Data processing method and system of intelligent fire-fighting internet-of-things integrated management platform

Abstract

The invention relates to the technical field of intelligent fire-fighting Internet of Things, and discloses a data processing method and system of an intelligent fire-fighting Internet of Things integrated management platform, and the method comprises the steps: obtaining and recording multiple operation performance data of each sensor in a stable operation state; when the data reported by the sensor is received and the data is abnormal in the analysis process, identifying and marking the data as an abnormal event; carrying out comprehensive analysis on the abnormal event; according to a comprehensive analysis result, the actual operation state of the sensor is judged, and the data processing method aims at solving the problem that in a complex industrial environment, an intelligent fire-fighting internet-of-things comprehensive management platform causes sensor data abnormity due to electromagnetic interference, and then misjudgment is caused. According to the method, through a refined data analysis and judgment mechanism, it is ensured that the platform can accurately recognize the real operation state of the sensor, and fire early warning is prevented from being delayed due to misjudgment.

Description

Technical Field

[0001] This invention relates to the technical field of smart fire protection Internet of Things (IoT), and in particular to a data processing method and system for a smart fire protection IoT integrated management platform. Background Technology

[0002] In modern industrial production environments, intelligent fire protection IoT integrated management platforms are widely deployed to improve the timeliness of fire early warning and the efficiency of safety hazard investigation. This platform is compatible with and manages various fire protection sensors. Newer fire protection sensors, while adhering to general communication rules, typically include vendor-specific data segments containing length declarations and specific content to provide more refined monitoring and diagnostic capabilities. The platform's data access module must be able to identify and process these vendor-specific data segments to fully utilize the additional functionality of new sensors and maintain good compatibility with future new equipment.

[0003] The parser in the platform, responsible for parsing data from various devices, is designed to process the standard data portion, then read the "length" field at the beginning of the manufacturer-specific data segment. Based on this length value, it skips a corresponding number of bytes, thus achieving compatibility with custom protocols of different manufacturers and lengths. This "trust input length" design is considered efficient and flexible in most cases because it allows the platform to correctly process the remaining parts of the data packet even without fully understanding or parsing all manufacturer-specific data content. This avoids writing a complete set of proprietary protocol parsing logic for each new sensor model, significantly reducing integration costs and development cycles.

[0004] However, in certain industrial areas, such as near the control rooms of large steel smelters, equipment rooms of high-voltage power transmission and transformation stations, or the core areas of heavy machinery manufacturing workshops, the operation of nearby large production equipment generates strong local electromagnetic interference. These interference sources can release powerful electromagnetic pulses instantaneously, potentially causing momentary errors in the data packets on the communication links of fire sensors deployed in these areas when they are reporting data. Specifically, the "length" field in the data packet, used to declare the length of manufacturer-specific data segments, may experience bit flipping, causing a length value that originally represented "10 bytes" to be incorrectly interpreted as "1024 bytes" or even a larger abnormal value. This physical-level data corruption is due to the impact of accidental external environmental factors on the integrity of digital signal transmission, rather than an internal logical error of the sensor. Summary of the Invention

[0005] This invention provides a data processing method and system for a smart fire protection IoT integrated management platform, aiming to solve the technical problem that in a strong electromagnetic interference environment, abnormal data parsing in the smart fire protection IoT integrated management platform leads to misjudgment of sensor status, which in turn affects the accuracy and timeliness of fire early warning.

[0006] Firstly, this application discloses a data processing method for a smart fire protection IoT integrated management platform, including: Acquire and record multiple performance data for each sensor under stable operating conditions; When data reported by the sensor is received and an anomaly occurs during the data parsing process, the data is identified and marked as an abnormal event. Conduct a comprehensive analysis of abnormal events; Based on the comprehensive analysis results, the actual operating status of the sensor is determined; The comprehensive analysis includes: The current abnormal data is compared with the description of the normal behavior pattern of the sensor to determine whether the current abnormal data conforms to the description of the normal behavior pattern. Check whether the time of the abnormal event is related to the operating time of production equipment known to generate strong electromagnetic interference; and check whether other sensors located near the abnormal sensor report abnormalities during the same period. The judgment includes: if the abnormal situation is significantly inconsistent with the description of the normal behavior of the sensor and cannot be explained by correlation with strong electromagnetic interference events, then the sensor is judged to be actually faulty. If the abnormal situation does not match the description of the sensor's normal behavior, it is interpreted in relation to a strong electromagnetic interference event. If so, the abnormal situation is determined to be an apparent anomaly caused by external interference. If so, the fire data reported by the sensor is trusted and sent to the fire judgment process for processing.

[0007] Through this technical solution, this application can effectively distinguish between sensor anomalies caused by their own malfunctions and apparent anomalies caused by external strong electromagnetic interference, thus avoiding the impact on the accuracy of fire warnings due to misjudgment of sensor status. This solves the problem in the background technology where electromagnetic interference leads to data parsing errors, which in turn causes misjudgment of sensor status and downgrade of fire warning processing.

[0008] Furthermore, based on the comprehensive analysis results, the actual operating status of the sensor is determined, including: Event data from sensors, production control systems, and network monitoring modules is continuously collected and initially standardized with timestamps. Sensors include an additional internal processing status code and data acquisition timestamp in their manufacturer-specific data segments. Define event types including covert electromagnetic interference events, sensor data bit flipping events, sensor anomaly report events, official production equipment startup events, and network latency fluctuation events, and extract the event type's occurrence timestamp, related sensor identifiers, equipment identifiers, and other auxiliary information; Construct a directed acyclic graph as a causal relationship network, where nodes represent event types and edges represent causal relationships; Real-time inferences are made based on causal relationship networks to assess the true cause of anomalies.

[0009] Through this technical solution, this application can build a causal relationship network and make real-time inferences, thereby uncovering the true causes of abnormal events at a deeper level, thus improving the accuracy and reliability of judging the actual operating status of sensors.

[0010] Based on this, event data from sensors, production control systems, and network monitoring modules are continuously collected and undergo preliminary timestamp standardization. Specifically, sensors include an internal processing status code and data acquisition timestamp in their manufacturer-specific data segments, including: Acquire event data from different data sources and perform preliminary timestamp standardization processing to convert all timestamps into a unified reference time; For each data source, a dynamic time offset feature record is established. The dynamic time offset feature record continuously monitors the difference between the timestamp of the event reported by the data source and the platform's internal time base, and calculates and updates the time offset and fluctuation range in real time by combining the time synchronization mechanism and time drift characteristics of the data source. Based on the event source data in the event data, obtain the currently effective time offset and fluctuation range from the dynamic time offset feature record; Based on the time offset and fluctuation range, as well as the network transmission delay data at the time of the event in the event data, the preliminary standardized timestamps in the event data are finely adjusted, and the nonlinear characteristics of the time offset are taken into account in the adjustment process. A timestamp calibration confidence assessment mechanism is introduced. After each calibration, the calibration confidence is calculated based on the difference between the timestamps before and after calibration, the stability of the time offset characteristic records, and the real-time fluctuation of network transmission delay. If the calibration confidence level is lower than the preset threshold, the event timestamp in the event data is marked as low confidence.

[0011] Through this technical solution, this application significantly improves the accuracy and consistency of timestamps for multi-source heterogeneous event data by recording dynamic time offset features and fine-grained timestamp adjustment, and introduces a calibration confidence assessment mechanism to ensure the reliability of subsequent causal association analysis.

[0012] Furthermore, a dynamic time offset feature record is established for each data source. This record continuously monitors the difference between the timestamp of events reported by the data source and the platform's internal time base. Combining the data source's time synchronization mechanism and time drift characteristics, it calculates and updates the time offset and fluctuation range in real time, including: Continuously monitor the difference between the timestamps of events reported by the data source and the platform's internal time base; Detecting sudden and drastic changes in ambient temperature or pressure parameters; When a sudden and drastic change in ambient temperature or pressure is detected, a high-frequency data sampling mode is activated to intensively collect timestamp difference data from the sensor. Adjust the calculation window and update frequency of the time offset based on the rate of change of environmental parameters; A prediction function based on nonlinear regression is used, with the ambient temperature and pressure change rate as input, to predict the instantaneous frequency drift of the internal clock crystal oscillator of the sensor. The time offset is corrected by combining historical drift data; An adaptive weighting mechanism is introduced to assign higher weights to data with timestamp differences in order to update the time offset and its fluctuation range.

[0013] Through this technical solution, this application achieves more accurate and real-time dynamic updates of sensor time offset and its fluctuation range by combining changes in environmental parameters and nonlinear regression prediction, thereby improving the adaptability and accuracy of timestamp calibration.

[0014] In some preferred implementations, the difference between the timestamps of events reported by the data source and the platform's internal time base is continuously monitored, including: It continuously receives event data reported from the data source and extracts timestamps from the event data; The extracted timestamps are initially verified, including checking the integrity of the timestamp format and whether the timestamp values ​​are within a reasonable time range. For the timestamps that pass the initial verification, calculate the difference between the timestamps and the platform's internal time base to obtain timestamp difference data; Real-time volatility analysis is performed on timestamp difference data. The volatility analysis includes: calculating the instantaneous rate of change of multiple consecutive timestamp difference values ​​and comparing it with a preset normal volatility threshold. When the instantaneous rate of change exceeds the preset normal fluctuation threshold, the abnormal timestamp identification process is triggered.

[0015] Through this technical solution, this application can promptly detect and identify abnormal timestamps by performing preliminary verification and real-time fluctuation analysis on timestamps, providing a foundation for subsequent fine-tuning and calibration, and improving the robustness of data processing.

[0016] More specifically, a preliminary verification is performed on the extracted timestamps. This preliminary verification includes checking the integrity of the timestamp format and whether the timestamp values ​​are within a reasonable time range, including: The extracted timestamps are initially verified, including checking the integrity of the timestamp format and whether the timestamp values ​​are within a reasonable time range. After the initial verification is passed, the temperature and pressure parameters inside the sensor are obtained; Based on temperature and pressure parameters, as well as the physical characteristics of the clock crystal oscillator corresponding to the sensor model, the instantaneous frequency drift of the internal clock crystal oscillator of the sensor is calculated in real time. The accuracy of timestamp generation is evaluated based on the instantaneous frequency drift. If the generation precision of the timestamp is lower than the preset precision threshold, the timestamp is marked as a low-precision timestamp.

[0017] By combining the sensor's internal environmental parameters and the characteristics of the clock crystal oscillator, this application can more accurately assess the generation accuracy of timestamps, thereby identifying and marking low-precision timestamps and providing more refined quality control for subsequent data processing.

[0018] As a further improvement, real-time volatility analysis is performed on the timestamp difference data. This volatility analysis includes calculating the instantaneous rate of change of multiple consecutive timestamp difference values ​​and comparing it with a preset normal volatility threshold, including: It continuously receives event data from the new fire sensor and extracts the timestamp and the current power consumption mode information of the sensor from the event data; Based on the power consumption mode information, the typical drift characteristic parameters and recovery time parameters of the sensor clock crystal under the power consumption mode are obtained from the pre-stored correspondence between power consumption mode and clock drift characteristics. Based on typical drift characteristic parameters and recovery time parameters, as well as the historical switching frequency and duration of the sensor switching from one power consumption mode to another, the normal fluctuation threshold of timestamp difference values ​​is dynamically calculated and updated. Calculate the instantaneous rate of change of multiple consecutive timestamp differences; The instantaneous rate of change is compared with the dynamically updated normal fluctuation threshold.

[0019] By adopting this technical solution, this application dynamically adjusts the normal fluctuation threshold of timestamp difference values ​​by considering the power consumption mode and switching history of the sensor, making the fluctuation analysis closer to the actual operating state of the sensor and improving the accuracy of abnormal timestamp identification.

[0020] To enhance functionality, based on typical drift characteristic parameters and recovery time parameters, as well as the historical switching frequency and duration of the sensor switching from one power mode to another, the normal fluctuation threshold of timestamp differences is dynamically calculated and updated, including: It continuously receives event data reported by the sensor and extracts the timestamp, the current power consumption mode information of the sensor, and the vibration or shock intensity data inside the sensor from the event data. Based on the power consumption mode information of the sensor, the typical drift characteristic parameters and recovery time parameters of the sensor clock crystal oscillator under the power consumption mode are obtained from the pre-stored correspondence between power consumption mode and clock drift characteristics. Based on the vibration or impact intensity inside the sensor, typical mechanical stress drift parameters of the sensor clock crystal under vibration or impact intensity are obtained from the pre-stored correspondence between vibration and impact and clock drift characteristics. Based on typical drift characteristic parameters, recovery time parameters, typical mechanical stress drift parameters, and the historical switching frequency and duration of the sensor switching from one power consumption mode to another, the normal fluctuation threshold of the timestamp difference value is dynamically calculated and updated.

[0021] Through this technical solution, this application further considers the vibration or impact intensity inside the sensor when dynamically calculating the normal fluctuation threshold, making the threshold setting more comprehensive and accurate, thereby improving the ability to identify abnormal fluctuations in timestamps.

[0022] To improve the solution, the normal fluctuation threshold for timestamp differences is dynamically calculated and updated, including: Taking power consumption mode information, vibration or shock intensity data, real-time temperature data, and historical switching frequency and duration as inputs, the predicted fluctuation range of timestamp difference values ​​is output through a multivariable nonlinear mapping function. Based on the predicted fluctuation range, determine the normal fluctuation threshold for timestamp differences; The multivariable nonlinear mapping function is adaptively adjusted by continuously monitoring the actual time drift data of the sensor under different power consumption modes, vibration or shock intensities and temperature conditions.

[0023] This technical solution introduces a multivariate nonlinear mapping function and adaptively adjusts it in conjunction with various environmental and operational parameters, enabling the normal fluctuation threshold of timestamp differences to more intelligently and accurately reflect the actual operating conditions of the sensor, thereby further improving the sensitivity and accuracy of anomaly detection.

[0024] Secondly, this application also discloses a data processing system for a smart fire protection IoT integrated management platform, comprising: The input end is used to acquire and record multiple operational performance data of each sensor under stable operating conditions; when data reported by the sensor is received and an anomaly occurs during the data parsing process, the data is identified and marked as an abnormal event. The analysis unit is used for comprehensive analysis of abnormal events; based on the results of the comprehensive analysis, it determines the actual operating status of the sensors; the comprehensive analysis includes: The current abnormal data is compared with the description of the normal behavior pattern of the sensor to determine whether the current abnormal data conforms to the description of the normal behavior pattern. The processing unit is used to check whether the time of the abnormal event is related to the operating time of production equipment known to generate strong electromagnetic interference; and to check whether other sensors geographically close to the abnormal sensor report abnormalities within the same time period. The judgment includes: if the abnormal situation is significantly inconsistent with the description of the normal behavior of the sensor and cannot be explained by correlation with strong electromagnetic interference events, then the sensor is judged to be actually faulty. If the abnormal situation does not match the description of the sensor's normal behavior, it is interpreted in relation to a strong electromagnetic interference event. If so, the abnormal situation is determined to be an apparent anomaly caused by external interference. If so, the fire data reported by the sensor is trusted and sent to the fire judgment process for processing.

[0025] This application provides a system that can effectively distinguish between sensor malfunctions and external interference. Through the collaborative work of the input end, analysis end, and processing end, it realizes intelligent identification, comprehensive analysis, and accurate judgment of abnormal events, thereby avoiding the impact of misjudging sensor status on the accuracy of fire early warning and improving the overall reliability of the smart fire protection IoT integrated management platform. Beneficial effects

[0026] The data processing method of the intelligent fire protection IoT integrated management platform disclosed in this application provides a benchmark for subsequent anomaly judgment by acquiring and recording multiple operational performance data of sensors under stable operating conditions. When anomalies occur in the data reported by the sensors during the parsing process, they can be promptly identified and marked as abnormal events. Crucially, this application conducts a comprehensive analysis of abnormal events, not only comparing the current abnormal data with the description of the sensor's normal behavior to determine if it conforms to normal behavior patterns, but also further checking whether the time of the abnormal event is related to known strong electromagnetic interference events, and cross-validating whether other sensors in nearby geographical locations reported abnormalities during the same period. Based on these comprehensive analysis results, this application can accurately judge the actual operating status of the sensors: if the abnormal situation significantly deviates from the description of the sensor's normal behavior patterns and cannot be explained by strong electromagnetic interference events, it is determined to be an actual fault; if the abnormal situation deviates from the sensor's normal behavior patterns but can be explained by strong electromagnetic interference events, it is determined to be a superficial anomaly caused by external interference. For superficial anomalies caused by external interference, this application trusts the fire data reported by the sensors and sends it to the fire judgment process for processing.

[0027] Through the above technical solution, this application effectively solves the problem in the background technology where strong electromagnetic interference causes data bit flipping in the manufacturer-specific data segment length field of the data packet, leading to out-of-bounds reading of the data parser buffer, ultimately resulting in the sensor being incorrectly judged as "faulty" or "offline," and causing the real fire data to be downgraded or suppressed. This application, by introducing a multi-dimensional and comprehensive anomaly analysis mechanism, can accurately distinguish between sensor malfunctions and external environmental interference, avoiding misjudgments of normally operating sensors. Especially when identifying apparent anomalies caused by external interference, this application can trust and promptly process the fire data reported by the sensor, thereby ensuring that the intelligent fire protection IoT integrated management platform can still trigger fire warnings promptly and accurately in complex industrial environments, effectively improving the timeliness and reliability of fire safety response and avoiding serious consequences caused by misjudgments. Attached Figure Description

[0028] Figure 1 This is a flowchart of a data processing method for a smart fire protection IoT integrated management platform provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the data processing system structure of a smart fire protection IoT integrated management platform provided in an embodiment of the present invention. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] Reference Figure 1 , Figure 1 This is a flowchart of a data processing method for a smart fire protection IoT integrated management platform provided in an embodiment of the present invention, including the following steps: Step S11: Acquire and record multiple performance data of each sensor under stable operating conditions; Step S12: When data reported by the sensor is received, and an anomaly occurs in the data during the parsing process, the data is identified and marked as an abnormal event; Step S13: Perform a comprehensive analysis of the abnormal event; Step S14: Based on the comprehensive analysis results, determine the actual operating status of the sensor.

[0031] The data processing method proposed in this application aims to address the problem of misjudgments caused by electromagnetic interference leading to abnormal sensor data in complex industrial environments within a smart fire protection IoT integrated management platform. This method, through refined data analysis and judgment mechanisms, ensures that the platform can accurately identify the true operating status of sensors, avoiding delays in fire warnings due to misjudgments.

[0032] Specifically, the key terms involved in this method include: "Sensors" refers to various detection devices deployed in the fire protection Internet of Things, such as smoke sensors, temperature sensors, and flame sensors. They are responsible for collecting environmental data and reporting it to the platform.

[0033] "Operating performance data" refers to the statistical characteristics or behavioral patterns of various parameters (such as signal strength, battery voltage, internal temperature, data transmission frequency, etc.) of the sensor under normal operating conditions. This data is used to establish a "normal behavior pattern description" of the sensor.

[0034] “Abnormal events” refer to situations where the data reported by the sensor has format errors, data bit flipping, or verification failures during the platform's parsing process, indicating that the data may be interfered with or that there is a problem with the sensor.

[0035] "Normal Behavior Pattern Description": This refers to the set of features formed by modeling the operational performance data acquired by the sensor under stable operating conditions, which describes the normal working mode of the sensor. Examples include the statistical distribution of data packet length, the value range of specific fields, and the periodicity of data reporting.

[0036] "Strong electromagnetic interference" refers to the instantaneous high-intensity electromagnetic radiation generated by large production equipment (such as electric arc furnaces, high-voltage switches, high-power motors, etc.) during operation, which may affect the communication and data integrity of nearby electronic equipment.

[0037] "Fire situation assessment process": refers to the internal logic module of the platform used to comprehensively analyze various fire-related data (such as smoke concentration, temperature change rate, etc.) and ultimately decide whether to trigger a fire alarm.

[0038] The embodiments of this application provide a data processing method for a smart fire protection IoT integrated management platform, which can effectively distinguish between actual sensor malfunctions and apparent anomalies caused by external interference.

[0039] First, acquire and record multiple performance data points for each sensor under stable operating conditions. For example, this can be achieved by allowing the sensors to operate in an interference-free, ideal environment for a period of time during initial deployment or periodic maintenance, continuously collecting parameters such as data packet length, checksum, internal status code, battery voltage, and signal strength. This data can be stored in a database and used to construct a "normal behavior pattern description" for each sensor. As a preferred implementation, statistical methods, such as calculating the mean, standard deviation, and data distribution range, can be used to quantify these performance data, thereby forming a baseline model.

[0040] Secondly, when data reported by a sensor is received, and an anomaly occurs during the parsing process, the data is identified and marked as an anomalous event. Specifically, when the platform's data parser processes data packets reported by the sensor, if it finds that the data packet format does not conform to the preset protocol specifications (for example, the length field of a manufacturer-specific data segment indicates a value exceeding the actual boundary of the data packet), or if the checksum verification fails, the data packet will be immediately identified as anomalous. Therefore, the system will generate an anomalous event record for this anomalous data packet, which includes information such as the anomaly type, the time of occurrence, and the sensor identifier involved.

[0041] Next, a comprehensive analysis of the abnormal event is conducted. This comprehensive analysis includes the following two aspects: On one hand, the current abnormal data is compared with the description of the sensor's normal behavior to determine whether the current abnormal data conforms to the description of normal behavior. For example, if the length of the manufacturer-specific data segment in the abnormal data packet far exceeds the statistical upper limit of the length of the sensor's historical normal data packets, it can be determined that it significantly deviates from the description of normal behavior. This comparison can be achieved by calculating the deviation or distance between the abnormal data and the description of normal behavior.

[0042] On the other hand, the system checks whether the timing of the abnormal event is related to the operating time of production equipment known to generate strong electromagnetic interference; and it also checks whether other sensors geographically close to the abnormal sensor reported abnormalities within the same time period. For example, the platform can maintain an operating schedule for a strong electromagnetic interference source, and when an abnormal event occurs, it queries this schedule to determine if there is any overlap in time. Simultaneously, the system queries other sensors located in the same physical area as the abnormal sensor to see if they reported similar abnormalities within a short period before and after the abnormal event. If multiple geographically close sensors report abnormalities within the same time period, there is a high probability of regional external interference.

[0043] Finally, based on the comprehensive analysis results, the actual operating status of the sensor is determined. Specifically, the determination process includes the following two scenarios: If the abnormal situation is significantly inconsistent with the normal behavior pattern of the sensor and cannot be explained by correlation with the strong electromagnetic interference event, then the sensor is determined to be actually faulty. For example, if a sensor's battery voltage is consistently below the normal range and there is no external interference event to explain it, then it can be determined to be a battery fault.

[0044] If the abnormal situation does not conform to the normal behavior pattern of the sensor, but can be interpreted in relation to the strong electromagnetic interference event, then the abnormal situation is determined to be a superficial anomaly caused by external interference. In this case, even if data parsing is abnormal, the system will trust the fire data reported by the sensor and send the fire data into the fire judgment process for processing. For example, if a smoke sensor reports data showing excessive smoke concentration during a strong electromagnetic interference event, even if its data packet length field is abnormal, because there is a reasonable explanation for the external interference and the data is related to the fire, the system will still send it into the fire judgment process to avoid delaying fire warnings due to misjudgment.

[0045] The data processing method of the intelligent fire protection IoT integrated management platform proposed in this application effectively solves the problem of sensor data misjudgment caused by electromagnetic interference in existing technologies by introducing multi-dimensional comprehensive analysis. Traditional methods often rely solely on the success or failure of data parsing to determine the sensor status. When data packets are damaged due to external interference, normally functioning sensors are easily marked as faulty or offline, leading to the degradation or suppression of real fire data.

[0046] The core innovation of this application lies in its focus not only on anomalies in data parsing but also in its in-depth and comprehensive analysis of abnormal events. First, by acquiring and recording multiple operational performance data points of sensors under stable operating conditions, a "normal behavior pattern description" for each sensor is constructed. This allows the platform to perform baseline comparisons with current abnormal data and determine the degree of deviation from normal behavior. Second, this application introduces checks on the correlation between abnormal events and strong electromagnetic interference events, as well as cross-validation of anomalies observed in other geographically close sensors during the same time period. These steps together constitute a robust diagnostic mechanism capable of distinguishing between anomalies caused by sensor malfunctions and apparent anomalies caused by external environmental interference.

[0047] For example, in a traditional smart fire protection platform, when a new type of fire sensor's data packet experiences a bit flip in its manufacturer-specific data segment length field due to strong electromagnetic interference, causing the parser to read out of bounds and incorrectly associate it with "low battery" status information, the sensor will be directly judged as "faulty." This misjudgment will cause the system to downgrade or even reject subsequent reports of real fire conditions from the sensor, thus delaying fire warnings.

[0048] In contrast, the method of this application, upon receiving the anomalous data, first identifies and marks it as an anomalous event. Subsequently, a comprehensive analysis is performed: on the one hand, the anomalous data (e.g., an unusually long manufacturer-specific data segment) is compared with the sensor's normal behavior description, revealing a significant discrepancy; on the other hand, it checks whether the timing of the anomalous event is related to the operation time of strong electromagnetic interference from nearby large production equipment, and whether other nearby sensors also reported anomalies during the same period. If the comprehensive analysis indicates that the anomaly does not conform to the sensor's normal behavior description but can be explained by association with strong electromagnetic interference events, then this method determines that the anomaly is a superficial anomaly caused by external interference. In this case, even if data parsing is abnormal, the system will still trust the fire data reported by the sensor and send it to the fire assessment process for processing.

[0049] Therefore, the method of this application can avoid the delays in fire warnings caused by misjudgment in traditional methods, significantly improving the reliability and security of the smart fire protection IoT integrated management platform in complex industrial environments. Through this refined data processing and judgment mechanism, the platform can more accurately identify the true status of sensors, ensuring the timeliness and effectiveness of critical safety decisions.

[0050] This application further proposes the method for judging the actual operating status of the sensor based on the comprehensive analysis results, specifically including: Event data from sensors, production control systems, and network monitoring modules is continuously collected and initially standardized with timestamps. Sensors include an additional internal processing status code and data acquisition timestamp in their manufacturer-specific data segments. Define event types including covert electromagnetic interference events, sensor data bit flipping events, sensor anomaly report events, official production equipment startup events, and network latency fluctuation events, and extract the timestamps of the events, related sensor identifiers, device identifiers, and other auxiliary information. Construct a directed acyclic graph as a causal relationship network, where nodes in the causal relationship network represent the event types and edges in the causal relationship network represent causal relationships; Real-time inferences are made based on the causal relationship network to assess the true cause of the anomaly.

[0051] Specifically, continuously collecting event data from sensors, production control systems, and network monitoring modules refers to acquiring real-time or historical event information from various data sources integrated into the smart fire protection IoT platform. The data reported by sensors not only includes the environmental parameters they monitor but also contains an internal processing status code and a data acquisition timestamp within the manufacturer-specific data segment. The internal processing status code indicates the sensor's operational status, such as self-test results, power status, and internal error codes; the data acquisition timestamp precisely records the moment the data was acquired within the sensor, which is crucial for subsequent time synchronization and causal analysis. Preliminary timestamp standardization of this event data aims to convert timestamps from different data sources, potentially using different time formats or time zones, into a unified internal platform reference time for accurate event sorting and correlation.

[0052] The definition of event types aims to abstract complex raw event data into categories with clear semantics, facilitating the construction and analysis of causal relationship networks. For example, covert electromagnetic interference events can refer to electromagnetic fluctuations that are not directly monitored but may affect sensor performance; sensor data bit flipping events refer to random bit errors that occur during data transmission or processing; sensor anomaly report events are fault or warning messages issued by the sensor itself; official production equipment start-up events refer to the normal start-up or shutdown operations of production equipment related to fire safety (such as large motors, high-power heating equipment, etc.), which may be accompanied by electromagnetic or mechanical interference; and network latency fluctuation events reflect delay changes along the data transmission path. Extracting the event type's occurrence timestamp, related sensor identifiers, equipment identifiers, and other auxiliary information provides necessary context and correlation evidence for subsequent causal analysis.

[0053] In practical applications, a directed acyclic graph (DAG) is constructed as a causal network to visualize and quantify the causal relationships between different event types. In this network, each node represents a predefined event type, such as a "production equipment startup event" or a "covert electromagnetic interference event." Directed edges represent causal dependencies between events; for example, an edge pointing from a "production equipment startup event" to a "covert electromagnetic interference event" might indicate that the startup of the production equipment is a cause of electromagnetic interference. Since causal relationships are typically unidirectional and do not form cycles, using a DAG effectively avoids circular reasoning and ensures the clarity of the causal chain.

[0054] Furthermore, real-time inference based on the causal relationship network to assess the true cause of the anomaly refers to using the constructed causal relationship network model, combined with real-time collected event data, and employing inference algorithms (such as Bayesian network inference, path analysis, etc.) to identify the root cause of the sensor anomaly. For example, when a sensor reports an anomaly, the system can backtrack along the edges in the causal relationship network to find the sequence of events that may have led to the anomaly before it occurred, thereby distinguishing whether it is a sensor malfunction, external interference, or other systemic problems.

[0055] This application's solution significantly improves the ability to determine the true cause of sensor anomalies by introducing multi-source event data collection, refined event type definitions, and a causal relationship network based on directed acyclic graphs. Compared to the aforementioned basic solutions that mainly rely on direct comparison and simple correlation analysis between abnormal data and normal behavior patterns, this solution can more deeply explore the inherent logic and causal chains between events. Specifically, by continuously collecting event data from sensors, production control systems, and network monitoring modules, and paying particular attention to sensor internal processing status codes and data acquisition timestamps, the system obtains more comprehensive contextual information, which helps distinguish between internal sensor faults and external environmental influences. For example, if the sensor's internal processing status code shows a self-test anomaly, even if there is no significant interference from the external environment, it may indicate a problem with the sensor itself.

[0056] Furthermore, by defining various event types, including covert electromagnetic interference events, sensor data bit flipping events, and official production equipment startup events, this solution abstracts complex real-world events into analyzable discrete units, laying the foundation for constructing a causal model. Constructing a directed acyclic graph (DAG) as a causal network allows the system to clearly represent and manage the causal dependencies between different event types. For example, the startup of production equipment may cause strong electromagnetic interference, which in turn triggers a brief data bit flipping event in nearby sensors. This network structure effectively avoids circular reasoning and allows the system to trace or predict along causal chains. Finally, real-time inference based on the causal network enables the platform to dynamically assess the true cause of current anomalies, such as determining whether the apparent anomaly is due to external strong electromagnetic interference or an actual malfunction of internal sensor components. This avoids false alarms or missed alarms that may result from judging solely based on surface phenomena, improving the accuracy and reliability of anomaly diagnosis.

[0057] Through the above technical solution, this application can significantly improve the diagnostic accuracy and efficiency of the smart fire protection IoT integrated management platform for sensor anomalies. Compared to basic solutions that rely solely on preset rules and simple correlation analysis, this solution, by constructing a refined causal relationship network, can more accurately identify the underlying causes of anomalies and effectively distinguish between actual sensor malfunctions and apparent anomalies caused by external environmental interference. This not only reduces the risk of false alarms and missed alarms and improves the reliability of fire situation assessment, but also provides maintenance personnel with more insightful fault diagnosis information, enabling them to take countermeasures more quickly and accurately, ensuring the stable operation of the fire protection system and the safety of life and property.

[0058] This application further proposes the continuous collection of event data from sensors, production control systems, and network monitoring modules, followed by preliminary timestamp standardization processing. Specifically, the sensors include an internal processing status code and a data acquisition timestamp in their manufacturer-specific data segments, including: Acquire event data from different data sources and perform preliminary timestamp standardization processing to convert all timestamps into a unified reference time; A dynamic time offset feature record is established for each data source. The dynamic time offset feature record continuously monitors the difference between the timestamp of the event reported by the data source and the platform's internal time base, and calculates and updates the time offset and fluctuation range in real time by combining the time synchronization mechanism and time drift characteristics of the data source. Based on the event source data in the event data, obtain the currently effective time offset and fluctuation range from the dynamic time offset feature record; Based on the time offset and the fluctuation range, as well as the network transmission delay data at the time of the event in the event data, the preliminary standardized timestamp in the event data is finely adjusted, and the adjustment process takes into account the nonlinear characteristics of the time offset. A timestamp calibration confidence assessment mechanism is introduced. After each calibration, the calibration confidence is calculated based on the difference between the timestamps before and after calibration, the stability of the time offset characteristic records, and the real-time fluctuation of network transmission delay. If the calibration confidence level is lower than a preset threshold, the event timestamp in the event data is marked as low confidence.

[0059] Specifically, acquiring event data from different data sources and performing preliminary timestamp standardization involves the platform receiving event data from various data sources such as sensors, production control systems, and network monitoring modules. The platform first converts the timestamps contained in this data into a unified internal reference time base, such as Coordinated Universal Time (UTC) or a local synchronization time. The purpose is to provide a unified starting point for subsequent fine-tuning.

[0060] Specifically, establishing a dynamic time offset characteristic record for each data source can be understood as the system maintaining a dedicated record for each data source, such as a sensor, production control system, or network monitoring module. This record continuously monitors the difference between the timestamp of the event reported by the data source and the platform's internal time base. It also considers the data source's own time synchronization mechanism (e.g., NTP client configuration, internal crystal oscillator accuracy) and its inherent time drift characteristics (e.g., affected by temperature, voltage, etc.) to calculate and update the current time offset (i.e., the average deviation from the platform's reference time) and its fluctuation range in real time. The purpose is to capture and quantify the unique clock behavior of each data source, providing a basis for subsequent precise calibration.

[0061] In practical applications, obtaining the currently effective time offset and fluctuation range from the dynamic time offset feature record based on the event source data in the event data means that when an event data is received, the system identifies which specific data source (e.g., a specific sensor 100) reported the event data, and then queries the pre-established dynamic time offset feature record to extract the latest calculated and updated time offset and fluctuation range from that data source. The purpose is to provide personalized calibration parameters for the timestamp calibration of the current event data.

[0062] Furthermore, based on the time offset and the fluctuation range, as well as the network transmission delay data at the time of the event in the event data, the preliminary standardized timestamp in the event data is finely adjusted. This adjustment process considers the nonlinear characteristics of the time offset, meaning that, based on the preliminary standardized timestamp, the system comprehensively considers the time offset and fluctuation range of the data source obtained from dynamic time offset feature records, as well as the actual network transmission delay experienced by the event data during transmission. The fine-tuning process employs more complex algorithms, such as nonlinear regression models or Kalman filtering, to correct the timestamp to more accurately reflect the true time of the event. This adjustment pays particular attention to potential nonlinear changes in the time offset; for example, clock drift is not constant but may vary with environmental factors or equipment operating status. Its purpose is to minimize the impact of various factors on timestamp accuracy and obtain a high-precision event occurrence time.

[0063] Furthermore, a timestamp calibration confidence assessment mechanism is introduced. After each calibration, the calibration confidence score is calculated based on the difference between the timestamps before and after calibration, the stability of the time offset characteristic records, and the real-time fluctuations in network transmission latency. This means that after each fine-tuning of the timestamps, the system immediately assesses the reliability of the calibration. The assessment is based on factors including the magnitude of the change in timestamp values ​​before and after calibration, whether the time offset characteristic records of the corresponding data source have been stable recently (e.g., whether the fluctuation range is too large), and whether the network latency during the transmission of the event data fluctuates abnormally. By comprehensively considering these factors, a quantitative confidence score is calculated. The purpose is to provide a reference for the reliability of timestamps in subsequent event analysis and avoid using low-quality timestamps for critical judgments.

[0064] Therefore, marking the event timestamp in the event data as low confidence if the calibration confidence score is lower than a preset threshold means that if the calculated calibration confidence score is lower than a pre-set threshold, the timestamp of the event data will be specifically marked as "low confidence". This implies that the accuracy of the timestamp may be highly questionable, and it should be used with caution or given a lower weight in subsequent causal correlation analysis or fault diagnosis. The purpose is to identify and isolate potentially inaccurate timestamps to prevent them from negatively impacting system decisions.

[0065] This application, through the aforementioned technical solution, significantly improves the accuracy and reliability of event timestamps in the intelligent fire protection IoT integrated management platform. Compared to basic solutions that only perform preliminary standardization processing, this solution greatly reduces timestamp errors by dynamically tracking and compensating for clock drift, network latency, and nonlinear time offsets from various data sources. This high-precision timestamp ensures that the order and time intervals of events are accurately reflected during comprehensive analysis of abnormal events, especially when performing causal network inference, thereby avoiding misjudgments caused by inaccurate timestamps, such as misjudging external interference as actual sensor malfunctions, or vice versa. Furthermore, the introduced calibration confidence assessment mechanism enables the system to identify and isolate potentially erroneous timestamps, further enhancing the robustness of data processing and ensuring the accuracy and reliability of the platform's judgment of the actual operating status of sensors, thereby improving the efficiency and accuracy of intelligent fire protection early warning and response.

[0066] In some preferred embodiments, it is assumed that a smart fire protection IoT integrated management platform is deployed in a large industrial park, which includes various types of sensors 100 (such as smoke sensors, temperature sensors, pressure sensors, etc.), a production control system 200, and a network monitoring module 300. Due to differences in hardware, deployment environment, and network conditions, these data sources have varying degrees of offset and drift between their internal clocks and the platform's internal time base.

[0067] Specifically, when the platform receives fire event data from a smoke sensor 100, the event data includes a pre-normalized timestamp. The platform first identifies that the event data originates from that specific smoke sensor 100. Next, the system queries the dynamic time offset characteristic record established for that smoke sensor 100. This record may show that the sensor 100's internal clock is typically 50 milliseconds behind the platform time, and its drift fluctuation range is within ±10 milliseconds. This record is updated in real-time based on the sensor 100's historical data and environmental parameters (such as temperature). Simultaneously, the system also obtains real-time network transmission latency data during the transmission of the fire event data, for example, 20 milliseconds.

[0068] Based on this information, the system uses a nonlinear regression model to fine-tune the initial standardized timestamp. For example, if the initial timestamp is T_raw, the system calculates a more accurate calibration timestamp T_calibrated based on the dynamic time offset (-50ms), fluctuation range (±10ms), and network latency (20ms), taking into account the clock crystal drift characteristics of the sensor 100 at the current temperature. For example, T_calibrated = T_raw - (-50ms) - 20ms + f(temperature, drift_model), where f is a function that considers nonlinear drift.

[0069] After calibration, the system calculates the confidence level of the calibration. If the timestamps before and after calibration differ significantly, or if the time offset characteristic record of the sensor 100 shows recent drastic fluctuations, or if there are abnormal peaks in network transmission latency, the calculated confidence level may be low. For example, if the confidence level is below a preset threshold of 0.8, the timestamp of the fire event data will be marked as "low confidence." This means that in subsequent fire assessment processes, the timestamp information of this event will be treated with caution, and may need to be combined with other high-confidence data or manually verified to avoid false alarms or missed alarms due to inaccurate timestamps. In this way, the platform can ensure that the timestamps of key events used for decision-making have the highest accuracy and reliability.

[0070] In some embodiments described above, a method for establishing dynamic time offset characteristic records for each data source is proposed. This method continuously monitors the difference between the timestamp of events reported by the data source and the platform's internal time base, and combines the data source's time synchronization mechanism and time drift characteristics to calculate and update the time offset and fluctuation range in real time. However, in actual industrial environments, the ambient temperature or pressure parameters of the sensor may change drastically and instantaneously. These environmental fluctuations significantly affect the frequency stability of the sensor's internal clock crystal, causing nonlinear changes in its time drift characteristics. Consequently, the time offset and fluctuation range calculated based solely on conventional monitoring and fixed models may not accurately reflect the sensor's true time behavior in dynamic environments, thus affecting the accuracy of subsequent fine-tuning of event data timestamps and calibration confidence assessment.

[0071] In response, this application further proposes establishing dynamic time offset feature records for each data source. These dynamic time offset feature records continuously monitor the difference between the timestamps of events reported by the data source and the platform's internal time base, and, in conjunction with the data source's time synchronization mechanism and time drift characteristics, calculate and update the time offset and fluctuation range in real time, including: Continuously monitor the difference between the timestamps of events reported by the data source and the platform's internal time base; Detecting sudden and drastic changes in the ambient temperature or pressure parameters; When a sudden and drastic change in the ambient temperature or pressure parameter is detected, a high-frequency data sampling mode is activated to intensively collect the timestamp difference data of the sensor; The calculation window and update frequency of the time offset are adjusted based on the rate of change of the environmental parameters. Using a prediction function based on nonlinear regression, the ambient temperature and pressure change rates are taken as inputs to predict the instantaneous frequency drift of the internal clock crystal oscillator of the sensor. The time offset is corrected by combining historical drift data; An adaptive weighting mechanism is introduced to assign higher weights to the timestamp difference data in order to update the time offset and its fluctuation range.

[0072] Specifically, continuously monitoring the difference between the timestamps of events reported by data sources and the platform's internal time base means that the system continuously receives data from various sensors and extracts the timestamps of the events that occurred. These timestamps are then compared with a unified time base maintained internally by the platform to quantify the time offset of each sensor relative to the platform's base.

[0073] The detection of sudden and drastic changes in ambient temperature or pressure parameters refers to the system acquiring real-time ambient temperature and pressure data of the sensor's location through an environmental monitoring module integrated inside or outside the sensor. When these parameters experience a rapid rise or fall exceeding a preset threshold within a short period, it is identified as a sudden and drastic change. This change typically has a physical impact on the clock crystal oscillator inside the sensor, causing a change in its oscillation frequency.

[0074] In practical applications, when a sudden and drastic change in ambient temperature or pressure parameters is detected, a high-frequency data sampling mode is activated to intensively collect timestamp difference data from the sensors. This means the system temporarily increases the frequency of timestamp data acquisition from the affected sensors. For example, sampling may be adjusted from once per minute to multiple times per second to more precisely capture the dynamic process of time drift during environmental changes.

[0075] Furthermore, adjusting the calculation window and update frequency of the time offset in conjunction with the rate of change of environmental parameters means dynamically adjusting the size of the historical data window used to calculate the time offset and the frequency of time offset updates based on the rate of change of environmental temperature or pressure. For example, when the rate of environmental change is rapid, a smaller calculation window and a higher update frequency are used to ensure that the calculation of the time offset can respond promptly to environmental changes.

[0076] As a preferred implementation, a prediction function based on nonlinear regression is used, taking the rate of change of ambient temperature and pressure as input, to predict the instantaneous frequency drift of the sensor's internal clock crystal oscillator. This involves establishing a nonlinear relationship between the rate of change of environmental parameters and the frequency drift of the sensor's clock crystal oscillator using machine learning or statistical models. This function can predict the degree of instantaneous drift of the sensor clock based on real-time environmental change trends.

[0077] Among them, correcting the time offset by combining historical drift data refers to comprehensively analyzing the instantaneous frequency drift predicted by the nonlinear regression function with the historical drift data of the sensor under similar environmental conditions, calibrating and optimizing the prediction results to improve the accuracy of the correction.

[0078] Furthermore, an adaptive weighting mechanism is introduced to assign higher weight to timestamp difference data in order to update the time offset and its fluctuation range. This means that when calculating and updating the time offset and its fluctuation range, higher weight is given to timestamp difference data collected at high frequencies during periods of drastic environmental change. This implies that data collected at critical moments has a greater influence on the calculation results of the time offset, thereby enabling the updated time offset to more accurately reflect the sensor's true time behavior in the current environment.

[0079] Through the aforementioned technical solution, this application significantly improves the accuracy and real-time performance of sensor time offset calculation in complex industrial environments by the intelligent fire protection IoT integrated management platform. Especially when there are sudden and drastic changes in ambient temperature or pressure, this solution effectively addresses the nonlinear drift of the internal clock crystal oscillator of the sensor, avoiding false alarms or missed alarms caused by inaccurate time synchronization. Consequently, the platform's fine-tuning of sensor event data becomes more precise, thereby improving the reliability of anomaly identification, comprehensive analysis, and judgment of the actual operating status of sensors, ultimately enhancing the early warning capabilities and decision-making efficiency of the entire intelligent fire protection system.

[0080] In some preferred embodiments, suppose a smart fire protection IoT integrated management platform is deployed in a chemical plant area, which includes a large number of smoke sensors, temperature sensors, etc. During a certain production process, a large reactor suddenly starts up, causing the local ambient temperature in the surrounding area to rise rapidly from 25°C to 40°C in a short period of time, accompanied by slight pressure fluctuations.

[0081] A temperature sensor 100 reported a sudden and drastic change in the ambient temperature parameter. Once this change is detected, the system immediately initiates a high-frequency data sampling mode for the temperature sensor 100, for example, increasing the collection frequency of timestamp difference data from once per minute to five times per second to intensively collect timestamp difference data during the period of rapid temperature rise.

[0082] Simultaneously, the system dynamically adjusts the calculation window for the time offset by considering the rate of change of the current ambient temperature (e.g., an increase of 1°C per second), making it smaller and increasing the update frequency to respond quickly to such changes. Next, a pre-trained prediction function based on nonlinear regression, taking the received rate of change of ambient temperature (and possibly the rate of change of pressure) as input, predicts the instantaneous frequency drift of the internal clock crystal oscillator of the temperature sensor 100. For example, it predicts that its clock may be 0.5 milliseconds fast per hour.

[0083] Subsequently, the system combines the drift records of the temperature sensor 100 in historical data under similar temperature change conditions to correct the predicted instantaneous frequency drift, making it closer to reality. Finally, when updating the dynamic time offset feature records of the temperature sensor 100, the system introduces an adaptive weighting mechanism, assigning higher weight to timestamp difference data collected at high frequencies during periods of drastic temperature changes. This means that data from these critical moments have a greater decisive influence on the final calculated time offset and fluctuation range, thus ensuring that the time offset of the temperature sensor 100 can be accurately calculated and updated in real time even under conditions of drastic environmental fluctuations. In this way, even if the sensor's internal clock experiences nonlinear drift due to environmental changes, the reported event timestamps can be accurately calibrated, avoiding misjudgments caused by time synchronization errors.

[0084] This application further proposes the following steps for continuously monitoring the difference between the timestamps of events reported by the data source and the platform's internal time base: Continuously receive event data reported by the data source and extract timestamps from the event data; The extracted timestamps are preliminarily verified, including checking the integrity of the timestamp format and whether the timestamp value is within a reasonable time range. For the timestamp that has passed the initial verification, calculate the difference between the timestamp and the internal time base of the platform to obtain timestamp difference data; Real-time volatility analysis is performed on the timestamp difference data. The volatility analysis includes: calculating the instantaneous rate of change of multiple consecutive timestamp difference values ​​and comparing it with a preset normal volatility threshold. When the instantaneous rate of change exceeds the preset normal fluctuation threshold, the abnormal timestamp identification process is triggered.

[0085] Specifically, continuously receiving event data reported by the data sources means that the platform continuously receives data packets or event reports from various connected sensors, production control systems, and network monitoring modules. These data packets or event reports contain information collected by the sensors at specific times, as well as the timestamps of their generation or reporting. The platform needs to have efficient data receiving capabilities to handle concurrent data streams from a large number of data sources.

[0086] The preliminary verification of the extracted timestamps aims to ensure their basic validity and reliability. This preliminary verification includes two aspects: First, checking the format integrity of the timestamps. For example, verifying whether the timestamp conforms to the expected date-time string format (such as ISO 8601) or Unix timestamp format, ensuring its length, separators, and other structural elements are correct. Second, checking whether the timestamp value is within a reasonable time range. For example, determining whether the timestamp falls within a reasonable time window before or after the current system time (such as from the past 24 hours to the next hour) to exclude obviously erroneous or maliciously tampered timestamps.

[0087] In practical applications, for the timestamp that passes initial verification, the difference between the timestamp and the platform's internal time base is calculated to obtain timestamp difference data. This means that after the timestamp passes initial verification, it is compared with a high-precision time base maintained by the platform itself, thereby quantifying the deviation between the data source clock and the platform clock. This difference data is the basis for subsequent analysis and calibration.

[0088] Furthermore, real-time volatility analysis is performed on the timestamp difference data to detect abnormal fluctuations in the timestamp differences, which may indicate sensor clock drift, abnormal network latency, or external interference. The volatility analysis specifically includes calculating the instantaneous rate of change of multiple consecutive timestamp difference values, i.e., examining the speed and magnitude of change of the timestamp differences over a short period, and comparing it with a preset normal volatility threshold. This threshold can be preset or dynamically adjusted based on the sensor type, communication protocol, environmental conditions, and historical data.

[0089] Therefore, when the instantaneous rate of change exceeds the preset normal fluctuation threshold, the abnormal timestamp identification process is triggered. This means that once a fluctuation in timestamp differences is detected to exceed the normal range, the system will immediately initiate a dedicated process to further investigate and handle this anomaly, rather than simply treating it as normal data. This helps to promptly identify and isolate potential problematic sensors or network anomalies.

[0090] Through the aforementioned technical solution, this application significantly improves the accuracy and robustness of dynamic time offset feature recording. By introducing a preliminary verification mechanism, invalid or abnormal timestamp data is effectively eliminated, ensuring data quality from the source. The addition of real-time fluctuation analysis enables the system to detect abnormal fluctuations in timestamp differences more sensitively and promptly, thereby identifying problems such as sensor clock drift, abnormal network latency, or external interference earlier. This refined monitoring and anomaly triggering mechanism avoids misjudgments caused by data quality issues or minor anomalies, ensuring more accurate calculations of time offset and fluctuation range. This provides a more reliable foundation for subsequent refined timestamp adjustments and calibration confidence assessments, ultimately improving the accuracy of the smart fire protection IoT integrated management platform in judging the actual operating status of sensors.

[0091] In some preferred embodiments, assume a smoke sensor operates within an industrial park, and its reported event data includes a timestamp. The platform continuously receives the event data reported by the sensor. When an event data packet is received, the timestamp is first extracted, for example, "2023-10-27T10:30:05.123Z". Subsequently, the timestamp is preliminarily validated to check if its format conforms to the ISO 8601 standard and to determine if its value is within a reasonable range of the current system time (e.g., no earlier than 24 hours ago and no later than 1 hour later). If the validation passes, the difference between the timestamp and the platform's internal high-precision time base (e.g., the platform's current time is "2023-10-27T10:30:05.000Z") is calculated, yielding the timestamp difference data, for example, 123 milliseconds. The platform continuously collects this timestamp difference data and performs real-time volatility analysis on multiple consecutive difference values. For example, the instantaneous rate of change of the most recent 5 timestamp difference values ​​is calculated. If the instantaneous rate of change of a timestamp difference suddenly jumps from the normal ±5 ms / sec to ±50 ms / sec, exceeding the preset normal fluctuation threshold (e.g., ±10 ms / sec), the system will immediately trigger the abnormal timestamp identification process. At this point, the platform will not simply use this abnormally fluctuating timestamp difference to update the dynamic time offset feature record. Instead, it will initiate further diagnostics, such as checking whether the sensor is subject to strong electromagnetic interference, whether there are network connection problems, or whether the sensor's internal clock crystal oscillator has malfunctioned. This avoids misjudgments caused by abnormal data and ensures the accuracy of the time offset feature record.

[0092] In some embodiments described above in this application, the initial verification of the extracted timestamp only checks the integrity of the timestamp format and whether the value is within a reasonable time range. However, this initial verification may not effectively identify low-precision timestamps caused by environmental factors affecting the internal clock crystal of the sensor, thus affecting the accuracy of subsequent data processing.

[0093] In response, this application further proposes a preliminary verification of the extracted timestamp, which includes checking the format integrity of the timestamp and whether the timestamp value is within a reasonable time range, including: The extracted timestamps are preliminarily verified, including checking the integrity of the timestamp format and whether the timestamp value is within a reasonable time range. After the initial verification is passed, the temperature and pressure parameters inside the sensor are obtained; Based on the temperature and pressure parameters, as well as the physical characteristics of the clock crystal oscillator corresponding to the sensor model, the instantaneous frequency drift of the clock crystal oscillator inside the sensor is calculated in real time. The generation accuracy of the timestamp is evaluated based on the instantaneous frequency drift. If the generation accuracy of the timestamp is lower than a preset accuracy threshold, the timestamp is marked as a low-precision timestamp.

[0094] Specifically, after the initial verification is passed, the system further acquires the temperature and pressure parameters inside the sensor. These parameters are key environmental factors affecting the stability of the internal clock crystal oscillator. For example, temperature changes can cause the crystal oscillator frequency to drift, thus affecting the accuracy of the timestamp. The physical characteristic parameters of the clock crystal oscillator corresponding to the sensor model refer to the frequency response curve and aging characteristics of the crystal oscillator used in a specific sensor model under different temperature and pressure conditions. By combining these real-time acquired temperature and pressure parameters with the preset crystal oscillator physical characteristic parameters, the instantaneous frequency drift of the internal clock crystal oscillator of the sensor can be calculated in real time. This instantaneous frequency drift reflects the actual timekeeping deviation of the internal clock of the sensor under the current environmental conditions. Subsequently, based on the calculated instantaneous frequency drift, the generation accuracy of the current timestamp can be evaluated. For example, the larger the drift, the lower the generation accuracy. If the evaluated timestamp generation accuracy is lower than the preset accuracy threshold, the timestamp will be marked as a low-precision timestamp for special processing or discarding in subsequent data processing to avoid misjudgment due to inaccurate timestamps.

[0095] This application's solution, based on preliminary verification, introduces real-time monitoring of the sensor's internal environmental parameters (such as temperature and pressure) and combines this with the physical characteristics of the sensor's clock crystal oscillator, enabling dynamic and accurate evaluation of timestamp generation accuracy. Traditional methods only check the timestamp's format and range, failing to delve into the physical characteristics of the timestamp generation source, thus potentially overlooking accuracy issues caused by internal clock drift. This solution calculates the instantaneous frequency drift in real time and uses this as the basis for evaluating timestamp generation accuracy, allowing the platform to identify timestamps that, while formatted correctly, lack sufficient actual precision. This allows for a more comprehensive assessment of timestamp reliability, providing a more accurate time reference for subsequent event correlation and causal inference.

[0096] Through the above technical solution, this application can significantly improve the depth and accuracy of the verification of sensor data timestamps in the intelligent fire protection IoT integrated management platform. Compared with the preliminary verification that only checks the format and range, this solution can identify low-precision timestamps caused by environmental factors affecting the internal clock crystal oscillator of the sensor, thereby effectively avoiding misjudgments or association errors caused by inaccurate timestamps. This not only improves the reliability of data processing, but also provides a more solid data foundation for subsequent comprehensive analysis of abnormal events and sensor status judgment, ensuring the timeliness and accuracy of fire early warning and response.

[0097] In some preferred embodiments, it is assumed that a smoke sensor is operating in an industrial environment. The event data reported by the sensor includes a timestamp. The platform first performs a preliminary check on the format integrity and numerical range of the timestamp. If the check passes, the platform further obtains the real-time temperature monitored by the sensor as 80°C and the pressure as 1.2 atmospheres. Simultaneously, based on the model of the smoke sensor, it queries the typical frequency drift characteristics of its internal clock crystal oscillator under the conditions of 80°C and 1.2 atmospheres. Based on these parameters, the system calculates the instantaneous frequency drift of the crystal oscillator in real time as +100 ppm (parts per million). According to a preset evaluation model, if the drift exceeds 50 ppm, the timestamp generation accuracy is considered low. Therefore, this timestamp is marked as a low-precision timestamp. In subsequent processing, the platform may choose to weight this low-precision timestamp, or when performing time-sensitive event association, it may prioritize data from other high-precision sensors to avoid misjudgments caused by insufficient timestamp accuracy, such as incorrectly associating a brief electromagnetic interference event as a fire event.

[0098] In some embodiments of this application, when performing real-time fluctuation analysis on timestamp difference data, a preset normal fluctuation threshold is typically used for comparison. However, in practical applications, the operating environment of sensors is complex and variable. Especially when the sensor is in different power consumption modes, the drift characteristics of its internal clock crystal oscillator will change significantly, causing the fixed preset threshold to fail to accurately reflect the true timestamp fluctuation of the sensor under different operating conditions. This may lead to misjudgment, such as misjudging the timestamp fluctuation caused by normal power consumption mode switching as abnormal, or failing to detect the real anomaly caused by power consumption mode changes in a timely manner. If the above problems are not solved, the accuracy and reliability of the smart fire protection IoT integrated management platform in judging the sensor status will be affected. To address this, this application further proposes a more refined real-time fluctuation analysis method for timestamp difference data, which dynamically adjusts the normal fluctuation threshold to adapt to the characteristic changes of the sensor under different operating states.

[0099] In response, this application further proposes the aforementioned real-time volatility analysis of the timestamp difference data. The volatility analysis includes: calculating the instantaneous rate of change of multiple consecutive timestamp difference values ​​and comparing it with a preset normal volatility threshold, including: It continuously receives event data from the new fire sensor and extracts the timestamp and the power consumption mode information of the sensor from the event data; Based on the power consumption mode information, the typical drift characteristic parameters and recovery time parameters of the sensor clock crystal oscillator under the power consumption mode are obtained from the pre-stored correspondence between power consumption mode and clock drift characteristics. Based on the typical drift characteristic parameters and recovery time parameters, as well as the historical switching frequency and duration of the sensor switching from one power consumption mode to another, the normal fluctuation threshold of the timestamp difference value is dynamically calculated and updated. Calculate the instantaneous rate of change of multiple consecutive timestamp differences; The instantaneous rate of change is compared with the dynamically updated normal fluctuation threshold.

[0100] Specifically, the novel fire sensor refers to a sensor with multiple power consumption modes, whose clock crystal oscillator drift characteristics change with the power consumption mode. Examples include low-power wide-area network (LPWAN) sensors and self-powered sensors. The power consumption mode information can be understood as the sensor's current operating state, such as sleep mode, standby mode, and active mode, reflecting the operating conditions of the sensor's internal clock crystal oscillator. The correspondence between power consumption modes and clock drift characteristics refers to the typical drift characteristic parameters of the sensor's clock crystal oscillator under each power consumption mode, established in advance through experiments, simulations, or data provided by the manufacturer, and the recovery time parameter after switching from one mode to another. This aims to provide basic data for dynamic threshold calculation. The typical drift characteristic parameters may include indicators such as clock drift rate and drift stability. The recovery time parameter refers to the time required for the clock crystal oscillator to reach a stable drift state after the sensor switches from one power consumption mode to another.

[0101] Specifically, the dynamic calculation and updating of the normal fluctuation threshold of the timestamp difference value refers to comprehensively considering the sensor's current power consumption mode, its corresponding typical drift characteristic parameters and recovery time parameters, as well as the historical frequency and duration of the sensor switching between different power consumption modes, and adjusting the normal fluctuation range of the timestamp difference value in real time through an algorithm model. For example, when the sensor is in a low-power mode, its clock crystal may have a large drift, and the threshold will be appropriately relaxed; while when the sensor is in a high-precision working mode, the threshold will be tightened.

[0102] Meanwhile, if the sensor frequently switches between different power consumption modes, and a certain recovery time is required after each switch, the threshold will be dynamically adjusted accordingly during the recovery period to avoid misjudging fluctuations during the normal recovery process as abnormal. The calculation method of the instantaneous rate of change is similar to that described in the above embodiments, but the benchmark for comparison is no longer a fixed preset threshold, but a normal fluctuation threshold that is dynamically updated according to the above method.

[0103] This application, through the aforementioned technical solution, significantly improves the accuracy of the smart fire protection IoT integrated management platform in identifying sensor timestamp anomalies. Compared to using a fixed preset threshold method, this application can adaptively adjust the normal fluctuation threshold of timestamp differences based on the actual operating conditions of the sensor (especially changes in power consumption mode). This effectively avoids misjudging normal timestamp fluctuations caused by sensor power consumption mode switching or low-power operation as anomalies, thereby reducing the false alarm rate. Simultaneously, for genuine anomalies caused by external interference or sensor malfunction, since the threshold more accurately reflects the normal state, they can also be detected more sensitively, reducing the false alarm rate. This dynamic, adaptive threshold management mechanism enables the platform to more reliably determine the actual operating status of sensors, providing a more solid data foundation for the stable operation of the smart fire protection system.

[0104] In some preferred embodiments, a specific example is given below. Assume a novel fire sensor periodically switches between "active mode" and "sleep mode" during normal operation. In "active mode," the drift characteristics of the sensor's clock crystal are relatively stable, and the platform can set a narrow normal fluctuation threshold. However, when the sensor switches to "sleep mode," to save power, its internal clock crystal may reduce its frequency or enter a low-power state, causing changes in the timestamp drift characteristics, and requiring a 5-second recovery time to stabilize after the mode switch.

[0105] The proposed solution continuously receives event data reported by the sensor and extracts its timestamps and power consumption mode information. When the sensor switches from "active mode" to "sleep mode," the system obtains the typical drift characteristic parameters (e.g., a slightly higher drift rate) and 5-second recovery time parameters of the sensor's clock crystal oscillator in "sleep mode" based on a pre-stored correspondence between power consumption modes and clock drift characteristics. Simultaneously, the system records the sensor's historical mode switching frequency and duration. Based on this information, the system dynamically calculates and updates the normal fluctuation threshold for timestamp differences. For example, during the 5-second recovery time after a mode switch, the threshold may be appropriately relaxed to allow for normal fluctuations caused by the mode switch; while during stable operation in "sleep mode," the threshold is adjusted according to the typical drift characteristics of that mode. When the calculated instantaneous rate of change of timestamp differences is compared with this dynamically updated threshold, only anomalies that truly exceed the normal fluctuation range of that mode are flagged, while normal fluctuations caused by power consumption mode switching do not trigger anomaly alarms. This ensures the accuracy and robustness of timestamp anomaly detection across different power consumption modes.

[0106] In some embodiments described above, a normal fluctuation threshold for timestamp differences is proposed, based on power consumption mode information, typical drift characteristic parameters, recovery time parameters, and the historical switching frequency and duration of the sensor switching from one power consumption mode to another. However, in practical applications, the sensor's clock crystal oscillator is affected not only by power consumption modes but also by external physical environments such as vibration or shock. These physical stresses can also cause clock drift, thus affecting the accuracy of the timestamps. If only power consumption modes are considered and these physical stresses are ignored, the calculated normal fluctuation threshold may not fully reflect the sensor's true behavior in complex environments, potentially leading to misjudgments of abnormal situations.

[0107] In response, this application further proposes a method for dynamically calculating and updating the normal fluctuation threshold of the timestamp difference value based on the aforementioned typical drift characteristic parameters and recovery time parameters, as well as the historical switching frequency and duration of the sensor switching from one power consumption mode to another. This method includes: It continuously receives event data reported by the sensor and extracts the timestamp, the current power consumption mode information of the sensor, and the vibration or impact intensity data inside the sensor from the event data. Based on the power consumption mode information of the sensor, the typical drift characteristic parameters and recovery time parameters of the sensor clock crystal oscillator under the power consumption mode are obtained from the pre-stored correspondence between power consumption mode and clock drift characteristics. Based on the vibration or impact intensity inside the sensor, the typical mechanical stress drift parameters of the sensor clock crystal under the vibration or impact intensity are obtained from the pre-stored correspondence between vibration and impact and clock drift characteristics. Based on the typical drift characteristic parameters, the recovery time parameters, the typical mechanical stress drift parameters, and the historical switching frequency and duration of the sensor switching from one power consumption mode to another, the normal fluctuation threshold of the timestamp difference value is dynamically calculated and updated.

[0108] Specifically, vibration or shock intensity data within the sensor refers to real-time data collected by the sensor's built-in accelerometer, gyroscope, or other vibration / shock sensors, used to quantify the level of mechanical stress experienced by the sensor. This data can include vibration frequency, amplitude, peak impact acceleration, etc. Its purpose is to capture physical disturbances that may cause instantaneous or persistent drift in the clock crystal oscillator frequency.

[0109] The relationship between vibration and shock and clock drift characteristics can be understood as a database or model established in advance through experiments, simulations, or empirical summaries. This database or model describes the frequency drift, drift direction, and recovery characteristics of the internal clock crystal oscillator of a specific sensor model under different vibration or shock intensities. Its purpose is to provide accurate physical stress-related drift parameters for subsequent threshold calculations.

[0110] In practical applications, typical mechanical stress drift parameters are specifically found from the above-mentioned correspondence between vibration / impact and clock drift characteristics, based on the vibration or impact intensity of the sensor. These parameters represent the typical frequency deviation that the sensor's clock crystal oscillator may produce under specific mechanical stress conditions. For example, when a sensor is subjected to an impact of a certain intensity, its internal crystal oscillator may undergo slight deformation, causing a momentary frequency shift; this parameter quantifies this shift.

[0111] This application, through the aforementioned technical solution, significantly improves the calculation accuracy and adaptability of the normal fluctuation threshold for timestamp differences. By fully considering the clock drift characteristics of the sensor under different power consumption modes and under the influence of mechanical stress such as vibration or impact, the obtained threshold can more accurately define the normal fluctuation range of the sensor timestamp. This enables the platform to more accurately identify truly abnormal timestamps when performing real-time fluctuation analysis on the sensor-reported data, rather than misjudging normal drift caused by physical stress as abnormal. Compared to solutions that only consider power consumption modes, this application effectively reduces false alarms and missed alarms, improving the accuracy and reliability of the smart fire protection IoT integrated management platform in judging the sensor's operating status, especially in scenarios with complex physical environments such as industrial sites, where its advantages are even more pronounced.

[0112] In response, this application further proposes a method for dynamically calculating and updating the normal fluctuation threshold of the timestamp difference value, which includes: The power consumption mode information, vibration or shock intensity data, real-time temperature data, and historical switching frequency and duration are taken as inputs, and a multivariate nonlinear mapping function is used to output the predicted fluctuation range of the timestamp difference value. Based on the predicted fluctuation range, determine the normal fluctuation threshold of the timestamp difference value; The multivariable nonlinear mapping function is adaptively adjusted by continuously monitoring the actual time drift data of the sensor under different power consumption modes, vibration or shock intensities, and temperature conditions.

[0113] Specifically, the power consumption mode information refers to the current operating state of the sensor, such as low-power mode, normal operating mode, or high-performance mode. Different power consumption modes affect the frequency stability of the internal clock crystal oscillator of the sensor. The vibration or shock intensity data refers to the external mechanical stress experienced by the sensor, such as vibration or shock caused by equipment operation or environmental vibration. These mechanical stresses may cause instantaneous drift of the crystal oscillator frequency. The real-time temperature data refers to the current temperature of the sensor's environment. Temperature changes are one of the key factors affecting crystal oscillator frequency drift. The historical switching frequency and duration refer to the frequency at which the sensor switches between different power consumption modes and the duration of continuous operation in each mode. This historical data reflects the dynamic characteristics of the sensor's long-term operation. These input parameters are fed into a multivariate nonlinear mapping function, which can capture the complex nonlinear relationships between these variables, thereby more accurately predicting the dynamic behavior of sensor clock drift.

[0114] The multivariate nonlinear mapping function can be understood as a complex mathematical model, such as a model built based on machine learning algorithms like neural networks, support vector regression, or Gaussian process regression. The purpose of this function is to map multiple interrelated input variables (power consumption pattern information, vibration or shock intensity data, real-time temperature data, historical switching frequency and duration) to a single output variable (the predicted fluctuation range of timestamp differences). Through this function, a more accurate predicted fluctuation range that considers the interaction of multiple complex factors can be obtained.

[0115] In practical applications, the predicted fluctuation range refers to the expected range of variation in the difference between the sensor timestamp and the platform's internal time base under current operating and environmental conditions. Based on this predicted fluctuation range, a normal fluctuation threshold for the timestamp difference value can be determined. For example, the upper and lower limits of the predicted fluctuation range can be used as the normal fluctuation threshold, or a certain safety margin can be added to the predicted fluctuation range.

[0116] Furthermore, the multivariate nonlinear mapping function is not fixed but adaptively adjusted by continuously monitoring the actual time drift data of the sensor under different power consumption modes, vibration or shock intensities, and temperature conditions. This means that the function can continuously learn and optimize its internal parameters based on new observation data, thereby improving the accuracy of predictions and the generalization ability of the model. For example, when the system receives new sensor time drift data, this data can be used to retrain or fine-tune the mapping function, enabling it to better adapt to changes in drift characteristics that may occur during long-term sensor operation.

[0117] This application significantly improves the accuracy and robustness of setting the normal fluctuation threshold for timestamp differences. Compared to traditional methods that rely solely on typical drift characteristic parameters and recovery time parameters, this application introduces a multivariate nonlinear mapping function and combines real-time temperature data, vibration or shock intensity data, and historical operating data to more precisely characterize the complex dynamic behavior of sensor clock drift. Therefore, the determined normal fluctuation threshold can more accurately reflect the true drift characteristics of the sensor under various complex operating conditions and environments, effectively avoiding false alarms or missed alarms caused by improper threshold settings. Furthermore, the adaptive adjustment capability of this multivariate nonlinear mapping function allows the system to continuously learn and optimize to adapt to potential performance degradation or environmental changes that may occur during long-term sensor operation, thus ensuring the long-term effectiveness and accuracy of the threshold setting. This highly adaptive and precise threshold setting enables the intelligent fire protection IoT integrated management platform to more effectively distinguish between apparent anomalies caused by external interference and actual sensor malfunctions when judging the actual operating status of sensors, thereby improving the reliability of the fire assessment process and the accuracy of decision-making, reducing operation and maintenance costs, and enhancing the overall system's intelligence level.

[0118] refer to Figure 2 , Figure 2 This is a schematic diagram of the data processing system structure of a smart fire protection IoT integrated management platform provided in an embodiment of the present invention, including: The input terminal is used to acquire and record multiple operational performance data of each sensor under stable operating conditions; when data reported by the sensor is received and an anomaly occurs during the data parsing process, the data is identified and marked as an abnormal event. The analysis unit is used to perform a comprehensive analysis of the abnormal event; based on the results of the comprehensive analysis, it determines the actual operating status of the sensor; wherein the comprehensive analysis includes: The current abnormal data is compared with the description of the normal behavior pattern of the sensor to determine whether the current abnormal data conforms to the description of the normal behavior pattern. The processing unit is used to check whether the time of the abnormal event is related to the operating time of production equipment known to generate strong electromagnetic interference; and to check whether other sensors geographically close to the abnormal sensor report abnormalities within the same time period. The judgment includes: if the abnormal situation is significantly inconsistent with the description of the normal behavior pattern of the sensor and cannot be explained by associating it with the strong electromagnetic interference event, then the sensor is determined to be an actual malfunction. If the abnormal situation does not match the description of the normal behavior of the sensor, and is associated with the strong electromagnetic interference event, then the abnormal situation is determined to be an apparent abnormality caused by external interference; if so, then the fire data reported by the sensor is trusted, and the fire data is sent to the fire judgment process for processing.

[0119] In modern industrial environments, intelligent fire protection IoT integrated management platforms play a crucial role in improving the timeliness of fire early warning and the efficiency of safety hazard investigation. However, traditional data processing methods, when dealing with data anomalies caused by strong electromagnetic interference, risk misjudging normally functioning sensors as faulty, potentially leading to the downgrading or suppression of real fire data and severely impacting the timeliness of emergency response. For example, suppose that near the control room of a large steel smelter, strong electromagnetic interference generated by nearby large production equipment causes a manufacturer-specific data segment length field in the reported data packet to flip. Based on the design principle of "trusting input length," the platform's data parser reads an abnormally large length value and attempts to skip the data based on this value, resulting in a "buffer overflow read." This incorrectly interprets the "low battery" status information of another adjacent sensor (e.g., a smoke detector) in memory as diagnostic information for the current data packet. If this problem is not addressed, the logic module responsible for equipment status monitoring within the platform will incorrectly classify this actually powered and functioning new fire protection sensor as "faulty" or "offline." Furthermore, when a real fire occurs in the area and fire data is reported, the rules set responsible for fire judgment in the platform will downgrade or directly suppress the reported real fire data because the sensor is considered unreliable. This ultimately causes the platform to fail to trigger an effective fire warning in a timely manner, thus delaying the emergency response.

[0120] To address this issue, this application proposes a data processing system for a smart fire protection IoT integrated management platform. Through the collaborative operation of its input, analysis, and processing ends, the system achieves refined processing and accurate judgment of abnormal sensor data. The system can acquire and record multiple operational performance data of sensors under stable operating conditions. When data analysis reveals anomalies, it identifies and marks abnormal events and performs comprehensive analysis. This comprehensive analysis includes comparing the current abnormal data with descriptions of the sensor's normal behavior patterns and checking the correlation between the abnormal event and strong electromagnetic interference events or other sensor anomalies. Finally, based on the comprehensive analysis results, the actual operating status of the sensor is judged, thereby effectively distinguishing between actual sensor malfunctions and apparent anomalies caused by external interference. This ensures accurate processing of fire data, avoids misjudgments, and improves the platform's reliability and safety.

[0121] The data processing system proposed in this application aims to address the problem of misjudgments caused by electromagnetic interference leading to abnormal sensor data in complex industrial environments within a smart fire protection IoT integrated management platform. Through a sophisticated data analysis and judgment mechanism, this system ensures that the platform can accurately identify the true operating status of sensors, preventing delays in fire warnings due to misjudgments.

[0122] In some embodiments of this application, the specific steps and principles of the data processing method have been described in detail in the above embodiments, and will not be repeated here. It should be emphasized that the data processing system of this application, through its specific functional module design, concretely implements the above method steps as operable system components.

[0123] The core innovation of the data processing system of the smart fire protection IoT integrated management platform proposed in this application lies in its modular system design, which enables more refined and accurate diagnosis of abnormal sensor data. Traditional data processing systems often integrate data parsing, anomaly detection, and status updates into a single module. When data is damaged due to external interference, the system struggles to distinguish whether the fault lies with the sensor itself or with external environmental influences, easily leading to misjudgments.

[0124] In contrast, this system constructs a multi-dimensional, collaborative diagnostic mechanism by clearly defining the functions of the input, analysis, and processing ends. The input end focuses on data acquisition, preliminary analysis, and anomaly labeling, ensuring that all potentially problematic data is captured. The analysis end is responsible for comparing abnormal data with the normal behavior patterns of sensors, providing a basis for judging internal behavioral anomalies. The processing end focuses on the correlation analysis of external environmental factors, such as strong electromagnetic interference and regional sensor anomalies, providing external explanations for abnormal events. This collaborative system architecture allows the platform to comprehensively consider the internal behavior of sensors and the influence of the external environment, thereby avoiding misjudgments caused by a single judgment criterion in traditional systems. For example, when a sensor causes an error in the data packet length field due to electromagnetic interference, a traditional system might directly mark it as a fault; however, this system can identify the correlation with strong electromagnetic interference through the processing end, and combine it with the analysis end's judgment of the sensor's normal behavior patterns, ultimately identifying it as a superficial anomaly, thus trusting the reported fire data and avoiding delays in fire warnings. As a result, this system significantly improves the robustness and decision-making accuracy of the smart fire protection IoT integrated management platform in complex industrial environments.

[0125] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A data processing method of a smart fire-fighting Internet of Things comprehensive management platform, characterized in that, The method comprises the following steps: acquiring and recording multiple performance data of each sensor in a stable running state; when receiving the data reported by the sensor and the data appears abnormal in the analysis process, identifying and marking the data as an abnormal event; comprehensively analyzing the abnormal event; judging the actual running state of the sensor according to the comprehensive analysis result; wherein the comprehensive analysis comprises: comparing the current abnormal data with the normal behavior rule description of the sensor to determine whether the current abnormal data conforms to the normal behavior rule description; checking whether the time of the abnormal event is related to the running time of the production equipment known to produce strong electromagnetic interference; and checking whether other sensors close to the abnormal sensor report abnormal conditions in the same period; the judgment comprises: if the abnormal condition is significantly inconsistent with the normal behavior rule description of the sensor and cannot be explained by association with the strong electromagnetic interference event, it is determined that the sensor is actually faulty; if the abnormal condition is inconsistent with the normal behavior rule description of the sensor and can be explained by association with the strong electromagnetic interference event, it is determined that the abnormal condition is a superficial abnormality caused by external interference; if so, the fire data reported by the sensor is trusted, and the fire data is sent to the fire judgment process for processing. 2.The data processing method of the intelligent fire-fighting Internet of Things comprehensive management platform according to claim 1, characterized in that, The judgment of the actual running state of the sensor according to the comprehensive analysis result comprises: continuously collecting event data from sensors, production control systems and network monitoring modules, and performing preliminary standardization processing on the time stamp, wherein the sensor additionally contains an internal processing state code and a data acquisition time stamp in the manufacturer-specific data segment; defining event types including hidden electromagnetic interference events, sensor data bit flip events, sensor abnormal report events, production equipment official start events and network time delay fluctuation events, and extracting the event type occurrence time stamp, related sensor identifier, device identifier and other auxiliary information; building a directed acyclic graph as a causal association network, wherein the nodes in the causal association network represent the event types, and the edges in the causal association network represent the causal relationship; real-time inference is performed according to the causal association network to evaluate the real cause of the abnormality. 3.The data processing method of the intelligent fire-fighting Internet of Things comprehensive management platform according to claim 2, characterized in that, The continuous collection of event data from sensors, production control systems and network monitoring modules, and the preliminary standardization processing of the time stamp, wherein the sensor additionally contains an internal processing state code and a data acquisition time stamp, comprises: acquiring event data from different data sources and performing preliminary standardization processing on the time stamp, converting all time stamps to a unified reference time; establishing a dynamic time offset feature record for each data source, which calculates and updates the time offset and fluctuation range in real time by continuously monitoring the difference between the time stamp of the event reported by the data source and the internal time reference of the platform, and combining the time synchronization mechanism and time drift characteristics of the data source; According to the event source data in the event data, the current effective time offset and fluctuation range are obtained from the dynamic time offset feature record; Based on the time offset and the fluctuation range, and the network transmission delay data when the event occurs in the event data, the preliminary standardized timestamp in the event data is finely adjusted, and the adjustment process considers the nonlinear characteristics of time offset; A timestamp calibration confidence evaluation mechanism is introduced, and after each calibration, the calibration confidence is calculated according to the difference between the timestamps before and after calibration, the stability of the time offset feature record, and the real-time fluctuation of the network transmission delay; If the calibration confidence is lower than the preset threshold, the event timestamp in the event data is marked as low confidence.

4. The data processing method of the intelligent fire-fighting Internet of Things comprehensive management platform according to claim 3, characterized in that, The dynamic time offset feature record is established for each data source, and the dynamic time offset feature record is calculated and updated in real time by continuously monitoring the difference between the timestamp of the event reported by the data source and the internal time reference of the platform, and combining the time synchronization mechanism and the time drift characteristics of the data source, including: Continuously monitoring the difference between the timestamp of the event reported by the data source and the internal time reference of the platform; Detecting the instantaneous sharp change of the environmental temperature or pressure parameter; When the instantaneous sharp change of the environmental temperature or pressure parameter is detected, a high-frequency data sampling mode is started, and the timestamp difference data of the sensor is intensively collected; Combined with the change rate of the environmental parameter, the calculation window and the update frequency of the time offset are adjusted; Using a prediction function based on nonlinear regression, the instantaneous frequency drift of the internal clock crystal oscillator of the sensor is predicted as input; Combined with historical drift data, the time offset is corrected; An adaptive weighting mechanism is introduced, and higher weight is given to the timestamp difference data to update the time offset and its fluctuation range.

5. The data processing method of the intelligent fire-fighting Internet of Things comprehensive management platform according to claim 4, characterized in that, The continuously monitoring the difference between the timestamp of the event reported by the data source and the internal time reference of the platform includes: Continuously receiving event data reported by the data source, and extracting timestamps from the event data; The extracted timestamps are preliminarily checked, and the preliminary check includes checking the format integrity of the timestamp, and whether the timestamp value is within a reasonable time range; For the timestamps that pass the preliminary check, the difference between the timestamp and the internal time reference of the platform is calculated to obtain timestamp difference data; Real-time volatility analysis is performed on the timestamp difference data, and the volatility analysis includes calculating the instantaneous change rate of a plurality of continuous timestamp difference values, and comparing with a preset normal fluctuation threshold; When the instantaneous change rate exceeds the preset normal fluctuation threshold, an abnormal timestamp identification process is triggered.

6. The data processing method of the intelligent fire-fighting Internet of Things comprehensive management platform according to claim 5, characterized in that, The preliminary check of the extracted timestamp includes checking the format integrity of the timestamp, and whether the timestamp value is within a reasonable time range, including: The time stamp is preliminarily checked, and the preliminary check includes checking the format integrity of the time stamp and whether the time stamp value is within a reasonable time range; After the preliminary check passes, the temperature and pressure parameters inside the sensor are obtained; Based on the temperature and pressure parameters and the clock crystal physical characteristic parameters corresponding to the sensor model, the instantaneous frequency drift of the internal clock crystal of the sensor is calculated in real time; According to the instantaneous frequency drift, the generation accuracy of the time stamp is evaluated; If the generation accuracy of the time stamp is lower than a preset accuracy threshold, the time stamp is marked as a low-accuracy time stamp.

7. The data processing method of the intelligent fire-fighting Internet of Things comprehensive management platform according to claim 5, characterized in that, The time stamp difference data is analyzed in real time, and the fluctuation analysis includes calculating the instantaneous change rate of a plurality of continuous time stamp difference values and comparing the instantaneous change rate with a preset normal fluctuation threshold, including: Continuously receiving event data from a new fire sensor, and extracting a time stamp and power consumption mode information of the sensor from the event data; According to the power consumption mode information, the typical drift characteristic parameters and recovery time parameters of the sensor clock crystal in the power consumption mode are obtained from a pre-stored power consumption mode and clock drift characteristic correspondence; Based on the typical drift characteristic parameters and recovery time parameters, and the historical switching frequency and duration of the sensor from one power consumption mode to another power consumption mode, the normal fluctuation threshold of the time stamp difference value is dynamically calculated and updated; The instantaneous change rate of a plurality of continuous time stamp difference values is calculated; The instantaneous change rate is compared with the dynamically updated normal fluctuation threshold. 8.The data processing method of the intelligent fire-fighting Internet of Things comprehensive management platform according to claim 7, characterized in that, The normal fluctuation threshold of the time stamp difference value is dynamically calculated and updated based on the typical drift characteristic parameters and recovery time parameters, and the historical switching frequency and duration of the sensor from one power consumption mode to another power consumption mode, including: Continuously receiving event data reported by the sensor, and extracting a time stamp, power consumption mode information of the sensor and vibration or impact intensity data inside the sensor from the event data; According to the power consumption mode information of the sensor, the typical drift characteristic parameters and recovery time parameters of the sensor clock crystal in the power consumption mode are obtained from a pre-stored power consumption mode and clock drift characteristic correspondence; According to the vibration or impact intensity inside the sensor, the typical mechanical stress drift parameters of the sensor clock crystal under the vibration or impact intensity are obtained from a pre-stored vibration impact and clock drift characteristic correspondence; Based on the typical drift characteristic parameters, the recovery time parameters, the typical mechanical stress drift parameters, and the historical switching frequency and duration of the sensor from one power consumption mode to another power consumption mode, the normal fluctuation threshold of the time stamp difference value is dynamically calculated and updated. 9.The data processing method of the intelligent fire-fighting Internet of Things comprehensive management platform according to claim 8, characterized in that, The normal fluctuation threshold of the time stamp difference value is dynamically calculated and updated, including: The power consumption mode information, vibration or impact intensity data, real-time temperature data, and historical switching frequency and duration are inputted into a multivariate nonlinear mapping function, and a predicted fluctuation range of the timestamp difference value is outputted; According to the predicted fluctuation range, a normal fluctuation threshold of the timestamp difference value is determined; The multivariate nonlinear mapping function is adaptively adjusted by continuously monitoring actual time drift data of the sensor under different power consumption modes, vibration or impact intensity, and temperature conditions.

10. A data processing system of a smart fire-fighting Internet of Things comprehensive management platform, characterized in that, Comprise: An input end is configured to acquire and record multiple performance data of each sensor in a stable running state, and identify and mark the data as an abnormal event when the data reported by the sensor is received and an abnormality occurs in the analysis process; An analysis end is configured to comprehensively analyze the abnormal event; According to the comprehensive analysis result, the actual running state of the sensor is judged; wherein the comprehensive analysis comprises: Comparing the current abnormal data with the normal behavior rule description of the sensor to determine whether the current abnormal data conforms to the normal behavior rule description; A processing end is configured to check whether the time when the abnormal event occurs is associated with the running time of a production equipment known to generate strong electromagnetic interference, and whether other sensors close to the abnormal sensor report abnormal conditions in the same period; The judgment comprises: if the abnormal condition is significantly inconsistent with the normal behavior rule description of the sensor and cannot be explained by association with the strong electromagnetic interference event, it is determined that the sensor is actually faulty; If the abnormal condition is inconsistent with the normal behavior rule description of the sensor and can be explained by association with the strong electromagnetic interference event, it is determined that the abnormal condition is a surface abnormality caused by external interference; if so, the fire data reported by the sensor is trusted, and the fire data is sent to a fire judgment process for processing.

Images