Building intelligent management system and method based on AI

Abstract

The invention discloses an AI-based building intelligent management system and method, belongs to the field of intelligent buildings, and aims to solve the problems of energy waste, environment regulation lag, slow security response, passive equipment maintenance and the like in traditional building management. The system adopts a layered architecture design and comprises a data acquisition layer, a data processing layer, a decision-making layer and an application layer, wherein the data acquisition layer acquires environment, personnel flow and equipment operation data through a sensor, a camera, an intelligent electric meter and the like; the data processing layer cleans and preprocesses the data, extracts features, and performs parallel processing based on cloud computing or edge computing; the decision-making layer integrates AI algorithms such as deep learning, machine learning and reinforcement learning to realize data prediction, state judgment and strategy optimization; the application layer provides a management platform and a mobile terminal interface, and realizes the functions of intelligent ventilation, energy management, security monitoring and equipment maintenance; according to the system and the method, through AI-driven full-process intelligent management and control, the security response efficiency is improved, and active maintenance of equipment is realized.

Description

Technical Field

[0001] This invention relates to the field of intelligent buildings, and in particular to an AI-based intelligent building management system and method. Background Technology

[0002] In the field of smart buildings, inefficient energy management is one of the core pain points of traditional building operations, severely restricting the economy and sustainability of building management and urgently requiring technological breakthroughs. Traditional building energy management relies on manual operation and fixed operating modes, lacking dynamic adaptability. Lighting systems mostly use timed switches or manual control, and even in areas with sufficient natural light during the day, lighting equipment often runs at full power. Air conditioning systems mostly operate continuously at preset temperatures, unable to adjust operating parameters according to changes in indoor and outdoor temperature differences or the number of people. For example, it is common for air conditioners to run idle after get off work in office areas and for equipment to continue operating at high energy consumption in meeting rooms when no one is present. Traditional systems lack accurate energy consumption monitoring and analysis methods. Data from smart meters and water meters are only used for basic metering and have not been used to deeply analyze energy consumption patterns, making it difficult to identify energy consumption anomalies and areas with energy-saving potential. This results in low energy utilization rates in most buildings, with energy waste rates reaching 15%-25%.

[0003] Traditional energy management lacks predictive and optimization capabilities, failing to forecast energy demand at different times based on historical data and real-time operating conditions. It can only passively respond to energy consumption changes and cannot proactively formulate energy-saving strategies. For example, it cannot predict the energy consumption differences between weekday morning peak hours and nighttime off-peak hours, making it difficult to achieve refined energy allocation by time period and region, further exacerbating energy waste. This management model, reliant on manual labor and lacking intelligent control, not only drives up building operating costs but also contradicts the current demands for green building and low-carbon development. Therefore, there is an urgent need for a building intelligent management system integrating AI technology to solve the problem of inefficient energy management and achieve precise, intelligent, and energy-efficient energy utilization. Summary of the Invention

[0004] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0005] The technical solution adopted by the present invention to solve its technical problem is: the building intelligent management system based on AI described in the present invention includes: a data acquisition layer, which consists of sensors, cameras, smart meters and smart water meters distributed in various areas of the building, used to collect environmental data, personnel flow data and equipment operation data in the building;

[0006] The data processing layer cleans, preprocesses, and extracts features from the data acquired by the data acquisition layer, and performs parallel processing using a cloud computing platform or edge computing device.

[0007] The AI ​​decision-making layer integrates multiple AI algorithms and models. It analyzes and predicts the processed data through deep learning algorithms, classifies and judges the operating status of equipment using machine learning algorithms, and optimizes intelligent control strategies using reinforcement learning algorithms.

[0008] The application layer provides management function interfaces for users, including the management platform used by administrators and the mobile application used by users.

[0009] As a preferred technical solution of this application, in the data acquisition layer, the temperature sensor collects indoor temperature data at intervals of 5-10 minutes. The humidity sensor collects indoor humidity data at equal intervals. Air quality sensors collect data on carbon dioxide concentration in the air. Formaldehyde concentration data Personnel flow sensors use infrared sensing and video analysis technology to count the number and direction of people entering and leaving in real time. Cameras use target detection and tracking algorithms to obtain information such as personnel flow trajectory and dwell time. Smart meters and smart water meters collect data on electricity consumption and water consumption, respectively.

[0010] As a preferred technical solution of this application, in the data processing layer, a data cleaning algorithm is used to remove noise, outliers and duplicate data from the data, Fourier transform and other methods are used to extract frequency domain features from the time series environmental data, and a convolutional neural network (CNN) is used to extract image features from the video image data.

[0011] As a preferred technical solution of this application, in the decision layer, the deep learning algorithm uses a multilayer perceptron (MLP) model to predict changes in environmental parameters, as shown in the formula: ,in The mapping function represents the multilayer perceptron model; the machine learning algorithm uses a support vector machine (SVM) to classify and judge the equipment operation data, and converts the feature vectors of the equipment operation data... Input a trained SVM model and output the device state category; take ventilation system control as an example of reinforcement learning algorithm, define the agent as ventilation system controller, the environment as air quality and personnel activity in the building, the action space as the operation of opening / closing the ventilation system, wind speed adjustment, etc., and the reward function is determined according to factors such as the degree of air quality improvement and energy consumption.

[0012] As a preferred technical solution of this application, in the intelligent ventilation control function of the application layer, when the indoor carbon dioxide concentration exceeds 1000ppm or there is an odor, the system automatically starts the ventilation equipment and adjusts the ventilation volume according to the number of people and the size of the space. In a densely populated conference room, when the carbon dioxide concentration reaches 1200ppm, the ventilation system runs at the maximum wind speed for 15-20 minutes and then adjusts to a suitable wind speed for continuous ventilation.

[0013] As a preferred technical solution of this application, in the energy management function of the application layer, by analyzing the power consumption data collected by the smart meter, the energy demand at different times is predicted by using AI algorithms, the lighting system automatically adjusts the brightness or switches on and off according to the ambient light intensity and the activity of people, and the air conditioning system dynamically adjusts the temperature setpoint and operating mode according to the indoor and outdoor temperature, the number of people and their activities, which is expected to reduce energy consumption by 15%-25%.

[0014] As a preferred technical solution of this application, the security monitoring function of the application layer adopts target detection and behavior analysis algorithms to monitor the behavior of people in the video in real time. When abnormal behavior is detected, an alarm is issued. Intrusion detection is achieved by setting virtual warning lines. Fire warning is achieved by using image recognition technology to detect smoke and flames in the video and linking relevant fire-fighting equipment.

[0015] As a preferred technical solution of this application, the equipment maintenance management function of the application layer monitors parameters such as equipment operating speed, vibration, and current, and uses machine learning algorithms to establish a fault prediction model. When the model predicts that the equipment may fail in the future, it notifies maintenance personnel in advance to carry out inspection and maintenance, and optimizes the maintenance plan based on the equipment maintenance history and operating conditions.

[0016] An AI-based intelligent building management method includes the following steps:

[0017] Environmental data, personnel flow data, and equipment operation data are collected within the building through sensors, cameras, smart meters, and smart water meters distributed throughout the building.

[0018] The collected data is cleaned, preprocessed, and feature extracted, and then processed in parallel using a cloud computing platform or edge computing device.

[0019] The processed data is analyzed and predicted using deep learning algorithms, the equipment operating status is classified and judged using machine learning algorithms, and the intelligent control strategy is optimized using reinforcement learning algorithms. Based on the optimized control strategy, the application layer provides users with functions such as intelligent ventilation control, energy management, security monitoring, and equipment maintenance management.

[0020] As a preferred technical solution of this application, in the data acquisition step, the specific data and method of acquisition are consistent with the data acquisition layer content described in claim 2; in the data processing step, the specific processing method is consistent with the data processing layer content described in claim 3; in the AI ​​algorithm processing step, the specific algorithm application is consistent with the decision layer content described in claim 4; and in the management function implementation step, the execution method of each management function is consistent with the application layer content described in claims 5-8.

[0021] The beneficial effects of this invention are as follows:

[0022] 1. This invention utilizes smart meters, temperature and humidity sensors, and other devices in the data acquisition layer to acquire real-time energy consumption data and environmental parameters. After cleaning and analysis by the data processing layer, the decision-making layer uses AI algorithms to predict energy demand at different times. Simultaneously, it dynamically adjusts the operating status of the lighting and air conditioning systems in the application layer. The lighting automatically switches on or dims based on light intensity and human activity, while the air conditioning optimizes temperature settings based on indoor and outdoor temperatures and the number of people. Through this intelligent control logic, energy waste under the traditional fixed mode is effectively avoided, and it is expected to reduce the overall energy consumption of the building, significantly improve energy utilization efficiency, and reduce operating costs.

[0023] 2. The data acquisition layer of this invention uses temperature and humidity sensors, air quality sensors, and personnel flow sensors distributed in various areas to capture environmental and personnel data in real time. After feature extraction by the data processing layer, the decision-making layer analyzes the changing trends of environmental parameters through AI algorithms, and the application layer's intelligent ventilation control function responds dynamically accordingly. When the carbon dioxide concentration exceeds the threshold or there is an odor, the ventilation equipment is automatically activated. In densely populated meeting rooms, when the concentration reaches the threshold, the ventilation equipment operates at maximum speed and then adjusts to a suitable speed. Through this closed-loop management, indoor environmental parameters are precisely adjusted, keeping carbon dioxide concentration, formaldehyde concentration, etc., within a reasonable range, providing a comfortable and healthy environment for personnel.

[0024] 3. This invention's system relies on high-definition cameras in the data acquisition layer to collect video image data from various areas. After the image features are extracted by the convolutional neural network (CNN) in the data processing layer, the decision layer uses AI algorithms such as target detection and behavior analysis to monitor abnormal human behavior, identify smoke, flames, and intrusions in real time. Once an anomaly is detected, the application layer's security monitoring function immediately issues an alarm and triggers an emergency response by linking fire-fighting equipment. Through this AI-driven proactive security mechanism, the system solves the problems of delayed response and independent operation of subsystems in traditional manual monitoring, improves the accuracy of security incident identification and handling efficiency, and ensures the safety of people and property in buildings.

[0025] 4. The data acquisition layer of this invention collects the operating parameters of equipment such as elevators and air conditioners in real time. After preprocessing by the data processing layer, the decision layer uses machine learning algorithms such as support vector machine (SVM) to establish a fault prediction model and identify potential equipment failure risks in advance. The application layer's equipment maintenance management function pushes early warning information to maintenance personnel based on this and optimizes maintenance plans by combining equipment maintenance history. Through this predictive maintenance mode, sudden equipment failures caused by the lag of traditional manual troubleshooting are avoided, operational interruptions caused by equipment downtime are reduced, unnecessary maintenance costs are reduced, and the service life of equipment is extended. Attached Figure Description

[0026] The invention will now be further described with reference to the accompanying drawings.

[0027] Figure 1 This is a diagram of the overall system architecture of the present invention;

[0028] Figure 2 This is a flowchart of the data acquisition and processing of the present invention;

[0029] Figure 3 This is the AI ​​decision-making and execution control interaction diagram of the present invention. Detailed Implementation

[0030] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0031] like Figure 1-3 As shown, the AI-based intelligent building management system of the present invention adopts a layered architecture design, mainly including a data acquisition layer, a data processing layer, a decision-making layer and an application layer;

[0032] Data Acquisition Layer: Composed of sensors, cameras, smart meters, smart water meters, and other devices distributed throughout the building. These devices are responsible for collecting various data within the building, such as temperature sensors collecting indoor temperature data, humidity sensors collecting indoor humidity data, air quality sensors collecting data on the concentration of pollutants such as carbon dioxide, formaldehyde, and PM2.5 in the air, personnel flow sensors collecting the number and movement information of people in various areas, cameras collecting video image data, and smart meters and smart water meters collecting data on electricity consumption and water consumption, respectively.

[0033] Data Processing Layer: Primarily responsible for cleaning, preprocessing, and feature extraction of the data acquired by the data acquisition layer. Data cleaning algorithms remove noise, outliers, and duplicate data to improve data quality. Feature extraction algorithms transform the raw data into feature vectors suitable for AI algorithm processing. For example, for video image data, Convolutional Neural Networks (CNNs) are used for image feature extraction; for time-series environmental data, methods such as Fourier Transform are used to extract its frequency domain features. Parallel data processing is performed using cloud computing platforms or edge computing devices to improve processing efficiency and reduce data transmission latency.

[0034] AI Decision Layer: This is the core of the system, integrating various AI algorithms and models. It analyzes and predicts processed data through deep learning algorithms, such as using Recurrent Neural Networks (RNNs) and their variant, Long Short-Term Memory (LSTM), to perform time-series analysis on energy consumption data and predict future energy demand; it uses classification algorithms such as Support Vector Machines (SVMs) and Random Forests to classify equipment operating status data and determine whether the equipment is operating normally; and it uses reinforcement learning algorithms to optimize intelligent control strategies, such as learning how to optimally control the operation of ventilation, lighting, air conditioning, and other equipment based on different environmental conditions and personnel activities through continuous interaction with the environment.

[0035] Application layer: Provides users with interfaces for various management functions, including a management platform for administrators and a mobile application for users; administrators can use the management platform to view various data of the building and the operating status of equipment in real time, and make remote control and management decisions; users can use the mobile application to query environmental information and equipment status in the building, and submit service requests, etc.

[0036] Data collection:

[0037] Environmental data acquisition: Temperature and humidity sensors and air quality sensors are installed in various rooms, corridors, meeting rooms, and other areas of the building to collect environmental data in real time at 5-10 minute intervals; for example, the data collected by the temperature and humidity sensors is recorded as follows. and , respectively representing the first The region in the first Temperature and humidity at any given time; carbon dioxide concentration data collected by the air quality sensor is recorded as... Formaldehyde concentration data is recorded as wait;

[0038] Personnel flow data collection: Install personnel flow sensors at building entrances, elevator entrances, main passages, etc., and use infrared sensing, video analysis and other technologies to count the number and direction of personnel entering and exiting in real time; through video analysis technology, use target detection algorithms such as YOLO algorithm to identify human targets in the video, and use tracking algorithms such as Hungarian algorithm to track people in real time, thereby obtaining information such as personnel flow trajectory and dwell time;

[0039] Equipment operation data acquisition: For equipment such as elevators, air conditioners, and lighting, the equipment's operating parameters are collected through the built-in intelligent monitoring module or external sensors. These parameters include elevator floor level, operating speed, and fault alarm information; air conditioner cooling / heating power, fan speed, and operating mode; and lighting equipment on / off status and brightness. This data is then transmitted to the data acquisition layer via wired or wireless communication.

[0040] AI algorithms:

[0041] Application of deep learning algorithms in environmental data analysis: Constructing a multilayer perceptron (MLP) model, using collected environmental data as input, to predict changes in environmental parameters over a future period; for example, inputting temperature, humidity, and carbon dioxide concentration data from the current moment and several previous moments to predict temperature changes over the next 1-2 hours. The formula is as follows: in The mapping function represents the multilayer perceptron model. The model is trained with a large amount of historical data, and the weights and biases of the model are adjusted so that it can accurately predict changes in environmental parameters.

[0042] Application of machine learning algorithms in equipment fault prediction: Support Vector Machine (SVM) algorithm is used to classify equipment operating data to determine whether the equipment is in normal operating condition. First, feature engineering is performed on the equipment operating data to extract features such as equipment operating time, temperature, and vibration, and these features are then combined into feature vectors. Then, the SVM model is trained using historical fault data and normal operation data to obtain a classification model. When new equipment operation data is available, its feature vector is input into the trained SVM model, and the model outputs the equipment's status category as "normal or faulty".

[0043] Application of reinforcement learning algorithms in intelligent control: Taking the control of a ventilation system as an example, the agent is defined as the ventilation system controller, the environment is the air quality and personnel activity within the building, the action space is the operation of opening / closing the ventilation system, adjusting the wind speed, etc., and the reward function is determined based on factors such as the degree of air quality improvement and energy consumption. Through continuous interaction with the environment, the agent learns which actions to take under different environmental conditions to obtain the maximum reward, thereby achieving intelligent control of the ventilation system. For example, when the air quality is poor and there are many people, the agent learns that increasing the ventilation volume can improve the air quality and obtain a positive reward; while excessive ventilation leads to energy waste and obtains a negative reward. Through continuous learning and optimization, the agent can find the optimal control strategy.

[0044] Management functions:

[0045] Intelligent ventilation control: Based on data collected by air quality sensors and personnel flow information, combined with the analysis results of AI algorithms, the system automatically controls the operation of the ventilation system. When the indoor carbon dioxide concentration exceeds a set threshold such as 1000 ppm or there is an odor, the system automatically starts the ventilation equipment and adjusts the ventilation volume according to the number of people and the size of the space. For example, in a crowded conference room, when the carbon dioxide concentration reaches 1200 ppm, the ventilation system will run at maximum wind speed for 15-20 minutes to reduce the carbon dioxide concentration to a normal level, and then adjust to a suitable wind speed for continuous ventilation.

[0046] Energy Management: By analyzing electricity consumption data collected by smart meters, AI algorithms are used to predict energy demand at different times and optimize energy allocation. For lighting systems, brightness or on / off status is automatically adjusted based on ambient light intensity and human activity. For example, in areas with ample natural light during the day, lighting is automatically dimmed or turned off; lighting is automatically turned off after people have left the room for a period of time. For air conditioning systems, temperature setpoints and operating modes are dynamically adjusted based on indoor and outdoor temperatures, the number of people, and their activities. In summer, when there are fewer people indoors and the outdoor temperature is not particularly high, the air conditioning setpoint is appropriately increased to reduce energy consumption. Through these measures, energy consumption is expected to be reduced by 15%-25%.

[0047] Security monitoring: Utilizing video image data collected by cameras, AI video analysis technology enables functions such as personnel behavior recognition, intrusion detection, and fire early warning. Employing target detection and behavior analysis algorithms, it monitors personnel behavior in real time within the video feed. Upon detecting abnormal behavior such as fighting or falling, it immediately issues an alarm to notify security personnel. For intrusion detection, virtual warning lines are set up; when a person or object crosses the line, an alarm mechanism is triggered. For fire early warning, image recognition technology detects smoke and flames in the video; once an anomaly is detected, the fire alarm system is quickly activated, and related equipment such as fire pumps and sprinkler systems are activated.

[0048] Equipment maintenance management: By monitoring and analyzing equipment operation data in real time, AI algorithms are used to predict potential equipment failures, allowing for advance maintenance and reducing losses caused by equipment malfunctions. For example, for elevators, by monitoring parameters such as operating speed, vibration, and current, machine learning algorithms are used to build a fault prediction model. When the model predicts that the elevator may malfunction in the near future, maintenance personnel are notified in advance to conduct inspections and maintenance, preventing sudden elevator failures that could trap people. Simultaneously, the system can also optimize maintenance plans based on the equipment's maintenance history and operating conditions, rationally allocating maintenance time and resources to reduce maintenance costs.

[0049] Example 1:

[0050] Scenario setting: This example uses a 50-story modern office building as an example. The building houses many companies covering multiple industries such as finance, technology, and media, with a large number of daily office workers, approximately 5,000. The building has complex internal functional areas, including open-plan offices, private offices, meeting rooms, restaurants, gyms, etc., which places high demands on building management.

[0051] Data Acquisition: Sensors and cameras are widely deployed throughout the office building; temperature and humidity sensors, carbon dioxide sensors, formaldehyde sensors, and PM2.5 sensors are installed in each office area, meeting room, corridor, etc., collecting environmental data every 5 minutes to ensure timely detection of changes in air quality and temperature and humidity; for example, in a 200-square-meter open-plan office area, 5 temperature and humidity sensors and 3 air quality sensors are evenly distributed to ensure comprehensive and accurate data collection; personnel flow sensors are installed at the office building entrances, elevator entrances, and main passageways, utilizing infrared sensing and visual... Video analytics technology is used to track the number and flow of people entering and exiting in real time. At elevator entrances, video analytics employs the YOLO algorithm to identify human targets in videos and uses the Hungarian algorithm to track people in real time, thereby obtaining information such as their movement trajectory and dwell time. Smart meters are installed inside elevators to collect real-time data on elevator power consumption. Smart water meters are installed in air conditioning rooms to monitor the water consumption of the air conditioning system. Meanwhile, high-definition cameras are installed in various corners of the office building, covering office areas, public areas, parking lots, etc., to collect video image data, providing data support for security monitoring and personnel behavior analysis.

[0052] Data Processing and Analysis: After data acquisition, the data first enters the data processing layer. Data cleaning algorithms are used to remove noise, outliers, and duplicate data. For example, for data collected by temperature and humidity sensors, if there are sudden temperature changes or abnormally high humidity values, the data cleaning algorithm identifies these as outliers and corrects or removes them. Fourier transform is used to extract frequency domain features from the time-series environmental data, transforming the raw time-series data into feature vectors suitable for AI algorithm processing. For video image data, a convolutional neural network (CNN) is used to extract image features, extracting key features such as people, objects, and scenes. The processed data is then transmitted to the decision-making layer, where a multilayer perceptron (MLP) model is used to predict environmental parameter changes over the next 1-2 hours. Taking temperature prediction as an example, the current temperature, humidity, and carbon dioxide concentration data from the previous few moments are input, and the prediction is processed using a formula... Predicting future temperature changes Simultaneously, the Support Vector Machine (SVM) algorithm is used to classify the operating data of equipment such as elevators and air conditioners to determine whether the equipment is in normal operating condition; the feature vectors of the equipment operating data are then processed. Input a trained SVM model and output the device state category;

[0053] System Execution: Based on the analysis results of the AI ​​algorithm, the system automatically executes corresponding operations. When the intelligent ventilation control module detects that the indoor carbon dioxide concentration exceeds 1000 ppm, it automatically starts the ventilation equipment. In densely populated conference rooms, if the carbon dioxide concentration reaches 1200 ppm, the ventilation system will run at maximum wind speed for 15 minutes to quickly reduce the carbon dioxide concentration, and then adjust to a suitable wind speed for continuous ventilation to maintain good indoor air quality. In terms of energy management, the lighting system automatically adjusts brightness or switches on / off according to the ambient light intensity and personnel activity. In areas with sufficient natural light during the day, it automatically dims or turns off the lighting equipment. It automatically turns off the lighting 30 minutes after personnel leave the office area. The air conditioning system dynamically adjusts the temperature setpoint and operating mode according to indoor and outdoor temperatures, the number of people, and their activities. In summer, when there are fewer people indoors and the outdoor temperature is not particularly high, the air conditioning setpoint is increased by 2°C to reduce energy consumption.

[0054] Implementation Results: After implementing the AI-based intelligent building management system of this invention, the air quality of the office building was significantly improved; the average indoor carbon dioxide concentration decreased by 250 ppm, from the previous average of 1100 ppm to 850 ppm; formaldehyde concentration decreased by 35%, and PM2.5 concentration also decreased significantly; in terms of energy conservation, through intelligent energy management, the energy consumption of the lighting system decreased by 32%, and the energy consumption of the air conditioning system decreased by 23%; the formula for calculating energy savings is as follows: ,in To save energy, Energy consumption before implementation This refers to energy consumption after implementation; taking the lighting system as an example, the monthly electricity consumption for lighting before implementation was... The monthly electricity consumption for lighting after implementation is [degree] The monthly energy savings of the lighting system are calculated based on the degree of energy consumption. Overall energy consumption was reduced by 21%, effectively lowering the office building's operating costs while providing a more comfortable and healthier working environment for office workers and improving work efficiency.

[0055] Example 2:

[0056] Scenario setting: This example uses a large residential community as an example. The community has 20 high-rise residential buildings and about 1,500 households. The community is equipped with public facilities such as kindergartens, gyms, and supermarkets. The residents live in a relatively concentrated area, have diverse living needs, and have high requirements for the comfort and safety of the living environment.

[0057] Functional focus: In residential settings, the system places greater emphasis on security monitoring and improving residents' living comfort; through high-definition cameras, smart access control and other equipment, it achieves strict control and real-time monitoring of people entering and leaving the community to ensure the safety of residents' lives and property; at the same time, through environmental monitoring equipment and smart home appliance control systems, it adjusts indoor and outdoor environmental parameters in real time to provide residents with a comfortable living environment;

[0058] Implementation Process: Facial recognition access control systems and high-definition cameras were installed at the entrances and exits of the residential area and at the entrances of individual buildings. Facial recognition technology was used to verify the identity of people entering and exiting, while video image data was collected by the cameras to monitor people's activities in real time. Multiple cameras were installed in internal roads, parking lots, and public areas to form a comprehensive monitoring network. Smart meters were installed in the elevators of each building to collect elevator power consumption data. Smart meters and smart water meters were installed in residents' homes to monitor residents' electricity and water consumption in real time. Temperature and humidity sensors and air quality sensors were installed in residents' homes and public areas to collect environmental data every 10 minutes. After data collection, the data was cleaned, preprocessed, and feature extracted. Data cleaning algorithms were used to remove noise and outliers from the data, and convolutional neural networks were used to refine the video image data. The system extracts facial and behavioral features of individuals; uses a multilayer perceptron model to predict changes in environmental parameters over a future period, such as trends in indoor temperature and humidity; employs a support vector machine algorithm to classify operational data from elevators, smart home appliances, and other devices to determine if they are functioning correctly; and executes corresponding operations based on the analysis results. When the security monitoring module detects abnormal behavior, such as strangers loitering in the community for an extended period or illegally entering, it immediately issues an alarm to notify security personnel. Through the smart home appliance control system, the system automatically adjusts the operating status of home appliances based on residents' lifestyles and environmental data; for example, in summer, when the indoor temperature exceeds 28°C, the air conditioner automatically starts and sets the temperature to 26°C; at night, after residents fall asleep, the system automatically adjusts light brightness and appliance operating modes to reduce noise and energy consumption.

[0059] Implementation Results: After implementing this system, the incidence of security incidents in the community significantly decreased; the alarm accuracy rate for intrusion incidents reached 96%, and the false alarm rate for abnormal behavior detection decreased by 52%, effectively ensuring residents' safety; residents' living comfort was significantly improved, with indoor temperature and humidity maintained within a suitable range through intelligent adjustment of indoor environmental parameters, maintaining a temperature of 24-26℃ and humidity of 40%-60%; resident satisfaction increased substantially, as shown in the questionnaire survey. The formula for calculating resident satisfaction is... ,in For residents' satisfaction, To satisfy the number of residents, The total number of residents; the survey results show that the total number of residents who participated in the survey... Households, of which the number of satisfied residents Households, then resident satisfaction This represents a 20 percentage point increase compared to before implementation, indicating that residents are more satisfied with the living environment in their community.

[0060] Example 3:

[0061] Scenario setting: This embodiment takes a large commercial complex as an example. The commercial complex integrates shopping, dining, entertainment and office space, covering an area of ​​approximately 100,000 square meters. It has a 5-story shopping center, 2 office buildings, 1 hotel and an underground parking lot. The daily traffic is huge, reaching tens of thousands of people during peak hours. The functional areas are complex and diverse, and the requirements for the efficiency and coordination of building management are extremely high.

[0062] Addressing Complex Situations: Faced with the complexities of densely populated commercial complexes and diverse functional areas, the system comprehensively collects data from various areas using numerous sensors and cameras. Personnel flow sensors, temperature and humidity sensors, air quality sensors, and cameras are installed in shops, corridors, and entrances of shopping malls to monitor personnel flow, environmental parameters, and safety conditions in real time. In office building areas, the system focuses on collecting data on office environment and equipment operation. In hotel areas, it monitors guest room environment and service demand data. Through real-time analysis and processing of this massive amount of data, the system can make quick and accurate decisions to address various complex situations.

[0063] Collaborative Management: The system enables collaborative management of different functional areas. In terms of energy management, energy is allocated rationally based on the usage time and demand of different areas. For example, during peak business hours in shopping malls, the energy supply for air conditioning and lighting is increased; after office hours, energy consumption in non-essential areas is reduced. In terms of security monitoring, cameras and alarm systems in various areas are linked, so that if a security incident occurs in one area, other areas can respond quickly and assist in handling it. In terms of equipment maintenance management, the system provides unified monitoring and maintenance planning for the equipment throughout the commercial complex, improving equipment operating efficiency and lifespan.

[0064] Implementation Results: After implementing the system of this invention, the energy costs of commercial complexes were significantly reduced; through intelligent energy management, overall energy consumption was reduced by 18%; operational efficiency was greatly improved. Taking parking lots as an example, through license plate recognition and intelligent parking guidance systems, the average parking time was shortened from 15 minutes to 8 minutes; the formula for improving operational efficiency is as follows: ,in To improve operational efficiency. This represents the average time to complete the business before implementation. The average time to complete the business after implementation; therefore, the percentage increase in parking lot operational efficiency. In terms of security, the incidence of security incidents has decreased by 35%, providing a safer and more comfortable environment for businesses and customers, and effectively improving the economic and social benefits of the commercial complex.

[0065] The terms "front," "back," "left," "right," "top," and "bottom" all refer to the figures in the accompanying drawings. Figure 1 Based on the perspective of the observer, the side of the device facing the observer is defined as the front, the left side of the observer is defined as the left, and so on.

[0066] In the description of this invention, it should be understood that the terms "center", "longitudinal", "lateral", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting the scope of protection of this invention.

[0067] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An AI-based intelligent building management system, characterized in that, include: The data acquisition layer consists of sensors, cameras, smart meters, and smart water meters distributed throughout the building, used to collect environmental data, personnel flow data, and equipment operation data within the building; The data processing layer cleans, preprocesses, and extracts features from the data acquired by the data acquisition layer, and performs parallel processing using a cloud computing platform or edge computing device. The AI ​​decision-making layer integrates multiple AI algorithms and models. It analyzes and predicts the processed data through deep learning algorithms, classifies and judges the operating status of equipment using machine learning algorithms, and optimizes intelligent control strategies using reinforcement learning algorithms. The application layer provides management function interfaces for users, including the management platform used by administrators and the mobile application used by users.

2. The AI-based intelligent building management system according to claim 1, characterized in that, In the data acquisition layer, temperature sensors collect indoor temperature data at 5-10 minute intervals. The humidity sensor collects indoor humidity data at equal intervals. Air quality sensors collect data on carbon dioxide concentration in the air. Formaldehyde concentration data Personnel flow sensors use infrared sensing and video analysis technology to count the number and direction of people entering and leaving in real time. Cameras use target detection and tracking algorithms to obtain information such as personnel flow trajectory and dwell time. Smart meters and smart water meters collect data on electricity consumption and water consumption, respectively.

3. The AI-based intelligent building management system according to claim 1, characterized in that, In the data processing layer, data cleaning algorithms are used to remove noise, outliers and duplicate data from the data. Fourier transform and other methods are used to extract frequency domain features from time series environmental data. Convolutional neural networks (CNNs) are used to extract image features from video image data.

4. The AI-based intelligent building management system according to claim 1, characterized in that, In the decision-making layer, the deep learning algorithm uses a multilayer perceptron (MLP) model to predict changes in environmental parameters, as shown in the formula: ,in The mapping function represents the multilayer perceptron model; the machine learning algorithm uses a support vector machine (SVM) to classify and judge the equipment operation data, and converts the feature vectors of the equipment operation data... Input a trained SVM model and output the device state category; take ventilation system control as an example of reinforcement learning algorithm, define the agent as ventilation system controller, the environment as air quality and personnel activity in the building, the action space as the operation of opening / closing the ventilation system, wind speed adjustment, etc., and the reward function is determined according to factors such as the degree of air quality improvement and energy consumption.

5. The AI-based intelligent building management system according to claim 1, characterized in that, In the intelligent ventilation control function of the application layer, when the indoor carbon dioxide concentration exceeds 1000ppm or there is an odor, the system automatically starts the ventilation equipment and adjusts the ventilation volume according to the number of people and the size of the space. In a densely populated conference room, when the carbon dioxide concentration reaches 1200ppm, the ventilation system runs at the maximum wind speed for 15-20 minutes and then adjusts to a suitable wind speed for continuous ventilation.

6. The AI-based intelligent building management system according to claim 1, characterized in that, In the energy management function of the application layer, by analyzing the power consumption data collected by smart meters, AI algorithms are used to predict energy demand at different times. The lighting system automatically adjusts the brightness or switches on and off according to the ambient light intensity and the activity of people. The air conditioning system dynamically adjusts the temperature setpoint and operating mode according to the indoor and outdoor temperature, the number of people and their activities. It is expected to reduce energy consumption by 15%-25%.

7. The AI-based intelligent building management system according to claim 1, characterized in that, In the security monitoring function of the application layer, target detection and behavior analysis algorithms are used to monitor the behavior of people in the video in real time. When abnormal behavior is detected, an alarm is issued. Intrusion detection is achieved by setting virtual warning lines. Fire warning is achieved by using image recognition technology to detect smoke and flames in the video and to link relevant fire-fighting equipment.

8. The AI-based intelligent building management system according to claim 1, characterized in that, In the equipment maintenance management function of the application layer, by monitoring parameters such as equipment operating speed, vibration, and current, a fault prediction model is established using machine learning algorithms. When the model predicts that the equipment may fail in the future, it notifies maintenance personnel in advance to carry out inspection and maintenance, and optimizes the maintenance plan based on the equipment maintenance history and operating conditions.

9. An AI-based intelligent building management method, characterized in that, Includes the following steps: Environmental data, personnel flow data, and equipment operation data are collected within the building through sensors, cameras, smart meters, and smart water meters distributed throughout the building. The collected data is cleaned, preprocessed, and feature extracted, and then processed in parallel using a cloud computing platform or edge computing device. The processed data is analyzed and predicted using deep learning algorithms, the equipment operating status is classified and judged using machine learning algorithms, and the intelligent control strategy is optimized using reinforcement learning algorithms. Based on the optimized control strategy, the application layer provides users with functions such as intelligent ventilation control, energy management, security monitoring, and equipment maintenance management.

10. The AI-based intelligent building management method according to claim 9, characterized in that, In the data acquisition step, the specific data and methods acquired are consistent with the data acquisition layer content described in claim 2; in the data processing step, the specific processing methods are consistent with the data processing layer content described in claim 3; in the AI ​​algorithm processing step, the specific algorithm application is consistent with the decision layer content described in claim 4; in the management function implementation step, the execution methods of each management function are consistent with the application layer content described in claims 5-8.

Images