a green building energy-saving regulation model, inputting the multi-source fusion data set and the operation state detection result, and generating a device energy-saving regulation strategy; and executing the device energy-saving regulation strategy to obtain a strategy execution result. The application realizes the collaborative closed loop of accurate diagnosis and energy-saving regulation of the green building operation state, significantly improves the operation stability, energy-saving effect and comfort, and has strong project landing performance.

Description

Technical Field

[0001] This invention relates to the field of green building operation diagnosis and optimization technology, and in particular to an AI-based method and system for green building operation diagnosis and energy-saving control. Background Technology

[0002] With the development of green building and intelligent operation and maintenance technologies, AI-based building operation diagnosis and energy-saving control have become a research hotspot in the industry. Current technologies mostly collect multi-source data such as equipment operation, environment, and user behavior, and combine them with machine learning models to achieve status monitoring and strategy optimization, gradually replacing traditional manual operation and maintenance and fixed control modes. The aim is to balance building energy consumption, operational stability, and indoor comfort, meeting the needs of green building low-carbon and intelligent development.

[0003] However, existing technologies still face a series of interconnected bottlenecks in achieving deep energy conservation and intelligent operation and maintenance in green buildings: First, insufficient feature extraction capabilities make it difficult to capture complex spatiotemporal coupled anomalies. Existing technologies mostly employ single CNN or LSTM models, which cannot simultaneously integrate spatiotemporal information, making it difficult to identify complex anomaly patterns with both spatial correlation and temporal periodicity, such as "periodic overcooling caused by uneven regional air supply," resulting in low anomaly identification accuracy. Second, anomaly detection often relies on fixed thresholds or single algorithms, making it difficult to adapt to dynamic changes in building operating conditions, easily leading to false alarms or missed alarms, and failing to provide a reliable basis for subsequent control. Third, the root cause reasoning for anomalies lacks quantitative support, resulting in insufficient credibility of decision-making basis. Existing fault diagnosis systems can only provide a list of anomaly types, lacking probabilistic inferences about root causes, and cannot distinguish the probability of faults, making it difficult for operation and maintenance personnel to determine the priority of handling. Finally, in the existing technology, the diagnostic results and energy-saving control models are independent of each other. The optimization algorithm usually operates under ideal conditions assuming that the equipment is completely healthy. When the equipment performance degrades or there are slight abnormalities, the generated "optimal" strategy may exceed the actual capabilities of the equipment, causing the instructions to fail to be executed or even aggravating the equipment wear and tear.

[0004] How to solve the above-mentioned technical problems is the challenge facing this invention. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an AI-based method and system for green building operation diagnosis and energy-saving control that achieves a collaborative closed loop of accurate diagnosis and energy-saving regulation of green building operation status, significantly improves operational stability, energy-saving effect and comfort, and is highly feasible for engineering implementation.

[0006] The technical solution adopted by this invention to solve its technical problem is as follows: This invention provides an AI-based method for green building operation diagnosis and energy-saving control, comprising the following steps: Collect multi-source data on green buildings and preprocess the multi-source data to obtain a multi-source fusion dataset; A green building operation diagnostic model is constructed, which uses a multi-source fusion dataset as input to detect the operation status of green buildings and obtain the operation status detection results. Construct a green building energy-saving control model, input multi-source fusion dataset and operation status detection results, and generate equipment energy-saving control strategies; Implement energy-saving control strategies for equipment, obtain the results of strategy implementation, and update the green building operation diagnosis model and green building energy-saving control model based on the results of strategy implementation.

[0007] Preferably, the multi-source data for green buildings includes equipment operating parameters, environmental parameters, user behavior data, and third-party associated data; the equipment operating parameters include operating status data of air conditioning equipment, lighting equipment, water pumps, and fans; the environmental parameters include indoor and outdoor temperature and humidity, carbon dioxide concentration, and PM2.5 concentration; the user behavior data includes personnel flow data and terminal device usage frequency; the third-party associated data includes weather forecast data and time-of-use electricity pricing; the collection of multi-source data for green buildings also includes recording the time information of data collection and obtaining the unique equipment identifiers and spatial location information of air conditioning equipment, lighting equipment, water pumps, fans, and terminal devices.

[0008] Preferably, the preprocessing includes noise reduction, missing value imputation, data standardization, and establishing a spatiotemporal information mapping relationship for the device.

[0009] Preferably, the detection of the green building's operational status includes A CNN-LSTM hybrid model is used to extract features from a multi-source fusion dataset to obtain operational status features. These operational status features include equipment load features, environmental parameter fluctuation features, equipment runtime sequence change features, and spatial correlation features. An isolated forest method is used to detect outliers, combined with sliding window time series analysis to detect anomalies in the operational status characteristics, resulting in outliers and the current window deviation; the current window deviation includes the deviation of the rate of change and the deviation of the fluctuation amplitude of the current window. The isolated path length is calculated based on the outlier, the time anomaly score is calculated based on the current window deviation, and the isolated path length and the time anomaly score are normalized to obtain the normalized isolated path length and time anomaly score. The runtime anomaly score is obtained by weighting the normalized isolated path length with the time-series anomaly score. A threshold is set for the operation anomaly score. If the operation anomaly score is less than or equal to the threshold, the operation is considered normal; if the operation anomaly score is greater than the threshold, the operation is considered abnormal and an operation anomaly event is obtained. Based on abnormal events, fault tree combined with Bayesian network is used to perform root cause reasoning to obtain the cause of operational abnormality; the cause of operational abnormality includes controllable causes and hardware failure causes. By integrating abnormal operation events and their causes, the operation status detection results are obtained.

[0010] Preferably, the energy-saving control strategy for the generating equipment includes: Based on the operational status monitoring results, when the operation is normal or abnormal and the cause of the abnormality is controllable, the equipment operating parameters are used as decision variables. The NSGA-III multi-objective optimization algorithm is used to define the energy-saving control objective function and constraints. Iterative optimization is performed according to the energy-saving control objective function and constraints to obtain the strategy when the equipment is controllable. The strategy when the equipment is controllable includes the optimized equipment operating parameters and the optimized operating cycle. When the device malfunctions and the cause of the malfunction is hardware damage, a strategy is generated for when the device is uncontrollable; the strategy for when the device is uncontrollable includes stopping the device and generating a device maintenance command. By integrating strategies for when the equipment is controllable and strategies for when the equipment is not controllable, an energy-saving control strategy for the equipment is obtained.

[0011] Preferably, the energy-saving control objective function includes minimizing building energy consumption and minimizing indoor comfort deviation, wherein minimizing building energy consumption is measured by the total building electricity consumption, and minimizing indoor comfort is measured by the PMV index; the constraints include equipment physical constraints, environmental comfort constraints, and energy efficiency constraints, wherein the equipment physical constraints include physical constraints for normal operation and physical constraints for abnormal operation.

[0012] Preferably, the execution of the energy-saving control strategy for the equipment yields the following results: The energy-saving control strategy of the equipment is converted into equipment control commands through the Modbus communication protocol and transmitted to the controller for execution. After the equipment control commands have completed the optimized operating cycle, collect multi-source data on green buildings after the strategy has been implemented; When collecting multiple batches of green building multi-source data after the strategy execution, outlier filtering is performed on the green building multi-source data after the strategy execution to obtain processed multi-source data. Based on the processed multi-source data, the strategy execution result is calculated; the strategy execution result includes equipment response data, energy consumption change data, comfort index data, and abnormal improvement degree data.

[0013] This invention also provides an AI-based green building operation diagnosis and energy-saving control system, comprising: The data acquisition and preprocessing module is used to collect multi-source data on green buildings and preprocess it to output a multi-source fused dataset. The operational status diagnostic model construction and status detection module is used to build a green building operational diagnostic model. It takes a multi-source fusion dataset as input, performs operational status detection on the green building through the green building operational diagnostic model, and outputs the operational status detection results. The energy-saving control model construction and strategy generation module is used to construct a green building energy-saving control model. It takes a multi-source fusion dataset and the operation status detection results as input, and generates equipment energy-saving control strategies through the green building energy-saving control model. The strategy execution and result acquisition module is used to execute equipment energy-saving control strategies, collect multi-source data of green buildings after strategy execution and perform preprocessing, and calculate the strategy execution result based on the processed multi-source data.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described AI-based green building operation diagnosis and energy-saving control method.

[0015] The present invention also provides a computer storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described AI-based green building operation diagnosis and energy-saving control method.

[0016] The beneficial effects of this invention are as follows: it achieves a synergistic closed loop of precise diagnosis and energy-saving control of green building operation status, significantly improving operational stability, energy-saving effect, and comfort, and has strong engineering feasibility. A CNN-LSTM hybrid model is used for feature extraction. The CNN part can automatically learn the correlation patterns between data from different spatial points at the same time (such as the collaborative state of air conditioning units and terminal fans), while the LSTM part can capture the long-term patterns of single or multiple parameters evolving over time (such as the correlation between the daily cycle of energy consumption and human activity), thus enabling more accurate identification of complex anomaly patterns such as "periodic overcooling in a certain area due to uneven air supply." The isolated path length, reflecting the outlier degree of the global distribution, is combined with the temporal anomaly score, reflecting the local temporal deviation. A dynamic weight based on historical variance is designed, automatically assigning higher decision weights to indicators with more stable historical performance (smaller variance). This allows the system to automatically adjust the detection sensitivity according to the characteristics of different subsystems, significantly improving the robustness and accuracy of anomaly identification under complex and variable operating conditions compared to fixed thresholds or rules. By combining fault trees with Bayesian networks, a conditional probability table is learned from historical data, and the posterior probability is calculated using the Markov chain Monte Carlo method during inference. This allows the system to not only list possible causes of failure but also output quantified probability distributions, providing a direct and reliable basis for subsequent decision-making. The anomaly types and degrees output by the diagnostic model are dynamically converted into the narrowing of physical constraints on the equipment in the optimization model (e.g., lowering the maximum set temperature by 1.5℃). The NSGA-III multi-objective optimization algorithm is then used to resolve the Pareto optimal solution for energy consumption and comfort within this corrected constraint space. This ensures that the generated control strategy strictly conforms to the actual operating capacity boundary of the equipment, avoiding the "theoretically optimal but practically infeasible" settings that traditional methods provide after equipment performance degradation. Thus, robust energy-saving optimization under real-world operating conditions is achieved while ensuring system safety and equipment lifespan. Attached Figure Description

[0017] Figure 1 This is a diagram illustrating the method steps of the present invention.

[0018] Figure 2 This is a system module diagram of the present invention.

[0019] Figure 3 This is an internal structural diagram of Embodiment 3 of the present invention. Detailed Implementation

[0020] To clearly illustrate the technical features of this solution, the following detailed implementation method will be used to explain the solution.

[0021] Example 1: See Figure 1As shown, this embodiment is an AI-based method for green building operation diagnosis and energy-saving control, including the following steps: S1. Collect multi-source data on green buildings and preprocess the multi-source data to obtain a multi-source fusion dataset; Multi-source data for green buildings includes equipment operating parameters, environmental parameters, user behavior data, and third-party related data. Equipment operating parameters include the operating status data of air conditioning equipment, lighting equipment, water pumps, and fans. Environmental parameters include indoor and outdoor temperature and humidity, carbon dioxide concentration, and PM2.5 concentration. User behavior data includes personnel flow data and the frequency of use of terminal equipment. Third-party related data includes weather forecast data and time-of-use electricity pricing. Collecting multi-source data for green buildings also includes recording the time information of data collection and obtaining the unique identifiers and spatial location information of air conditioning equipment, lighting equipment, water pumps, fans, and terminal equipment.

[0022] Preprocessing includes noise reduction, missing value imputation, data standardization, and establishing a spatiotemporal information mapping relationship for the equipment.

[0023] It should be noted that the equipment operating parameters include the operating status data of air conditioning equipment, lighting equipment, water pumps, and fans, specifically air conditioning power, compressor frequency, water flow, fan speed, and lighting power; environmental parameters include indoor and outdoor temperature and humidity, carbon dioxide concentration, and PM2.5 concentration; user behavior data includes personnel flow data and terminal device usage frequency, with personnel flow data collected through infrared sensors or cameras and terminal device usage frequency recorded through smart sockets; third-party related data includes weather forecast data and time-of-use electricity pricing, with weather forecast data obtained from the meteorological bureau's API and time-of-use electricity pricing obtained from the power company's API; the collection of multi-source data for green buildings also includes recording the time information of data collection and obtaining the unique equipment identifiers and spatial location information of air conditioning equipment, lighting equipment, water pumps, fans, and terminal devices, including building, floor, and room number.

[0024] Noise reduction processing: The vibration signal and current signal are decomposed into three layers using db4 wavelet for noise reduction; Missing value imputation: If the number of consecutive missing records is ≤5, linear interpolation is used for imputation; if the number of consecutive missing records is >5, the mean of the corresponding historical data of the same period of the same device is used for imputation. Data standardization: Physical quantity parameters are mapped to the [0,1] interval using the min-max standardization method; energy consumption parameters are processed using the Z-score standardization method to make the mean 0 and the standard deviation 1 after processing. Establish a spatiotemporal information mapping relationship for devices: associate the unique device identifier of each device with the data it collects, its spatial location, and the timestamp of the data collection to form a structured table.

[0025] S2. Construct a green building operation diagnosis model, using a multi-source fusion dataset as input, to detect the operation status of green buildings and obtain the operation status detection results; Monitoring the operational status of green buildings includes A CNN-LSTM hybrid model is used to extract features from a multi-source fusion dataset to obtain operational status features. These operational status features include equipment load features, environmental parameter fluctuation features, equipment runtime sequence change features, and spatial correlation features. An isolated forest method is used to detect outliers, combined with sliding window time series analysis to detect anomalies in the operational status characteristics, resulting in outliers and the current window deviation; the current window deviation includes the deviation of the rate of change and the deviation of the fluctuation amplitude of the current window. The isolated path length is calculated based on outliers, and the temporal anomaly score is calculated based on the current window bias. The calculation formula is as follows:

[0026] in, and These represent the change deviation and fluctuation amplitude deviation of the current window, respectively. and These are the standard deviations of the change deviation and the standard deviations of the fluctuation amplitude, respectively. The isolated path length and the time anomaly score are normalized to obtain the normalized isolated path length and time anomaly score. The runtime anomaly score is obtained by weighting the normalized isolated path length with the time-series anomaly score; the calculation formula is as follows:

[0027] in, This represents the normalized isolated path length. As weight.

[0028]

[0029] in, and These represent the variance of the normalized isolated path length and the variance of the temporal anomaly score, respectively.

[0030] A threshold is set for the operation anomaly score. If the operation anomaly score is less than or equal to the threshold, the operation is considered normal; if the operation anomaly score is greater than the threshold, the operation is considered abnormal and an operation anomaly event is obtained. Based on abnormal events, fault tree combined with Bayesian network is used to perform root cause reasoning to obtain the cause of operational abnormality; the cause of operational abnormality includes controllable causes and hardware failure causes. By integrating abnormal operation events and their causes, the operation status detection results are obtained.

[0031] It should be noted that a CNN-LSTM hybrid model was used to extract features from the multi-source fusion dataset to obtain runtime features. In the CNN-LSTM hybrid model, the CNN part contains two convolutional layers with kernel sizes of 3×3 and 5×5, a stride of 1, and the activation function is ReLU. The LSTM part contains one LSTM layer with 128 hidden neurons. The runtime features include the following four categories, which were extracted from the multi-source fusion dataset using the CNN-LSTM hybrid model: (1) Equipment load characteristics: including the ratio of actual operating parameters of the equipment to its rated parameters, such as the current load rate of the air conditioning compressor and the power density of the lighting circuit; (2) Environmental parameter fluctuation characteristics: including the statistical variance or differential energy of environmental parameters within a sliding time window, such as the standard deviation of indoor temperature within a 15-minute window; (3) Characteristics of equipment operation sequence changes: including the periodicity of equipment status signals, duty cycle and start-stop time deviation, such as the deviation between the daily start time of the water pump and the standard time; (4) Spatial correlation characteristics: including correlation coefficients, gradients or balance of monitoring point data at different spatial locations, such as the correlation coefficients of temperature data from different rooms on the same floor.

[0032] Outlier detection using isolated forest: The number of decision trees in the isolated forest is set to 100, and the running state features are used for training to obtain the isolated path length of each feature point; Combining sliding window time series analysis: with a window length of 1 hour, calculate the rate of change deviation and fluctuation amplitude deviation for each feature within the current window, where: (1) Change rate deviation = mean of first-order differences of features within the current window - mean of first-order differences of windows in the same historical period; (2) Fluctuation amplitude deviation = standard deviation of features within the current window - standard deviation of historical windows in the same period; historical windows in the same period are windows at the same time in the past 7 days; Based on abnormal events, a fault tree combined with a Bayesian network is used for root cause inference. The fault tree is constructed based on domain knowledge and historical failure cases, including abnormal phenomena such as "insufficient air conditioning cooling" and "poor contact in the lighting system" and their corresponding potential root causes. For example, the potential root causes of "insufficient air conditioning cooling" include "filter blockage" and "refrigerant leakage". The conditional probability table of the Bayesian network is learned from historical failure data and updated quarterly. During inference, the current abnormal features are input into the Bayesian network, and the posterior probability of each potential root cause is calculated using the Markov chain Monte Carlo method. The one with the highest probability is taken as the cause of the operational anomaly. The causes of operational anomalies include controllable causes and hardware failure causes. Controllable causes refer to causes that can be improved by adjusting the equipment operating parameters, while hardware failure causes refer to causes that require repair or replacement of hardware.

[0033] S3. Construct a green building energy-saving control model, input multi-source fusion dataset and operation status detection results, and generate equipment energy-saving control strategies; Energy-saving control strategies for generating equipment include Based on the operational status monitoring results, when the operation is normal or abnormal and the cause of the abnormality is controllable, the equipment operating parameters are used as decision variables. The NSGA-III multi-objective optimization algorithm (with parameters set as follows: population size 100, number of iterations 200, crossover probability 0.9, mutation probability 0.1) is used to define the energy-saving control objective function and constraints. Iterative optimization is performed based on the energy-saving control objective function and constraints to obtain the strategy when the equipment is controllable. The strategy when the equipment is controllable includes the optimized equipment operating parameters and the optimized operating cycle. When an abnormal operation occurs and the cause of the abnormal operation is hardware failure, a strategy is generated for when the device is uncontrollable; the strategy for when the device is uncontrollable includes stopping the device and generating a device maintenance instruction. By integrating strategies for when the equipment is controllable and strategies for when the equipment is not controllable, an energy-saving control strategy for the equipment is obtained.

[0034] The energy-saving control objective function includes minimizing building energy consumption and minimizing indoor comfort deviation. Minimizing building energy consumption is measured by the total building electricity consumption, and minimizing indoor comfort is measured by the PMV index. The constraints include equipment physical constraints, environmental comfort constraints, and energy efficiency constraints. Equipment physical constraints include physical constraints for normal operation and physical constraints for abnormal operation, as expressed by the following formula:

[0035] in, Let the building energy consumption sub-objective function be... The objective function is the sub-objective function for indoor comfort deviation.

[0036]

[0037] in, As decision variables, For time parameters, A collection of electrical equipment. This refers to the instantaneous power of the equipment.

[0038]

[0039] in, This represents the total number of spatial regions. For regional weights, The predicted PMV values ​​for the region are calculated using a thermal comfort model, with parameters derived from multi-source data. For the target PMV value, For a moment.

[0040] Equipment physical constraints:

[0041] in, The rated operating range of the equipment is provided by the equipment manufacturer and is a known static parameter. The calculation is dynamic based on the type and severity of the anomaly in the output. For example, if the diagnosis is "air conditioning efficiency decreased by x%", the temperature adjustment range will be narrowed proportionally.

[0042] Environmental comfort constraints:

[0043] in, These are the preset lower and upper limits of the indoor comfort PMV index, respectively. This range is usually taken as [ 0.5,0.5][ [0.5, 0.5], but can be adjusted according to building type, user needs, or season (e.g., a senior activity center can be set to [ ]). 0.3,0.3][ 0.3,0.3]).

[0044] Energy efficiency constraints:

[0045] in, For equipment At any moment Energy efficiency ratio, This is the minimum energy efficiency ratio, which can be preset to a percentage of the device's rated COP (e.g., 80%), or set according to local energy efficiency standards. This is a preset or configured parameter.

[0046] S4. Execute the equipment energy-saving control strategy, obtain the strategy execution results, and update the green building operation diagnosis model and green building energy-saving control model based on the strategy execution results.

[0047] The execution of the equipment energy-saving control strategy yields the following results: The energy-saving control strategy of the equipment is converted into equipment control commands through the Modbus communication protocol and transmitted to the controller for execution. After the equipment control commands have completed the optimized operating cycle, collect multi-source data on green buildings after the strategy has been implemented; When collecting multiple batches of green building multi-source data after the strategy execution, outlier filtering is performed on the green building multi-source data after the strategy execution to obtain processed multi-source data. Based on the processed multi-source data, the strategy execution results are calculated; the strategy execution results include equipment response data, energy consumption change data, comfort index data, and abnormal improvement degree data.

[0048] It should be noted that the equipment response data is obtained by comparing the deviation between the set values ​​of the equipment control commands and the actual operating parameters, the response time, and the command execution success rate; the following accuracy is expressed as the percentage error between the command value and the actual value: Error percentage = |Actual value - Command value| / Command value × 100%; the response time is the time difference from the issuance of the command to the actual value first entering the tolerance zone of the command value; energy consumption change data is obtained by calculating the actual total energy consumption within the strategy execution cycle and comparing it with the baseline energy consumption; the baseline energy consumption is the average energy consumption of the same period in the 7 days before the strategy execution; energy consumption change rate = (actual total energy consumption - baseline energy consumption) / baseline energy consumption × 100%; comfort index data is obtained by statistically analyzing the compliance rate of PMV values ​​within the preset comfort range and calculating the spatial temperature uniformity index; PMV compliance rate = (number of sampling points with PMV values ​​within the comfort range / total number of sampling points) ×100%; Abnormal improvement rate data: obtained by comparing the change rate of the abnormal operation score before and after the strategy is implemented; the operation status characteristics after the strategy is implemented are re-extracted using the green building operation diagnosis model before the update, the abnormal operation score is calculated, and compared with the abnormal score before the strategy is implemented to obtain the abnormal improvement rate.

[0049] Example 2: See Figure 2 As shown, this embodiment is an AI-based green building operation diagnosis and energy-saving control system, including: The data acquisition and preprocessing module is used to collect multi-source data on green buildings and preprocess it to output a multi-source fused dataset. The operational status diagnostic model construction and status detection module is used to build a green building operational diagnostic model. It takes a multi-source fusion dataset as input, performs operational status detection on the green building through the green building operational diagnostic model, and outputs the operational status detection results. The energy-saving control model construction and strategy generation module is used to construct a green building energy-saving control model. It takes a multi-source fusion dataset and the operation status detection results as input, and generates equipment energy-saving control strategies through the green building energy-saving control model. The strategy execution and result acquisition module is used to execute equipment energy-saving control strategies, collect multi-source data of green buildings after strategy execution and perform preprocessing, and calculate the strategy execution result based on the processed multi-source data.

[0050] Example 3: This embodiment provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0051] This computer device can be a server, and its internal structure diagram can be as follows: Figure 3 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores server data. The network interface communicates with external terminals via a network connection. When the computer program is executed by the processor, it implements an AI-based method for green building operation diagnosis and energy-saving control.

[0052] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0053] Example 4: This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0054] If the functions implemented by the method are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art or the current technical solution, can be embodied in the form of a software product. This current computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0055] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0056] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0057] The technical features of this invention not described can be implemented by or using existing technology, and will not be repeated here. Of course, the above description is not a limitation of this invention, and this invention is not limited to the examples above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of this invention should also be within the protection scope of this invention.

Claims

1. A method for green building operation diagnosis and energy-saving control based on AI, characterized in that, Includes the following steps: Collect multi-source data on green buildings and preprocess the multi-source data to obtain a multi-source fusion dataset; A green building operation diagnostic model is constructed, which uses a multi-source fusion dataset as input to detect the operation status of green buildings and obtain the operation status detection results. Construct a green building energy-saving control model, input multi-source fusion dataset and operation status detection results, and generate equipment energy-saving control strategies; Implement the equipment energy-saving control strategy and obtain the strategy execution results.

2. The AI-based green building operation diagnosis and energy-saving control method according to claim 1, characterized in that, The multi-source data for green buildings includes equipment operating parameters, environmental parameters, user behavior data, and third-party associated data. The equipment operating parameters include operating status data of air conditioning equipment, lighting equipment, water pumps, and fans. The environmental parameters include indoor and outdoor temperature and humidity, carbon dioxide concentration, and PM2.5 concentration. The user behavior data includes personnel flow data and terminal device usage frequency. The third-party associated data includes weather forecast data and time-of-use electricity pricing. The collection of multi-source data for green buildings also includes recording the time information of data collection and obtaining the unique equipment identifiers and spatial location information of air conditioning equipment, lighting equipment, water pumps, fans, and terminal devices.

3. The AI-based green building operation diagnosis and energy-saving control method according to claim 2, characterized in that, The preprocessing includes noise reduction, missing value imputation, data standardization, and establishing a spatiotemporal information mapping relationship for the device.

4. The AI-based green building operation diagnosis and energy-saving control method according to claim 3, characterized in that, The detection of the operational status of green buildings includes A CNN-LSTM hybrid model is used to extract features from a multi-source fusion dataset to obtain operational status features. These operational status features include equipment load features, environmental parameter fluctuation features, equipment runtime sequence change features, and spatial correlation features. An isolated forest method is used to detect outliers, combined with sliding window time series analysis to detect anomalies in the operational status characteristics, resulting in outliers and the current window deviation; the current window deviation includes the deviation of the rate of change and the deviation of the fluctuation amplitude of the current window. The isolated path length is calculated based on the outlier, the time anomaly score is calculated based on the current window deviation, and the isolated path length and the time anomaly score are normalized to obtain the normalized isolated path length and time anomaly score. The runtime anomaly score is obtained by weighting the normalized isolated path length with the time-series anomaly score. A threshold is set for the operation anomaly score. If the operation anomaly score is less than or equal to the threshold, the operation is considered normal. If the abnormal operation score is greater than the threshold, it is determined that there is an abnormality in the operation and an abnormal operation event is obtained; Based on abnormal events, fault tree combined with Bayesian network is used to perform root cause reasoning to obtain the cause of operational abnormality; the cause of operational abnormality includes controllable causes and hardware failure causes. By integrating abnormal operation events and their causes, the operation status detection results are obtained.

5. The AI-based green building operation diagnosis and energy-saving control method according to claim 4, characterized in that, The energy-saving control strategy for the generating equipment includes Based on the operational status monitoring results, when the operation is normal or abnormal and the cause of the abnormality is controllable, the equipment operating parameters are used as decision variables. The NSGA-III multi-objective optimization algorithm is used to define the energy-saving control objective function and constraints. Iterative optimization is performed according to the energy-saving control objective function and constraints to obtain the strategy when the equipment is controllable. The strategy when the equipment is controllable includes the optimized equipment operating parameters and the optimized operating cycle. When the device malfunctions and the cause of the malfunction is hardware damage, a strategy is generated for when the device is uncontrollable; the strategy for when the device is uncontrollable includes stopping the device and generating a device maintenance command. By integrating strategies for when the equipment is controllable and strategies for when the equipment is not controllable, an energy-saving control strategy for the equipment is obtained.

6. The AI-based green building operation diagnosis and energy-saving control method according to claim 5, characterized in that, The energy-saving control objective function includes minimizing building energy consumption and minimizing indoor comfort deviation. The minimization of building energy consumption is measured by the total building electricity consumption, and the minimization of indoor comfort is measured by the PMV index. The constraints include equipment physical constraints, environmental comfort constraints, and energy efficiency constraints. The equipment physical constraints include physical constraints for normal operation and physical constraints for abnormal operation.

7. The AI-based green building operation diagnosis and energy-saving control method according to claim 6, characterized in that, The execution of the energy-saving control strategy for the equipment yields the following results: The energy-saving control strategy of the equipment is converted into equipment control commands through the Modbus communication protocol and transmitted to the controller for execution. After the equipment control commands have completed the optimized operating cycle, collect multi-source data on green buildings after the strategy has been implemented; When collecting multiple batches of green building multi-source data after the strategy execution, outlier filtering is performed on the green building multi-source data after the strategy execution to obtain processed multi-source data. Based on the processed multi-source data, the strategy execution result is calculated; the strategy execution result includes equipment response data, energy consumption change data, comfort index data, and abnormal improvement degree data.

8. An AI-based green building operation diagnosis and energy-saving control system, characterized in that, The method for performing the AI-based green building operation diagnosis and energy-saving control method according to any one of claims 1-7 includes: The data acquisition and preprocessing module is used to collect multi-source data on green buildings and preprocess it to output a multi-source fused dataset. The operational status diagnostic model construction and status detection module is used to build a green building operational diagnostic model. It takes a multi-source fusion dataset as input, performs operational status detection on the green building through the green building operational diagnostic model, and outputs the operational status detection results. The energy-saving control model construction and strategy generation module is used to construct a green building energy-saving control model. It takes a multi-source fusion dataset and the operation status detection results as input, and generates equipment energy-saving control strategies through the green building energy-saving control model. The strategy execution and result acquisition module is used to execute equipment energy-saving control strategies, collect multi-source data of green buildings after strategy execution and perform preprocessing, and calculate the strategy execution result based on the processed multi-source data.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the AI-based green building operation diagnosis and energy-saving control method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the AI-based green building operation diagnosis and energy-saving control method as described in any one of claims 1 to 7.

Images