A building monitoring and management system based on the Internet of Things

Abstract

The present invention relates to the technical field of alarm devices and discloses a building monitoring and management system based on the Internet of Things. The system comprises: when an alarm sensor generates an early warning signal, a circumstantial evidence confirmation module captures power grid harmonic data and compares it with a benign equipment feature library, and decides whether to trigger an alarm based on the matching result. The present invention fundamentally distinguishes between real fire conditions and equipment operation interference through a time-series mutual verification mechanism between power grid harmonics and alarm signals, significantly improving alarm accuracy. At the same time, the difference in multi-branch harmonic intensity is used to achieve passive positioning, eliminating the risk of false suppression in large buildings, and monitoring the health status of equipment through the harmonic entropy change trend, forming a complete safety closed loop from early warning to prevention.

Description

Technical Field

[0001] The present invention relates to a building monitoring and management system based on the Internet of Things, and belongs to the technical field of alarm devices. Background Art

[0002] The current mainstream system relies on a multi-sensor threshold trigger mechanism. When the smoke or temperature sensor detects that the parameters exceed the standard, the alarm is triggered. Although this method is direct, it has fundamental limitations. Taking large commercial complexes as an example, kitchen cooking fumes often trigger smoke alarms. The sensitive threshold set to reduce the false alarm rate causes the system to be overly sensitive to benign activities such as equipment start-up and shutdown and environmental fluctuations, resulting in a technical problem of high false alarm rate.

[0003] Existing improvements attempt to introduce multi-source data fusion algorithms to improve judgment accuracy by analyzing auxiliary parameters such as temperature, humidity, and population density. Such solutions require the deployment of additional sensors and rely on high-computing platforms for real-time calculations, which not only significantly increases system complexity and cost, but also makes its decision-making logic difficult to verify due to the characteristics of the algorithm. In complex fault scenarios such as sudden circuit overloads, it is still difficult to distinguish the essential differences between air conditioner startup harmonics and electrical spark noise.

[0004] A thorough analysis reveals three fundamental bottlenecks in existing technologies: 1. Alarm decision-making lacks the ability to dynamically perceive the building's operating context, failing to factor in equipment activity status; 2. The spatial correlation of cross-regional events in large buildings is difficult to verify through conventional means, posing the risk of false suppression; and 3. The system focuses solely on immediate alarm functionality, failing to exploit the implicit representations of early electrical equipment degradation within the grid's characteristics. Therefore, the technical challenge addressed by this invention is to develop a lightweight alarm mechanism that leverages endogenous signals for contextual self-awareness, possesses spatial positioning capabilities, and can provide early warning of equipment hazards. Summary of the Invention

[0005] The present invention provides a building monitoring and management system based on the Internet of Things. Its main purpose is to solve the problems of high false alarm rate due to the lack of situational awareness capability of the building alarm system, the risk of false suppression caused by spatial positioning ambiguity, and the lack of equipment health monitoring affecting the fire prevention effect.

[0006] To achieve the above objectives, the present invention provides a building monitoring and management system based on the Internet of Things, the system comprising:

[0007] an alarm sensor configured to generate an internal warning signal when a set warning condition is detected, the warning signal including location information;

[0008] The circumstantial evidence confirmation module is coupled to the building's power grid and is configured to: in response to an internal warning signal, capture time-domain harmonic data of the power grid within a predetermined time window before and after the warning signal is generated; perform a fast Fourier transform on the time-domain harmonic data to extract frequency-domain features of the harmonic data, which represent the activity of electrical devices within the power grid;

[0009] The processor is electrically connected to the alarm sensor and the circumstantial evidence confirmation module and is configured to: store a pre-calibrated baseline library of harmonic characteristics of benign equipment operation, where the baseline library is generated by actively operating high-power electrical equipment in a building when the building is in a non-alarm state and recording the harmonic frequency domain characteristics of the equipment during stable operation; perform pattern matching on the harmonic frequency domain characteristics extracted in real time by the circumstantial evidence confirmation module with the baseline library of harmonic characteristics of benign equipment operation to generate a matching result; and perform situational judgment on the internal warning signal based on the matching result: if the matching degree between the real-time harmonic frequency domain characteristics and any benign harmonic characteristics in the baseline library reaches a preset threshold, the warning signal is judged to be a benign event, and the conversion of the warning signal into an external alarm is suppressed; if the matching degree between the real-time harmonic frequency domain characteristics and any benign harmonic characteristics in the baseline library does not reach the preset threshold, and the power grid exhibits characteristics related to electrical faults, the warning signal is confirmed and an external alarm is triggered; if the power grid harmonic change does not reach a preset minimum change threshold near the time when the warning signal is generated, the warning signal is confirmed by default and an external alarm is triggered.

[0010] Preferably, the alarm sensor is any one of a smoke sensor, a temperature sensor, a gas sensor, a flame sensor, or a combination thereof; the circumstantial evidence confirmation module includes a current transformer, an analog-to-digital converter, and a microcontroller equipped with a fast Fourier transform module, and the microcontroller is used to process time domain harmonic data.

[0011] Preferably, pattern matching is achieved by calculating the Euclidean distance between the real-time harmonic frequency domain features and the benign harmonic features in the baseline library, wherein the matching degree reaches a preset threshold value means that the Euclidean distance is less than or equal to a preset distance threshold value:

[0012] ,

[0013] in, Represents the real-time harmonic frequency domain feature vector Compared with the benign harmonic feature vector in the baseline library The Euclidean distance between Indicates the preset distance threshold.

[0014] Preferably, the calibration of the baseline library of harmonic characteristics of benign equipment operation includes: starting any one or more high-power electrical equipment such as kitchen equipment, cleaning machines, large air conditioners, and elevators in the building in sequence; collecting the corresponding power grid time domain harmonic data when the equipment is started, running stably, and stopped; performing fast Fourier transform on the collected time domain harmonic data to extract its energy distribution characteristics in a specific frequency band to form the harmonic characteristics of benign equipment operation.

[0015] Preferably, the circumstantial evidence confirmation modules include at least two, which are respectively deployed in different power branches of the building; the processor is configured to: receive the signal strength of the harmonic data captured by the at least two circumstantial evidence confirmation modules and matched with the baseline feature library; determine the approximate position of the harmonic source based on the signal strength, wherein the position of the circumstantial evidence confirmation module with the largest signal strength is determined to be closest to the harmonic source; only when the approximate position of the harmonic source is adjacent to the position of the alarm sensor that generates the internal warning signal, the processor suppresses the warning signal from being converted into an external alarm.

[0016] Preferably, the processor is also configured to: when the real-time harmonic frequency domain characteristics do not match any benign harmonic characteristics in the baseline library to a preset threshold, and when the power grid exhibits a preset broadband random noise characteristic generated by electrical sparks, the processor immediately confirms the early warning signal and triggers the highest priority external alarm.

[0017] Preferably, the processor is further configured to: when calibrating a certain device, calculate and store the health information entropy baseline corresponding to the harmonic frequency domain characteristics of the device in a healthy operating state; in daily operation of the building, whenever the circumstantial confirmation module identifies that a certain device is operating, the processor calculates the current information entropy value corresponding to its real-time harmonic frequency domain characteristics; the processor generates pre-diagnostic information about the health status of a certain device based on the long-term change trend of the current information entropy value relative to the health information entropy baseline, wherein a continuous increase in the information entropy value indicates that the device has a degradation trend.

[0018] Preferably, the processor is configured to record all suppressed warning events, and the event log includes the warning type, occurrence time, alarm sensor location information and corresponding benign harmonic matching results.

[0019] Preferably, the system also includes a user interface module, which is configured to provide operation and maintenance personnel with the function of updating and maintaining the baseline library of harmonic characteristics of benign equipment operation, including one-click calibration of newly purchased equipment and recalibration of the characteristics of existing equipment.

[0020] Preferably, data is transmitted between the circumstantial evidence confirmation module and the processor via an Internet of Things communication protocol, which includes any one of MQTT, CoAP, or LoRaWAN, or a combination thereof.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] 1. By forcibly coupling the alarm sensor's early warning signal with the contextual characteristics of power grid harmonics, the system establishes a dual verification mechanism in alarm decision-making for the first time. When the smoke sensor triggers the early warning, the system does not respond directly, but instead synchronously captures the real-time status of the power grid harmonics. By comparing it with a preset benign device feature library, it naturally distinguishes between cooking fumes and actual fires. This temporal mutual verification mechanism between the early warning signal and the power context transforms the traditional single-point threshold judgment with a high false alarm rate into a closed-loop decision-making based on the evidence chain of environmental activities, thus avoiding false alarms caused by benign activities in principle.

[0023] 2. Multiple circumstantial evidence modules deployed in different power branches simultaneously record signal strength while capturing harmonic signatures. By comparing signal attenuation differences between the east and west wings, the system can automatically infer the harmonic source location based on the physical characteristics of the power grid, without the need for additional positioning hardware. When the west wing smoke warning and the east wing air conditioning harmonics coincide in time, the system determines spatial non-proximity to avoid false suppression due to coincidence. This design, which converts power grid propagation attenuation into spatial vectors, eliminates positioning blind spots in large buildings at a low cost.

[0024] 3. By reusing the harmonic analysis link, the system simultaneously calculates the harmonic information entropy value of kitchen equipment when identifying its operation, continuously tracks the slow drift trend of the entropy value relative to the healthy baseline, such as the spectrum disorder caused by motor bearing wear, and generates a pre-diagnosis work order. This allows the security system to naturally extend the early warning capability of electrical equipment degradation while completing the core alarm function. Operation and maintenance personnel shift from passively handling alarms to actively intervening in equipment health, suppressing the risk of electrical fires caused by equipment aging from the source.

[0025] 4. The benign device signature library allows operations and maintenance personnel to calibrate new equipment with one click, while the harmonic entropy change trend only indicates abnormal drift within a relaxed threshold. This design retains the final decision-making power of manual calibration while focusing machine computing power on continuous monitoring and anomaly screening. When a newly purchased industrial oven is connected to the power grid, operations and maintenance personnel can trigger calibration on-site to update system knowledge without the intervention of algorithm experts. The system thus forms an evolvable operating ecosystem. The simplicity of its operation path and the transparency of its decision logic surpass those of traditional adaptive algorithms. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a comparison diagram of harmonic signals of the induction cooker and the electric spark at different frequencies of the present invention;

[0027] Figure 2 This is a schematic diagram of the power grid signal acquisition and analysis process of the present invention;

[0028] Figure 3This is an event alarm sequence diagram of the building monitoring and management system based on the Internet of Things of the present invention;

[0029] Figure 4 Comparison chart of harmonic characteristics of benign equipment and faulty equipment;

[0030] Figure 5 Comparative analysis chart of false alarm rate and detection accuracy;

[0031] Figure 6 Multi-branch harmonic signal strength distribution diagram.

[0032] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0033] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0034] The present invention provides a building monitoring and management system based on the Internet of Things, the system comprising:

[0035] an alarm sensor configured to generate an internal warning signal when a set warning condition is detected, the warning signal including location information;

[0036] The circumstantial evidence confirmation module is coupled to the building's power grid and is configured to: in response to the internal warning signal, capture time-domain harmonic data of the power grid within a predetermined time window before and after the warning signal is generated, for example, within a time window of 500 milliseconds before and after the warning signal is generated; perform a fast Fourier transform on the time-domain harmonic data to extract frequency-domain features of the harmonic data, the frequency-domain features representing the activity of electrical devices within the power grid;

[0037] The processor is electrically connected to the alarm sensor and the circumstantial evidence confirmation module and is configured to: store a pre-calibrated baseline library of harmonic characteristics of benign equipment operation, where the baseline library is generated by actively operating high-power electrical equipment in a building when the building is in a non-alarm state and recording the harmonic frequency domain characteristics of the equipment during stable operation; perform pattern matching on the harmonic frequency domain characteristics extracted in real time by the circumstantial evidence confirmation module with the baseline library of harmonic characteristics of benign equipment operation to generate a matching result; and perform situational judgment on the internal warning signal based on the matching result: if the matching degree between the real-time harmonic frequency domain characteristics and any benign harmonic characteristics in the baseline library reaches a preset threshold, the warning signal is judged to be a benign event, and the conversion of the warning signal into an external alarm is suppressed; if the matching degree between the real-time harmonic frequency domain characteristics and any benign harmonic characteristics in the baseline library does not reach the preset threshold, and the power grid exhibits characteristics related to electrical faults, the warning signal is confirmed and an external alarm is triggered; if the power grid harmonic change does not reach a preset minimum change threshold near the time when the warning signal is generated, the warning signal is confirmed by default and an external alarm is triggered.

[0038] Preferably, the alarm sensor is any one of a smoke sensor, a temperature sensor, a gas sensor, a flame sensor, or a combination thereof; the circumstantial evidence confirmation module includes a current transformer, an analog-to-digital converter, and a microcontroller equipped with a fast Fourier transform module, and the microcontroller is used to process time domain harmonic data.

[0039] Preferably, pattern matching is achieved by calculating the Euclidean distance between the real-time harmonic frequency domain features and the benign harmonic features in the baseline library, wherein the matching degree reaches a preset threshold value means that the Euclidean distance is less than or equal to a preset distance threshold value:

[0040] ,

[0041] in, Represents the real-time harmonic frequency domain feature vector Compared with the benign harmonic feature vector in the baseline library The Euclidean distance between Indicates the preset distance threshold.

[0042] Preferably, the calibration of the baseline library of harmonic characteristics of benign equipment operation includes: starting any one or more high-power electrical equipment such as kitchen equipment, cleaning machines, large air conditioners, and elevators in the building in sequence; collecting the corresponding power grid time domain harmonic data when the equipment is started, running stably, and stopped; performing fast Fourier transform on the collected time domain harmonic data to extract its energy distribution characteristics in a specific frequency band to form the harmonic characteristics of benign equipment operation.

[0043] Preferably, the circumstantial evidence confirmation modules include at least two, which are respectively deployed in different power branches of the building; the processor is configured to: receive the signal strength of the harmonic data captured by the at least two circumstantial evidence confirmation modules and matched with the baseline feature library; determine the approximate position of the harmonic source based on the signal strength, wherein the position of the circumstantial evidence confirmation module with the largest signal strength is determined to be closest to the harmonic source; only when the approximate position of the harmonic source is adjacent to the position of the alarm sensor that generates the internal warning signal, the processor suppresses the warning signal from being converted into an external alarm.

[0044] Preferably, the processor is also configured to: when the real-time harmonic frequency domain characteristics do not match any benign harmonic characteristics in the baseline library to a preset threshold, and when the power grid exhibits a preset broadband random noise characteristic generated by electrical sparks, the processor immediately confirms the early warning signal and triggers the highest priority external alarm.

[0045] Preferably, the processor is further configured to: when calibrating a certain device, calculate and store the health information entropy baseline corresponding to the harmonic frequency domain characteristics of the device in a healthy operating state; in daily operation of the building, whenever the circumstantial confirmation module identifies that a certain device is operating, the processor calculates the current information entropy value corresponding to its real-time harmonic frequency domain characteristics; the processor generates pre-diagnostic information about the health status of a certain device based on the long-term change trend of the current information entropy value relative to the health information entropy baseline, wherein a continuous increase in the information entropy value indicates that the device has a degradation trend.

[0046] Preferably, the processor is configured to record all suppressed warning events, and the event log includes the warning type, occurrence time, alarm sensor location information and corresponding benign harmonic matching results.

[0047] Preferably, the system also includes a user interface module, which is configured to provide operation and maintenance personnel with the function of updating and maintaining the baseline library of harmonic characteristics of benign equipment operation, including one-click calibration of newly purchased equipment and recalibration of the characteristics of existing equipment. For example, after the new equipment is connected to the system, the user can perform a one-click calibration operation through the user interface module. Specifically, the operation and maintenance personnel select the target equipment through the graphical interface and click the calibration button. The system then suspends the start and stop tasks of other high-power equipment within the set calibration window period (the default is 10 seconds) to keep the grid interference at a minimum, and continuously collects power data through the circumstantial confirmation module deployed on the power branch. The collected signal is converted into a digital signal and then the processor performs a fast Fourier transform to extract the frequency domain features of the 50Hz to 3kHz frequency band. The system automatically determines the spectrum stability and determines whether the health calibration conditions are met in combination with the operating status of the equipment. If the conditions are met, the system stores the current frequency domain feature vector in the harmonic feature baseline library of benign equipment operation and binds it with the equipment code for storage. The entire process does not require user intervention and the feature entry of the equipment is completed through a single operation. It has the advantages of high consistency, easy deployment, and traceability, which can ensure the accuracy of the system in identifying the operating status of new equipment. These are all extended implementation methods that can be known to ordinary technicians in this field.

[0048] Preferably, data is transmitted between the circumstantial evidence confirmation module and the processor via an Internet of Things communication protocol, which includes any one of MQTT, CoAP, or LoRaWAN, or a combination thereof.

[0049] Example 1: The present invention provides a building monitoring and management system based on the Internet of Things, which includes an alarm sensor, a circumstantial evidence confirmation module, a processor and a user interface module. The alarm sensor monitors environmental changes and generates an early warning signal containing location information according to the set early warning conditions; after receiving the early warning signal, the circumstantial evidence confirmation module captures the time domain harmonic data of the power grid and extracts the frequency domain features through fast Fourier transform. The processor compares it with the harmonic feature baseline library of benign equipment operation to determine whether to trigger the alarm signal; the alarm sensor is used to monitor environmental changes in the building, including smoke, temperature, gas concentration, etc. When the set early warning conditions are detected, the alarm sensor generates an early warning signal containing the location, which provides a trigger basis for the circumstantial evidence confirmation module. The circumstantial evidence confirmation module is connected to the power grid of the building and responds to the alarm The sensor captures the power grid's time-domain harmonic data within a 500-millisecond time window before and after the warning signal, and performs a fast Fourier transform on the data to extract frequency-domain features. These frequency-domain features are used to characterize the operating status of electrical equipment within the power grid, thereby helping to distinguish between normal equipment operation and actual fire conditions. The processor connects the alarm sensor and the circumstantial evidence confirmation module, receives the real-time harmonic frequency-domain features from the circumstantial evidence confirmation module, and performs pattern matching on them with the data in the benign equipment operation feature library. If the matching degree reaches the preset threshold, the warning signal is considered a benign event and the system suppresses the alarm. If the matching degree does not reach the threshold and an electrical fault signal is detected, an external alarm is triggered. Pattern matching is achieved by calculating the Euclidean distance between the real-time harmonic frequency-domain features and the benign features in the baseline library. The specific formula is:

[0050] ,

[0051] in, Represents the real-time harmonic eigenvector Compared with the benign feature vector in the baseline library The Euclidean distance between This calculation method ensures that the system can accurately distinguish between normally operating equipment and electrical faults based on the preset distance threshold. The system not only determines whether the signal comes from benign equipment, but also immediately triggers the highest priority external alarm when an electrical fault occurs in the power grid (such as an electrical spark). If the power grid harmonic change near the warning signal does not reach the preset threshold, the system will confirm the warning signal by default and trigger an alarm.

[0052] The alarm sensor can be any one of smoke, temperature, gas or flame sensors or a combination thereof. This configuration provides a variety of sensor options to meet different monitoring needs. The circumstantial evidence confirmation module includes a current transformer, an analog-to-digital converter and a microcontroller. The microcontroller processes time-domain harmonic data through fast Fourier transform, thereby improving the accuracy and timeliness of the data. The circumstantial evidence confirmation module is deployed in different power branches of the building and can capture harmonic signals at multiple locations in the power grid. By analyzing the differences in signal strength, the system can infer the approximate location of the harmonic source. When the warning of the smoke sensor coincides with the operating time of other equipment, the system can determine whether to trigger an alarm through spatial positioning. The processor is also configured to calculate and store the harmonic frequency domain characteristics and information entropy baseline in a healthy operating state when calibrating the equipment. This design enhances the system's ability to monitor the health status of the equipment and issues early warnings when the equipment is aging or failing. The present invention also effectively improves the accuracy of alarms and avoids the risk of false alarms in traditional solutions through the timing mutual verification mechanism of power grid harmonics and alarm signals. The multi-branch deployment of circumstantial confirmation modules realizes spatial positioning, significantly reducing the risk of false suppression in large buildings. At the same time, the system reuses the harmonic analysis link to extend the equipment health monitoring function, so that the security system can provide early warning of potential problems with electrical equipment in addition to the core alarm function, thereby improving building safety.

[0053] Example 2: This example provides a building monitoring and management system based on the Internet of Things, including an alarm sensor, a circumstantial evidence confirmation module, a processor, and a user interface module. The alarm sensor is used to monitor abnormal changes in the building environment and, when the set warning conditions are met, generates a warning signal containing location information. The alarm sensor can select one or a combination of smoke, temperature, gas, or flame sensors to ensure that the system has multiple monitoring capabilities. The circumstantial evidence confirmation module is deployed in the building power grid and responds to the warning signal generated by the alarm sensor. Within 500 milliseconds before and after the signal occurs, the circumstantial evidence confirmation module collects the time domain harmonic data of the power grid and uses fast Fourier transform to generate a warning signal. The Fast Fourier Transform (FFT) extracts frequency domain features, which reflect the operating status of power grid equipment and help distinguish between normal equipment operation and an actual fire. The processor receives signals from the alarm sensor and the circumstantial evidence confirmation module and stores a pre-calibrated harmonic feature library of benign equipment operation. This library records the harmonic features of equipment operating stably in the building's non-alarm state. The real-time harmonic features extracted by the circumstantial evidence confirmation module are matched with the data in this library, and the Euclidean distance is calculated to determine whether to trigger an alarm. The user interface module provides interactive functions between the user and the system, allowing operation and maintenance personnel to calibrate the equipment, update the feature library, and view the health status of the equipment.

[0054] When an alarm sensor generates a warning signal, the system does not immediately trigger an alarm. Instead, it compares the harmonic data of the circumstantial confirmation module with the benign device feature library to determine whether the signal is a false alarm. For example, when a smoke sensor detects smoke, the system captures the power grid harmonic data and compares it. If the harmonic signature matches the data in the baseline library, the system suppresses the alarm signal. If the signature does not match and the harmonic signature conforms to the pattern of an electrical fault, the alarm is triggered. The system achieves pattern matching by calculating the Euclidean distance between the real-time harmonic frequency domain signature and the benign device signature in the baseline library. The specific formula is: ,in, Represents the real-time harmonic frequency domain feature vector Compared with the benign feature vector in the baseline library The Euclidean distance between The algorithm uses a preset distance threshold to ensure that the system can accurately distinguish the harmonic characteristics of normal equipment from the harmonics caused by electrical faults. The circumstantial evidence confirmation module uses multiple modules deployed on different power branches to analyze signal strength differences to infer the approximate location of the harmonic source. When the smoke sensor triggers an alarm and overlaps with the harmonic signals of other equipment, the system determines whether the signal comes from an adjacent area based on the signal strength difference, thereby avoiding false suppression. The system monitors the health status of electrical equipment by multiplexing harmonic analysis links, calculates the harmonic information entropy value of the equipment during operation, and compares it with the health baseline. If the information entropy value continues to increase, it indicates that the equipment may be deteriorating. The system generates a pre-diagnosis work order, prompting the operation and maintenance personnel to conduct an inspection.

[0055] When the alarm sensor detects smoke, temperature or other abnormal conditions, it generates an early warning signal containing location information and transmits it to the circumstantial evidence confirmation module and processor. After receiving the early warning signal, the circumstantial evidence confirmation module captures the time domain harmonic data of the power grid and then extracts the frequency domain features through fast Fourier transform (FFT). The processor compares the extracted real-time harmonic frequency domain features with the data in the benign equipment feature library and calculates the matching degree using the Euclidean distance algorithm. If the matching degree meets the preset threshold, the system suppresses the alarm signal; if the matching degree does not meet the preset threshold and the harmonic feature shows an electrical fault, the alarm is triggered. The system divides the signal into two parts according to the signal strength of the multi-branch circumstantial evidence module. The system analyzes and infers the approximate location of the harmonic source. If the location with the maximum signal strength is close to the alarm sensor, the system suppresses the alarm signal; otherwise, an alarm is triggered. The system calculates the harmonic information entropy value of the equipment in real time and compares it with the healthy baseline. If the information entropy value is abnormal, a pre-diagnosis work order is generated to prompt the operation and maintenance personnel to conduct an inspection. Through the time-series mutual verification mechanism of harmonics and alarm signals, this system effectively distinguishes between equipment activities and real fire signals, significantly reducing the false alarm rate. The attenuation characteristics of the power grid signal are used to achieve passive positioning, avoiding the cost of setting up additional positioning hardware in large buildings. The system can detect equipment failure hazards in advance and improve building safety.

[0056] Example 3: This embodiment builds a test space with a typical floor structure and partition layout, which is equipped with functional partitions such as a simulated kitchen, office area, and equipment room. Each partition is equipped with controllable lighting, air conditioning, and exhaust fans, and contains an industrial-grade high-power heating device, which can be used as a potential electrical fault source or a benign high-power electrical device to simulate a real fire. This heating device can simulate a variety of operating states according to preset programs, including normal startup, shutdown, stable operation, and electrical fault harmonic characteristics caused by internal component overload or short circuit under specific conditions; alarm sensor array, in each key area of ​​the simulated building, strategically deployed a variety of types of alarm sensors, including smoke sensors, infrared temperature sensors, and combustible gas sensors. The trigger thresholds of these sensors are calibrated in accordance with the Chinese national standard "Design Specifications for Automatic Fire Alarm Systems" to ensure that they can generate internal warning signals and accurately record location information when typical warning conditions are detected. For example, the smoke sensor is set to trigger when the smoke concentration reaches The circumstantial evidence confirmation module network has four circumstantial evidence confirmation modules deployed on multiple independent power branches of the simulated building. Each module consists of a current transformer, an analog-to-digital converter, and a microcontroller equipped with a fast Fourier transform (FFT) module. The current transformer is used to collect the time domain current data on the monitored power branch in real time. The analog-to-digital converter converts the analog current signal into a digital signal, which is then processed by the microcontroller at high speed FFT to extract the current. Fundamental frequency and its The frequency domain characteristics of subharmonics are determined by the circumstantial evidence confirmation module and the processor via the MQTT IoT communication protocol. The central processing unit (CPU) utilizes a high-performance industrial-grade embedded computing platform pre-installed with a proprietary software system that receives real-time data streams from all alarm sensors and the circumstantial evidence confirmation module. The processor stores a pre-calibrated and continuously updated baseline library of harmonic characteristics of benign equipment operation. This baseline library is constructed by sequentially starting high-power electrical equipment such as a typical kitchen induction cooker, a large computer room air conditioner, and an elevator drive motor in a simulated building under non-alarm conditions and recording their harmonic frequency domain characteristics during startup, stable operation, and shutdown. For example, the harmonic characteristics of a certain model of induction cooker in stable heating mode are recorded as a feature in the baseline library. The user interface module provides a graphical user interface that supports operations and maintenance personnel in performing one-click calibration of the benign equipment characteristic library, viewing pre-diagnostic information on equipment health status, and querying historical alarm event logs.

[0057] This experiment aims to verify the ability of the present invention to distinguish between real fire and equipment operation interference through the timing mutual verification mechanism of power grid harmonics and alarm signals, thereby improving the accuracy of alarms. In this experiment, the warning thresholds of the smoke sensor and temperature sensor in the simulated kitchen area have been set. In the non-alarm state, by starting the induction cooker and exhaust fan in the simulated kitchen and making them run stably, a small amount of oil smoke is generated by heating cooking oil, causing the smoke sensor or temperature sensor to generate internal warning signals. These signals are not real fires. In another set of control experiments, the heat dissipation and a small amount of odor caused by long-term continuous operation of a high-power laser printer in the office area are simulated, which may cause the temperature sensor or gas sensor to trigger an early warning. When the alarm sensor generates an internal warning signal, the circumstantial evidence confirmation module is started synchronously. Within a millisecond time window, the time domain harmonic data on the power branch where it is located is captured. The microcontroller performs a fast Fourier transform on the captured time domain data to extract its frequency domain features, which represent the activity context of the electrical equipment in the power grid. The processor receives the harmonic frequency domain features extracted by the circumstantial evidence confirmation module in real time. The frequency domain features are then pattern matched with the benign harmonic features pre-stored in the benign equipment operation harmonic feature baseline library. Pattern matching is achieved by calculating the Euclidean distance, where the matching degree reaches a preset threshold when the Euclidean distance is less than or equal to a preset distance threshold. In this experiment, the distance threshold is set to 0.05. The fundamental technical consideration in setting this distance threshold is to achieve a technical optimization balance between the system's ability to suppress benign events (i.e., specificity) and its ability to identify real fire conditions (i.e., sensitivity). Specifically, if the distance threshold is set too high, the system's specificity will be excessively compromised, potentially misclassifying some non-benign events as benign, resulting in missed alarms. Conversely, if the distance threshold is set too low, the system's sensitivity will be excessively compromised, causing the system to become overly sensitive to background noise or subtle fluctuations in normal equipment, easily generating spurious responses and leading to false alarms. Therefore, in actual engineering practice, the distance threshold is not determined by a single absolute value. Instead, it needs to be set within a reasonable engineering range that optimizes the overall technical effect, based on the inherent noise baseline of the sensor used, the typical dynamic range of the power grid harmonic signal, and the minimum detection accuracy required for the building fire prevention problem addressed by the present invention. Those skilled in the art can readily perform routine and reasonable settings or optimizations based on the above technical principles and specific implementation conditions. In actual deployment, achieving this optimal balance can typically be achieved through the following iterative optimization procedure: First, an initial distance threshold is set. Second, in a test environment simulating real-world building operations, the system will continuously operate and record warning events triggered by the alarm sensor and the power grid harmonic data captured by the circumstantial evidence confirmation module. This data is then used to evaluate the system's specificity and sensitivity at the current distance threshold. If the evaluation results do not meet the preset overall performance target—for example, maintaining a true fire identification rate of over 99% while keeping the false alarm rate for benign events below 5%—then the distance threshold is fine-tuned based on the evaluation results and the preset adjustment rules. If the false alarm rate is too high, the threshold is appropriately raised; if the risk of missed alarms increases, the threshold is appropriately lowered. This process continues iteratively until the distance threshold achieves the optimal balance between specificity and sensitivity, ensuring reliable alarm detection in various operating scenarios.

[0058] Based on the pattern matching results, the processor makes a situational judgment on the internal warning signal: if the matching degree between the real-time harmonic frequency domain feature and any benign harmonic feature in the baseline library reaches the preset threshold, the warning signal is judged to be a benign event, and the warning signal is suppressed from being converted into an external alarm; if the matching degree between the real-time harmonic frequency domain feature and any benign harmonic feature in the baseline library does not reach the preset threshold, and the power grid exhibits characteristics related to electrical faults, the warning signal is confirmed and an external alarm is triggered; if the power grid harmonic change does not reach the preset minimum change threshold near the time when the warning signal is generated, the warning signal is confirmed by default and an external alarm is triggered. In this experiment, the minimum change threshold is set to a total harmonic distortion rate less than In the test simulating cooking fumes triggering a smoke alarm, the system of the present invention successfully identified the harmonic characteristics of the induction cooker or exhaust fan as matching the baseline library in multiple tests, and judged the warning signal as a benign event and suppressed it, which shows that the timing mutual verification mechanism of the present invention can effectively distinguish benign interference from real fire. In the test simulating an electrical circuit short circuit causing local overheating and triggering a temperature sensor, the system of the present invention detected electrical fault-related harmonic characteristics that did not match the baseline library in all tests, which was manifested as follows: to The broadband random noise energy within the frequency band increased significantly, so the system immediately confirmed the warning signal and triggered the highest priority external alarm, which verified the system's rapid response and high accuracy in identifying real electrical fires. All suppressed warning events recorded by the processor generated an event log, which included the warning type, occurrence time, alarm sensor location information and the corresponding benign harmonic matching results.

[0059] This experiment aims to verify the ability of the present invention to achieve passive positioning through multi-branch harmonic intensity differences, thereby avoiding the risk of false suppression caused by spatial correlation ambiguity in large buildings. Circumstantial evidence confirmation modules A, B, C, and D were deployed on different power branches of a simulated building. Module A was deployed on the power branch of a simulated kitchen area, module B was deployed on the power branch of an adjacent office area, and modules C and D were deployed in the power branches of a distant equipment room and public area. The experimental scenario was set as follows: a high-power induction heating device was deliberately started in the simulated kitchen area to simulate a benign electrical device that generates a strong harmonic signal. At the same time, a small number of smoke bombs were used in the adjacent office area to simulate a smoke sensor triggering an alarm, but there was no actual fire in the office area at this time. When the smoke sensor in the office area triggered an internal alarm signal, all four circumstantial evidence confirmation modules synchronously captured harmonic data on their power branches. The processor received the signal strength of the harmonic data captured by at least two circumstantial evidence confirmation modules and matched with the baseline feature library. The signal strength was the sum of the energy of specific harmonic components, such as arrive The root mean square value of the subharmonics is used to characterize and normalize them for comparison. The processor determines the approximate location of the harmonic source based on the signal strength received by each module: the location of the circumstantial confirmation module with the largest signal strength is determined to be the location closest to the harmonic source. In this experiment, when the approximate location of the harmonic source is not adjacent to the location of the alarm sensor that generates the internal warning signal, the processor suppresses the warning signal and converts it into an external alarm. After receiving the warning signal from the smoke sensor in the office area, the processor simultaneously analyzes the harmonic data captured by the circumstantial confirmation modules A, B, C, and D. If the harmonic signal strength reported by module A is the highest and the harmonic characteristics are consistent with the harmonics of the induction heating equipment in the simulated kitchen area, the processor will determine the harmonic source. The processor determines that the harmonic source (from the kitchen) and the alarm sensor (located in the office area) are not adjacent to each other, and even if the smoke sensor triggers an alarm, the processor will suppress the external alarm, thereby avoiding false suppression due to coincidence. The experimental results show that the present invention can achieve passive positioning of the harmonic source through the physical propagation attenuation characteristics of power grid harmonics without the need for additional positioning hardware. This capability enables the system to effectively distinguish between events that overlap in time but are unrelated in space, thereby avoiding the risk of false suppression due to misjudgment of spatial correlation in large buildings.

[0060] This experiment aims to verify the ability of the present invention to predict the health status of electrical equipment by monitoring the long-term change trend of the harmonic information entropy value during equipment operation. In the equipment room area, an industrial fan is selected as the monitoring object. First, when the fan is in a healthy operating state, the harmonic frequency domain characteristics of the fan during stable operation are collected multiple times through the circumstantial confirmation module. The processor calculates and stores the health information entropy baseline corresponding to these harmonic frequency domain characteristics. The information entropy calculation method is based on the Shannon information entropy formula commonly used in this field, in which the amplitude of each harmonic component of the frequency domain feature vector can be used as a representation of the probability distribution. The core technical consideration for determining the health information entropy baseline is to accurately capture the harmonic energy distribution law of the equipment in normal operating mode, and provide a reliable reference starting point for subsequent abnormal trend judgment. If If the baseline calculation is too rough, it will lead to the unreliability of subsequent deviation judgment and affect the accuracy of pre-diagnosis. Therefore, the establishment of the baseline requires multiple rounds of high-precision, high-frequency data collection when the equipment is running stably and in a known good state, and a representative and robust baseline model is constructed through statistical methods; in the daily operation of the simulated building, the circumstantial confirmation module continuously identifies the operating status of the industrial fan. Whenever the fan is running, the processor calculates the current information entropy value corresponding to its real-time harmonic frequency domain characteristics. The processor generates pre-diagnostic information about the health status of the industrial fan based on the long-term change trend of the current information entropy value relative to the healthy information entropy baseline. The continuous growth of the information entropy value is defined as an indication of the equipment's degradation trend. For example, in this experiment, the current information entropy value is continuously The average value of the day exceeds the health information entropy baseline , and the growth trend continues to exceed Every day, the system generates a pre-diagnosis work order. To simulate equipment degradation, slight wear of the fan bearings is gradually introduced into the test, and the degree of wear is gradually increased over a period of time to simulate the process of equipment from health to early degradation and then to maintenance. In the industrial fan degradation simulation test, the harmonic information entropy value of the fan fluctuates within the reasonable range of the healthy information entropy baseline at the initial stage of operation. As the degree of bearing wear increases, the harmonic spectrum captured by the circumstantial confirmation module gradually becomes disordered, which is mainly manifested in an increase in the energy proportion of high-order harmonic components and a more diffuse energy distribution, causing the real-time information entropy value to begin to show a continuous growth trend. Based on the preset pre-diagnosis rules, the processor successfully generates a pre-diagnosis work order for the degradation trend of the industrial fan, prompting the operation and maintenance personnel to conduct an inspection. After checking according to the work order, the operation and maintenance personnel found that the fan bearings had shown initial signs of wear and carried out maintenance in a timely manner. The test results confirm that the present invention can effectively reuse the harmonic analysis link, naturally extending the security system from the traditional passive alarm function to the active early warning capability of electrical equipment degradation. By continuously monitoring the entropy change trend of harmonic information, the system can provide operation and maintenance personnel with forward-looking equipment health pre-diagnosis information, thereby realizing a complete safety closed loop from early warning to prevention, and reducing the risk of electrical fires caused by equipment aging.

[0061] Example 4: This example combines Figures 1 to 6 , this paper describes the implementation of a building monitoring and management system based on the Internet of Things. Figure 1 As shown in the figure, the horizontal axis is frequency (unit: Hz), ranging from 50Hz to 950Hz; the vertical axis represents the amplitude value of the harmonic signal, in percentage. The two curves in the figure represent the harmonic signals of the induction cooker and the electric spark respectively. The solid line curve represents the harmonic signal of the induction cooker (benign equipment), and its signal amplitude reaches the highest value at 50Hz, and the amplitude decreases rapidly with the increase of frequency; the dotted line curve represents the harmonic signal of the electric spark (faulty equipment), and its amplitude changes relatively slowly. When the frequency is high, the signal changes less. Through the diagram, we can intuitively observe the difference in harmonic signals between the electric spark and the induction cooker at each frequency point.

[0062] like Figure 2As shown in the figure, first, the power grid (220V / 380V, 50Hz power frequency) provides the current signal of each harmonic, which is non-invasively collected through the current transformer (CT). The current signal is then transmitted to the analog-to-digital converter (ADC) and sampled and converted with a sampling rate of 10kHz. Then, the signal is processed by the microcontroller. The FFT module in the microcontroller performs fast Fourier transform (FFT) on the current signal to extract the characteristics of the 50Hz fundamental frequency and the 63rd harmonic. The figure also shows the role of the timing control module. The early warning signal is collected and processed through a 500ms acquisition window, and finally the frequency domain characteristics are output. In the figure, the frequency domain signals with frequencies of 50Hz, 150Hz and 250Hz are extracted and presented in the form of a bar graph, showing the amplitude distribution at different frequencies, reflecting the harmonic characteristics of the power grid.

[0063] like Figure 3 As shown, first, the temperature sensor detects abnormally high temperature and generates a warning signal. The signal is transmitted to the processor through the cross-domain transmission module. After receiving the signal, the processor triggers the signal acquisition process to collect grid harmonic data with a signal range of 1kHz to 10kHz. Then, the collected signal is subjected to fast Fourier transform through FFT transformation to extract its frequency domain features. Then, the system queries the matching feature library and compares the current signal with the data in the existing benign equipment feature library. If there is no match, the electrical fault feature is further identified and an alarm is issued. If a match is detected, the system will pass the query result to the alarm system. If there is no match, no alarm signal will be generated. Finally, the system will send an emergency alarm notification to the operation and maintenance personnel based on the identified electrical fault features.

[0064] like Figure 4 As shown in the figure, the harmonic amplitude comparison diagram of the induction cooker (benign equipment) and the electrical spark (fault) at each frequency point (including the baseline threshold reference) contains three sets of annotated data curves, namely the solid circle curve corresponding to the induction cooker (benign equipment), the dotted triangle curve corresponding to the electrical spark (fault) and the dotted square curve corresponding to the baseline threshold. The harmonic amplitude of the induction cooker (benign equipment) is the highest at 50 Hz, reaching 100%, and then shows a rapid downward trend with increasing frequency, dropping to a low amplitude at 950 Hz, reflecting the characteristic of benign equipment that the harmonic energy is concentrated at low frequencies; the harmonic amplitude of the electrical spark (fault) changes relatively slowly in each frequency band. Although it is also close to 100% at 50 Hz, the decrease is smaller as the frequency increases, reflecting its characteristics of wider harmonic energy distribution and obvious high-frequency disturbance; in addition, the baseline threshold shown in the figure represents the reference judgment threshold set by the system, represented by a square symbol and a dotted line, and maintained at an approximately constant horizontal line. It is used to distinguish between electrical sparks (faults) and benign equipment operating conditions in actual judgment, and serves as an important basis for frequency domain feature matching and event discrimination in the alarm system.

[0065] Figure 5 This performance comparison chart (including percentages of false alarm rate and detection accuracy) shows the performance of different methods. The horizontal axis is labeled with the traditional threshold method, multi-sensor fusion, and the proposed method, representing three different building alarm strategies, respectively. The vertical axis is a percentage, quantifying two performance indicators: false alarm rate and detection accuracy. The chart uses light gray bars to represent false alarm rate, and dark gray bars to represent detection accuracy, clearly comparing the differences between the three methods in terms of false alarm control and accuracy. The traditional threshold method has a false alarm rate of approximately 35% and a detection accuracy of approximately 75%. The multi-sensor fusion solution achieves a detection accuracy of approximately 85% while controlling the false alarm rate to approximately 20%. After introducing a mutual verification mechanism for grid harmonic timing, the proposed method significantly reduces the false alarm rate to less than 10% and increases the detection accuracy to approximately 95%, significantly outperforming both the traditional threshold method and the multi-sensor fusion method. This chart fully demonstrates the effectiveness and engineering value of the proposed IoT-based building monitoring and management system in reducing false alarm rates and improving detection accuracy.

[0066] Figure 6 The figure shows the signal strength distribution of harmonic sources at different locations. The vertical axis represents signal strength, and the horizontal axis represents four distribution areas, labeled Module A (kitchen), Module B (office area), Module C (computer room), and Module D (public area). The signal strength curve for each module records response data under three scenarios: when the kitchen equipment is operating, when the office equipment is operating, and when the computer room equipment is operating. It can be seen that in Module A (kitchen), the signal strength is highest, approaching 90%, when the kitchen equipment is operating. In Module C (computer room), the signal strength peaks at over 80% when the computer room equipment is operating. Furthermore, the signal strength in Module B (office area) when the office equipment is operating is significantly higher than at other locations. This figure demonstrates the principle of the present invention, which uses multiple circumstantial evidence to confirm the distribution of modules at different power branches in a building and infers the location of harmonic sources by monitoring signal strength changes. Specifically, the system uses the attenuation characteristics of harmonics during spatial propagation to determine the proximity of the harmonic source to the alarm sensor location, thereby determining whether the warning signal was triggered by the equipment in the target area, further reducing the risk of false suppression due to misjudgment of spatial location.

[0067] Example 5: In a typical commercial building, the building management system is tasked with real-time monitoring and control of the operating status of electrical equipment, air conditioning systems, and security facilities. To ensure efficient and stable operation of the equipment, the system should be able to promptly issue an alarm signal when an electrical equipment failure occurs. To avoid false alarms, the system combines historical data and real-time data of the electrical equipment, and accurately determines the equipment status through multimodal sensors and intelligent algorithms. The building monitoring and management system of this embodiment includes the following key modules: a multimodal sensor module, a data processing module, an alarm determination module, and a circumstantial evidence confirmation module. The multimodal sensor module includes current sensors, temperature sensors, vibration sensors, etc., which are used to collect the operating status of electrical equipment in real time. Each sensor transmits the collected data to the central processing unit via the Internet of Things protocol. Each sensor collects data once per second to ensure that the system can quickly respond to environmental changes. The data processing module pre-processes the sensor data, removes noise and performs normalization. Subsequently, the data is cleaned through an adaptive filtering algorithm to ensure the accuracy of the signal. Through this pre-processing, the system can reduce the impact of external interference on the data and improve the accuracy of equipment monitoring. The alarm judgment module determines whether the electrical equipment has failed based on the processed data. The system analyzes the data through the set threshold judgment rules. For example, when the rate of change of the current exceeds the set value, it is determined that the equipment is abnormal. The module has adaptive capabilities and can dynamically adjust the threshold according to the operating characteristics of the equipment. The circumstantial evidence confirmation module further confirms the accuracy of the alarm signal by comparing the historical operating data of the electrical equipment. In this way, the system can reduce the risk of false alarms. For example, if a similar fault mode has appeared in the historical data, the system will automatically increase the priority of the fault judgment to ensure the accuracy of the alarm signal.

[0068] The multimodal sensor module transmits real-time data to the central processing unit via a wireless sensor network (WSN). The data transmission uses the MQTT protocol to ensure low latency and high reliability. Each sensor collects data once per second, performs preliminary preprocessing, and then transmits it to the data processing module via the IoT network. The data processing module first preprocesses the received sensor data, including filtering, normalization, and noise removal. The data then enters the fault detection algorithm, which analyzes characteristics such as current changes, temperature changes, and vibration frequency. Based on predefined rules, it determines whether the device is abnormal. For example, if the current change rate exceeds a set threshold, the device is considered to be in a fault state. The alarm determination module determines whether the device is faulty based on the comparison results of real-time data with historical data. Using the Euclidean distance algorithm, the system matches the current device state with historical fault patterns. If the match exceeds the threshold, an alarm signal is triggered. The circumstantial evidence confirmation module ensures the accuracy of the alarm signal by reviewing historical data to avoid false alarms.

[0069] In actual engineering applications, the selection of parameters is crucial to the performance of the system. In this embodiment, the setting basis of parameters such as alarm thresholds and fault judgment standards is as follows: the temperature threshold is set to 45°C. If the operating temperature of the equipment exceeds this value, the system will trigger the alarm process. This value is set according to the maximum load temperature of the equipment and the requirements of its normal working environment to ensure timely response when the equipment temperature is too high to avoid equipment damage. The fault judgment basis of electrical equipment mainly includes the current change rate and vibration frequency. If the current change rate exceeds the preset threshold, the equipment is judged to be faulty. If the vibration frequency exceeds the normal range, the equipment is also judged to be faulty. The setting of this standard is based on the historical failure mode and manufacturing of the equipment. According to the technical manual provided by the manufacturer, the circumstantial evidence confirmation module compares the historical data of the equipment with the current real-time data. When the system detects a similar fault mode, it will automatically determine that the equipment is faulty and trigger an alarm. This mechanism effectively reduces the false alarm rate and ensures that the system's alarm function is more accurate. In the building monitoring system, the modules work together through information flow and control signals. After data is collected from the multimodal sensor module, it is sent to the central processing unit via wireless transmission. After the data is processed, the alarm judgment module makes a fault judgment and further verifies the accuracy of the alarm signal through the circumstantial evidence confirmation module. The entire system ensures that equipment failures can be detected in a timely and accurate manner through intelligent collaboration.

[0070] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A building monitoring and management system based on the Internet of Things, characterized in that: The system comprises: an alarm sensor configured to generate an internal warning signal when a set warning condition is detected, the warning signal including location information; The circumstantial evidence confirmation module is coupled to the building's power grid and is configured to: in response to an internal warning signal, capture time-domain harmonic data of the power grid within a predetermined time window before and after the warning signal is generated; perform a fast Fourier transform on the time-domain harmonic data to extract frequency-domain features of the harmonic data, which represent the activity of electrical devices within the power grid; The processor is electrically connected to the alarm sensor and the circumstantial evidence confirmation module, and is configured to: store a pre-calibrated baseline library of harmonic characteristics of benign equipment operation, the baseline library being generated by actively operating high-power electrical equipment in a building in a non-alarm state and recording the harmonic frequency domain characteristics of the equipment during stable operation; pattern-match the harmonic frequency domain characteristics extracted in real time by the circumstantial evidence confirmation module with the baseline library of harmonic characteristics of benign equipment operation to generate a matching result; and perform situational judgment on the internal warning signal based on the matching result: if the matching degree between the real-time harmonic frequency domain characteristics and any benign harmonic characteristics in the baseline library reaches a preset threshold, the warning signal is judged to be a benign event, and the conversion of the warning signal into an external alarm is suppressed; if the matching degree between the real-time harmonic frequency domain characteristics and any benign harmonic characteristics in the baseline library does not reach the preset threshold, and the power grid exhibits characteristics related to electrical faults, the warning signal is confirmed and an external alarm is triggered; if the power grid harmonic change does not reach a preset minimum change threshold near the time when the warning signal is generated, the warning signal is confirmed by default and an external alarm is triggered; Pattern matching is achieved by calculating the Euclidean distance between the real-time harmonic frequency domain features and the benign harmonic features in the baseline library. The matching degree reaches a preset threshold when the Euclidean distance is less than or equal to a preset distance threshold: ,in, Represents the real-time harmonic frequency domain feature vector Compared with the benign harmonic feature vector in the baseline library The Euclidean distance between Indicates the preset distance threshold; The circumstantial evidence confirmation modules include at least two, which are deployed in different power branches of the building respectively; the processor is configured to: receive the signal strength of the harmonic data captured by the at least two circumstantial evidence confirmation modules and matched with the baseline feature library; determine the approximate location of the harmonic source based on the signal strength, wherein the location of the circumstantial evidence confirmation module with the largest signal strength is determined to be closest to the harmonic source; only when the approximate location of the harmonic source is adjacent to the location of the alarm sensor that generates the internal warning signal, the processor suppresses the warning signal from being converted into an external alarm.

2. The building monitoring and management system based on the Internet of Things according to claim 1, characterized in that: The alarm sensor is any one of a smoke sensor, a temperature sensor, a gas sensor, and a flame sensor, or a combination thereof; the circumstantial evidence confirmation module includes a current transformer, an analog-to-digital converter, and a microcontroller equipped with a fast Fourier transform module, and the microcontroller is used to process time domain harmonic data.

3. The building monitoring and management system based on the Internet of Things according to claim 1, characterized in that: Calibration of the baseline library of harmonic characteristics of benign equipment operation includes: sequentially starting one or more high-power electrical equipment in the building, including kitchen equipment, cleaning machines, large air conditioners, and elevators; collecting the corresponding power grid time-domain harmonic data when the equipment is started, in stable operation, and stopped; performing a fast Fourier transform on the collected time-domain harmonic data to extract energy distribution characteristics and form the harmonic characteristics of benign equipment operation.

4. The building monitoring and management system based on the Internet of Things according to claim 1, characterized in that: The processor is also configured to: when the real-time harmonic frequency domain characteristics do not match any benign harmonic characteristics in the baseline library to a preset threshold, and when the power grid exhibits a preset broadband random noise characteristic generated by electrical sparks, the processor immediately confirms the early warning signal and triggers the highest priority external alarm.

5. The building monitoring and management system based on the Internet of Things according to claim 3 is characterized in that: The processor is also configured to: when calibrating a certain device, calculate and store the health information entropy baseline corresponding to the harmonic frequency domain characteristics of the device in a healthy operating state; in the daily operation of the building, whenever the circumstantial confirmation module identifies that a certain device is operating, the processor calculates the current information entropy value corresponding to its real-time harmonic frequency domain characteristics; the processor generates pre-diagnostic information about the health status of a certain device based on the long-term change trend of the current information entropy value relative to the health information entropy baseline, wherein a continuous increase in the information entropy value indicates that the device has a degradation trend.

6. The building monitoring and management system based on the Internet of Things according to claim 1, characterized in that: The processor is configured to record all suppressed warning events, and the event log includes the warning type, occurrence time, alarm sensor location information and corresponding benign harmonic matching results.

Images