Fire water supply system safety early warning method, device and equipment and storage medium

Abstract

The application discloses a fire water system safety early warning method and device, equipment and storage medium, relates to the field of information processing technology, and comprises the following steps: acquiring real-time working condition data and real-time environmental parameters of a fire water system through a global Internet of Things network, and determining a dynamic alarm threshold based on the real-time environmental parameters; determining whether a real abnormal event exists in the fire water system through multi-source data cross-validation rules according to the real-time working condition data, the real-time environmental parameters and the dynamic alarm threshold; if yes, predicting a water level change trend of the fire water system through a time series model, and determining an abnormal cause of the fire water system in combination with the real-time working condition data and the real-time environmental parameters to generate an abnormal repair scheme and an abnormal early warning. The application breaks data islands, dynamically adjusts alarm thresholds and satisfies the needs of modern fire safety management for precision, high efficiency and active protection.

Description

Technical Field

[0001] This application relates to the field of information processing technology, and in particular to a method, device, equipment and storage medium for safety early warning of fire protection and water conservancy systems. Background Technology

[0002] Urban fire monitoring has evolved from early stand-alone alarm devices to digital remote terminals, but "real-time reliability of the hydraulic system" remains the lifeline of fire fighting and rescue: the water pressure of the fire pipeline network, the water level of the water tank, and the status of the pump group must be uniformly perceived and responded to in a coordinated manner within a second; otherwise, the fire will get out of control due to the information vacuum and equipment lag.

[0003] However, this critical data has long been scattered across independent devices or closed platforms, lacking an IoT network that covers "water source - pipeline network - end point" and can coordinate in real time, making it difficult to meet the needs of modern fire safety management for precise, efficient, and proactive protection. Summary of the Invention

[0004] The main purpose of this application is to provide a method, device, equipment and storage medium for early warning of fire protection water conservancy system safety, which aims to meet the needs of modern fire safety management for accurate, efficient and proactive protection.

[0005] To achieve the above objectives, this application proposes a safety early warning method for fire protection and water conservancy systems, the method comprising: The system acquires real-time operating data and environmental parameters of the fire protection and water conservancy system through a global Internet of Things network, and determines dynamic alarm thresholds based on the real-time environmental parameters. Based on the real-time operating data, the real-time environmental parameters, and the dynamic alarm threshold, the existence of a real abnormal event in the fire-fighting water conservancy system is determined by multi-source data cross-validation rules. If so, the water level change trend of the fire-fighting water system is predicted by time series model, and the causes of the anomaly of the fire-fighting water system are determined by combining the real-time operating data and the real-time environmental parameters, so as to generate an anomaly repair plan and an anomaly early warning.

[0006] In one possible implementation, acquiring real-time operating data and real-time environmental parameters of the fire protection and water conservancy system through a global Internet of Things (IoT) network includes: The system adopts a full-domain IoT architecture, which collects raw operating data and raw environmental parameters through data transmission terminals, water pressure sensors and water level sensors deployed in fire protection pipe networks, water storage tanks and pumping stations. The full-domain IoT architecture includes a terminal acquisition layer, an edge computing layer and a cloud decision layer. The original operating condition data and the original environmental parameters are synchronized in time and space by the edge controller to generate the real-time operating condition data and the real-time environmental parameters. The real-time operating condition data includes water pressure value, water level value, pipeline flow velocity, and equipment operating status. The real-time environmental parameters include temperature, humidity, and pipeline vibration intensity.

[0007] In one possible implementation, determining the dynamic alarm threshold based on the real-time environmental parameters includes: Extract the diurnal temperature difference and pipeline vibration intensity from the real-time environmental parameters, and establish a mapping relationship between environmental impact factors and dynamic alarm thresholds; When the diurnal temperature difference is detected to exceed the preset temperature difference threshold, the threshold adjustment coefficient is determined by combining historical data fitting and expert experience, and the dynamic alarm threshold corresponding to water pressure and water level fluctuations is determined. When the vibration intensity of the pipeline exceeds the preset value, the threshold adjustment coefficient is determined by combining historical data fitting and expert experience, and the sensitivity threshold for anomaly judgment is adjusted to form a dynamically updated alarm threshold system that supports real-time adaptation of thresholds to environmental changes.

[0008] In one possible implementation, determining whether a real abnormal event exists in the fire-fighting water system based on the real-time operating data, the real-time environmental parameters, and the dynamic alarm threshold using multi-source data cross-validation rules includes: Based on the real-time operating data, the water pressure deviation rate and water level drop rate are calculated. Combined with the temperature fluctuation amplitude and equipment malfunction factors in the real-time environmental parameters, the risk index is calculated using the risk index formula. When the risk index exceeds the dynamic alarm threshold, multi-source data cross-validation is triggered to determine the authenticity of the abnormal event through multi-source data cross-validation rules. The cross-validation rules are to detect whether two or more of the following features exist simultaneously: sudden drop in water pressure, drop in water level, sudden change in flow rate, and increase in vibration signal. If the judgment result is true, then a real abnormal event is confirmed to exist.

[0009] In one possible implementation, the step of predicting the water level change trend of the fire-fighting water system using a time series model, and combining the real-time operating data and the real-time environmental parameters to determine the cause of the anomaly in the fire-fighting water system, in order to generate an anomaly repair plan and an anomaly early warning, includes: A time series model was used to analyze the acquired historical water level data, water pump operating status, and external ambient temperature to predict the trend of water level changes. The isolated forest algorithm is used to detect the real-time water pressure value and pipeline flow velocity in the real-time environmental parameters, and outputs the leakage probability score of the sudden water pressure drop event in the fire water conservancy system. Based on the water level change trend and the leakage probability score, the causes of anomalies are analyzed using Bayesian networks. The causes of anomalies include pipe rupture, water pump failure, and pipeline leakage. Based on the causes of the anomalies, an anomaly repair plan is generated, and anomaly warnings are pushed out according to the anomaly level.

[0010] In one possible implementation, after predicting the water level change trend of the fire-fighting water system using a time series model, and combining the real-time operating data and the real-time environmental parameters to determine the cause of the anomaly in the fire-fighting water system, in order to generate an anomaly repair plan and an anomaly early warning, the method further includes: Based on the historical operation data, past alarm records, and equipment maintenance logs of the fire-fighting water conservancy system, a time series prediction network structure is used to construct an equipment health model. An incremental learning strategy is adopted to conduct supervised training on the equipment health model. The model weights are dynamically updated according to the newly added operating data at fixed intervals. When the model prediction error is detected to continuously exceed the preset model error range, the model retraining process is automatically triggered to correct the equipment health model by combining the latest data and external environmental parameters. The trained equipment health model predicts the remaining lifespan of key equipment in the fire protection and water conservancy system and the probability of leakage in the pipeline network, generates targeted maintenance plans and equipment maintenance work orders, and pushes them to relevant operation and maintenance personnel.

[0011] In one possible implementation, after predicting the water level change trend of the fire-fighting water system using a time series model, and combining the real-time operating data and the real-time environmental parameters to determine the cause of the anomaly in the fire-fighting water system, in order to generate an anomaly repair plan and an anomaly early warning, the method further includes: Obtain the geographic coordinate data uploaded by each terminal in the global Internet of Things network, and construct a digital topology map of fire protection facilities in combination with the actual layout of the pipeline network; When an abnormal event is detected in the equipment deployment area, the system determines whether the abnormal event is caused by a fault in the upstream pipe section or related equipment based on the digital topology map of the fire protection facilities. The system then matches the cross-regional linkage node corresponding to the equipment deployment area and sends a collaborative control command to the edge controller of the cross-regional linkage node to complete the cross-regional abnormal correction collaborative operation. After the anomaly correction collaborative operation is completed, the repair status data of the fire-fighting water conservancy system after restoration is collected in real time, and an anomaly correction report is generated.

[0012] Furthermore, to achieve the above objectives, this application also proposes a fire-fighting water conservancy system safety early warning device, which includes: The acquisition module is used to acquire real-time operating data and real-time environmental parameters of the fire protection and water conservancy system through the global Internet of Things network, and determine the dynamic alarm threshold based on the real-time environmental parameters. The judgment module is used to determine whether there is a real abnormal event in the fire-fighting water conservancy system based on the real-time operating data, the real-time environmental parameters and the dynamic alarm threshold, and through multi-source data cross-validation rules. The early warning module is used to predict the water level change trend of the fire-fighting water system through a time series model if the condition is met, and to determine the cause of the abnormality of the fire-fighting water system by combining the real-time operating data and the real-time environmental parameters, so as to generate an abnormality repair plan and an abnormality early warning.

[0013] In addition, to achieve the above objectives, this application also proposes a fire protection and water conservancy system safety early warning device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the fire protection and water conservancy system safety early warning method as described above.

[0014] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the fire protection and water conservancy system safety early warning method described above.

[0015] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the fire protection and water conservancy system safety early warning method described above.

[0016] This application provides a method, device, equipment, and storage medium for safety early warning of fire-fighting water conservancy systems. The method acquires real-time operating data and real-time environmental parameters of the fire-fighting water conservancy system through a global Internet of Things (IoT) network. Based on the real-time environmental parameters, a dynamic alarm threshold is determined. Then, based on the real-time operating data, the real-time environmental parameters, and the dynamic alarm threshold, a multi-source data cross-validation rule is used to determine whether there are real abnormal events in the fire-fighting water conservancy system. If so, a time series model is used to predict the water level change trend of the fire-fighting water conservancy system. Combining the real-time operating data and the real-time environmental parameters, the cause of the abnormality in the fire-fighting water conservancy system is determined to generate an anomaly repair plan and an anomaly early warning. This breaks down data silos, achieves real-time fusion and synchronization of global data, and reduces the false alarm rate caused by environmental interference by dynamically adjusting the alarm threshold, thus achieving proactive early warning and promoting the transformation of safety early warning for fire-fighting water conservancy systems from passive handling to proactive prevention and control. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating an embodiment of the fire protection and water conservancy system safety early warning method of this application. Figure 2 This is a flowchart illustrating Embodiment 2 of the fire protection and water conservancy system safety early warning method of this application; Figure 3 This is a schematic diagram of the module structure of the fire protection and water conservancy system safety early warning device according to an embodiment of this application; Figure 4 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the fire protection and water conservancy system safety early warning method in the embodiments of this application.

[0020] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0022] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0023] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device, big data service platform, or fire and water conservancy system safety early warning system capable of realizing the above functions. The following description uses a fire and water conservancy system safety early warning system as an example to illustrate this embodiment and the subsequent embodiments.

[0024] Based on this, the embodiments of this application provide a safety early warning method for fire protection and water conservancy systems, referring to... Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the fire protection and water conservancy system safety early warning method of this application.

[0025] In this embodiment, the fire protection and water conservancy system safety early warning method includes steps S11 to S13: Step S11: Obtain real-time operating data and real-time environmental parameters of the fire protection and water conservancy system through the global Internet of Things network, and determine the dynamic alarm threshold based on the real-time environmental parameters. It should be noted that the global IoT network refers to a network system that adopts a three-tier architecture of "terminal acquisition layer - edge computing layer - cloud decision layer," and achieves wide-area interconnection and unified data transmission and management of various terminal devices in the fire protection and water conservancy system through a specific networking method. Its core function is to break down the data silos in traditional fire monitoring systems and achieve full-domain linkage and real-time synchronization of data. Real-time operating data refers to various data reflecting the operating status of the fire protection and water conservancy system generated in real time during operation, including water pressure, water level, pipeline flow velocity, and equipment operating status. Real-time environmental parameters refer to relevant data of the external environment in which the fire protection and water conservancy system is located, including temperature, humidity, and pipeline vibration intensity. Dynamic alarm thresholds refer to the critical values ​​that are dynamically adjusted according to real-time environmental parameters to determine whether there are abnormalities in the system. Unlike fixed alarm thresholds, they can adapt to the abnormality judgment needs under different environmental conditions. Thus, by achieving comprehensive data acquisition and real-time transmission through the global IoT network, and combining it with dynamic adjustment of alarm thresholds based on environmental parameters, false alarms or missed alarms caused by environmental changes can be effectively avoided, improving the accuracy and reliability of subsequent abnormality judgment.

[0026] Specifically, in one possible implementation, the global IoT network adopts a hybrid networking mode combining LoRaWAN and 4G. The LoRaWAN protocol is used for low-power, long-distance data transmission between terminal devices and edge controllers, while the 4G network is used for high-speed data backhaul between the edge computing layer and the cloud decision layer. This networking method can meet the needs of full coverage in urban areas while ensuring the stability and efficiency of data transmission.

[0027] Furthermore, real-time operating data is acquired through data transmission terminals, water pressure sensors, and water level sensors deployed at key nodes such as fire-fighting pipelines, water storage tanks, and pumping stations. The water pressure sensors utilize MEMS pressure chips to ensure acquisition accuracy, while the water level sensors employ ultrasonic level gauges to improve measurement resolution. For example, in a city's urban area, the terminal acquisition layer of the nationwide IoT network includes 50 water pressure sensors, 30 water level sensors, and 20 data transmission terminals distributed across key nodes of the fire-fighting pipeline network in various areas. These data transmission terminals integrate LoRaWAN communication and GPS positioning modules, acquiring and uploading real-time operating data such as water pressure, water level, pipeline flow velocity, and equipment operating status at a fixed 3-second interval, as well as real-time environmental parameters such as temperature, humidity, and pipeline vibration intensity. The edge controller uses the NTP protocol to synchronize the time of each terminal, generating integrated data with timestamps and geographic coordinates.

[0028] Furthermore, the system extracts the diurnal temperature difference and pipeline vibration intensity from real-time environmental parameters. When the diurnal temperature difference exceeds 8°C, it combines the fitting results of historical data from the past 6 months with the experience of fire protection experts to determine the threshold adjustment coefficient as 1.1, thus relaxing the alarm threshold corresponding to water pressure and water level fluctuations by 10%. When the pipeline vibration intensity is detected to be higher than 0.3g, the threshold adjustment coefficient is determined to be 0.9, increasing the sensitivity threshold for anomaly detection by 10%, ultimately forming a dynamically updated alarm threshold system.

[0029] Step S12: Based on the real-time operating data, the real-time environmental parameters, and the dynamic alarm threshold, determine whether there are any real abnormal events in the fire-fighting water conservancy system through multi-source data cross-validation rules; It should be noted that multi-source data cross-validation rules refer to rules established by combining the changing characteristics of various types of data (such as water pressure and water level data in real-time operating conditions, and vibration and temperature data in real-time environmental parameters) to determine whether an abnormal event truly exists. Its core logic is to eliminate misjudgments caused by interference from a single data point through cross-verification of multi-dimensional data. A real abnormal event refers to an event that actually occurs in the fire protection and water conservancy system, affecting the normal operation of the system or posing a safety hazard, such as pipe rupture, pump failure, or pipeline leakage. The purpose of this step is to accurately identify real abnormal events through multi-source data cross-validation, avoiding false alarms or missed alarms caused by single-data judgments, and ensuring the accuracy of anomaly identification.

[0030] For details, please refer to steps S41-S43, which will not be repeated here.

[0031] Step S13: If yes, then predict the water level change trend of the fire-fighting water system through a time series model, and combine the real-time operating data and the real-time environmental parameters to determine the cause of the abnormality of the fire-fighting water system, so as to generate an abnormality repair plan and an abnormality warning.

[0032] It should be noted that the time series model refers to the ARIMA model. This model, based on the time-series characteristics of data, predicts the trend of data changes over a future period by analyzing the changing patterns of historical data. The water level change trend refers to the upward, downward, or stable change of the water level in the fire-fighting water system over a future period. Its prediction results can provide a forward-looking reference for anomaly handling, helping to formulate response strategies in advance. The anomaly cause refers to the specific reason for the actual abnormal event in the fire-fighting water system, including types such as pipe rupture, pump failure, and network leakage. Identifying the anomaly cause is a prerequisite for developing targeted repair plans. The anomaly repair plan refers to the specific plan developed to resolve the anomaly and restore normal system operation based on the identified anomaly cause, including equipment operation instructions and maintenance measures. Anomaly early warning refers to the notification containing key anomaly information pushed to relevant maintenance personnel or management systems based on the severity of the anomaly event, so that relevant personnel can respond and handle it in a timely manner. The core objective of this step is to predict water level change trends and identify the root causes of anomalies after confirming a genuine abnormal event, generate targeted remediation plans, and promptly issue early warnings to achieve rapid and effective handling of abnormal events, reduce their impact on fire-fighting water systems, and improve the system's emergency response capabilities and operational reliability.

[0033] For details, please refer to steps S51-S54, which will not be repeated here.

[0034] This embodiment acquires real-time operating data and environmental parameters of the fire-fighting water conservancy system through a global IoT network. Based on the real-time environmental parameters, a dynamic alarm threshold is determined. Then, based on the real-time operating data, the real-time environmental parameters, and the dynamic alarm threshold, a multi-source data cross-validation rule is used to determine whether there are any real abnormal events in the fire-fighting water conservancy system. If so, a time series model is used to predict the water level change trend of the fire-fighting water conservancy system. Combining the real-time operating data and the real-time environmental parameters, the cause of the abnormality in the fire-fighting water conservancy system is determined to generate an anomaly repair plan and an anomaly warning. This breaks down data silos, achieves real-time fusion and synchronization of global data, and reduces the false alarm rate caused by environmental interference by dynamically adjusting the alarm threshold, realizing proactive early warning and promoting the transformation of fire-fighting water conservancy system safety early warning from passive handling to proactive prevention and control.

[0035] Based on this, the embodiments of this application provide a safety early warning method for fire protection and water conservancy systems, referring to... Figure 2 , Figure 2 This is a flowchart illustrating Embodiment 2 of the fire protection and water conservancy system safety early warning method of this application.

[0036] In one feasible implementation, the acquisition of real-time operating data and real-time environmental parameters of the fire-fighting and water conservancy system through a global Internet of Things network includes: Step S21: Adopting a full-domain IoT architecture, raw operating condition data and raw environmental parameters are collected by data transmission terminals, water pressure sensors and water level sensors deployed in fire-fighting pipelines, water storage tanks and pumping stations. The full-domain IoT architecture includes a terminal acquisition layer, an edge computing layer and a cloud decision layer. It should be noted that the terminal acquisition layer refers to the bottom layer of the whole-domain IoT architecture, composed of various data acquisition devices deployed in key locations of the fire protection and water conservancy system, responsible for the real-time acquisition of raw data; the edge computing layer refers to the middle layer located between the terminal acquisition layer and the cloud decision-making layer, deployed with LoRaWAN gateways and edge controllers, undertaking preliminary data processing and local decision-making tasks; the cloud decision-making layer refers to the top layer of the whole-domain IoT architecture, receiving data transmitted from the edge layer based on 4G / Ethernet, responsible for centralized modeling, global optimization decision-making, and unified data management. The data transmission terminal refers to a terminal device integrating a LoRaWAN communication module and a GPS positioning module, supporting data acquisition and transmission, capable of uploading data once per second; the water pressure sensor refers to a sensing device using a MEMS pressure chip (accuracy ±0.5% FS) to detect water pressure data within the fire protection pipe network; the water level sensor refers to a sensing device using an ultrasonic water level gauge (resolution 1mm) to detect water level data in water storage tanks, fire pools, etc. Raw operating condition data refers to unprocessed data directly collected by the equipment that reflects the operating status of the fire water system, such as raw water pressure data and raw water level data. Raw environmental parameters refer to unprocessed data related to the external environment of the fire water system directly collected by the equipment, such as raw temperature data and raw vibration data.

[0037] The purpose of this step is to comprehensively and accurately capture the raw operational and environmental data of the fire protection and water conservancy system through a three-level global IoT architecture and by utilizing data acquisition devices deployed at key nodes, thereby solving the problems of scattered data acquisition, incomplete coverage, and insufficient accuracy in traditional systems.

[0038] Specifically, in one possible implementation, the deployment of data transmission terminals, water pressure sensors, and water level sensors follows the principle of "full coverage of key nodes and denser deployment in high-risk areas." At least one set of data acquisition equipment is deployed at key nodes such as the connections between main and branch pipes of the fire protection network, the inlet and outlet locations of water storage tanks, and near pump units in pumping stations. For high-risk locations such as aging pipe networks and areas prone to leakage, the sensor deployment density is increased by 50%. The full-domain IoT architecture adopts a LoRaWAN / 4G hybrid networking method. Devices at the terminal acquisition layer prioritize data transmission via the LoRaWAN protocol (coverage radius up to 5km, supporting city-level deployment). In areas where LoRa signals are blocked or transmission is obstructed, it automatically switches to 4G network transmission to ensure data transmission continuity.

[0039] For example, a city's fire-fighting water system deployed 500 data transmission terminals, 600 water pressure sensors, and 200 water level sensors at 500 key nodes, 20 water storage tanks, and 15 pumping stations across a 300km fire-fighting pipeline network. The water pressure sensors used MEMS pressure chips to collect real-time raw water pressure data (e.g., 0.75MPa, 0.82MPa) within the pipeline network. The water level sensors used ultrasonic level gauges to collect real-time raw water level data (e.g., 3.2m, 3.6m) within the water storage tanks. The data transmission terminals integrated LoRaWAN communication and GPS positioning modules to simultaneously collect raw environmental parameters such as ambient temperature (e.g., 23℃, 26℃) and vibration intensity (e.g., 0.15g, 0.2g) around the equipment. All raw data was uploaded every 2 seconds via the LoRaWAN protocol, achieving comprehensive data collection from the entire fire-fighting water system.

[0040] Step S22: The original operating condition data and the original environmental parameters are spatiotemporally synchronized by the edge controller to generate the real-time operating condition data and the real-time environmental parameters. The real-time operating condition data includes water pressure value, water level value, pipeline flow velocity, and equipment operating status. The real-time environmental parameters include temperature, humidity, and pipeline vibration intensity.

[0041] It should be noted that the edge controller refers to the core processing device based on the ARM Cortex-A72 chip, integrating Modbus protocol parsing and AI inference capabilities, and deployed at the edge computing layer. It is responsible for preprocessing and coordinating control of raw data. Spatiotemporal synchronization refers to the edge controller assigning a unified timestamp to the raw data uploaded by different acquisition devices based on the NTP protocol, and combining it with the GPS positioning information (latitude and longitude coordinates) of the data transmission terminal to achieve data alignment in time and space, forming "spatiotemporally integrated" data. Real-time operating data refers to standardized data that accurately reflects the current operating status of the fire protection water conservancy system after spatiotemporal synchronization processing. This includes water pressure values ​​(calibrated standard water pressure data), water level values ​​(calibrated standard water level data), pipeline flow velocity (calculated based on water pressure changes and pipeline parameters), and equipment operating status (such as the start-up and shutdown status of pump sets, the opening and closing status of valves, etc.). Real-time environmental parameters refer to standardized data that accurately reflects the external environmental status of the system after spatiotemporal synchronization processing. This includes temperature (calibrated ambient temperature), humidity (calibrated ambient humidity), and pipeline vibration intensity (vibration data after filtering).

[0042] Specifically, in one possible implementation, the spatiotemporal synchronization processing flow of the edge controller for raw data includes: first, cleaning the raw data to remove outliers (such as data exceeding reasonable range due to sensor malfunction) and duplicate data; then, based on the NTP protocol, uniformly calibrating the timestamps of all cleaned raw data to UTC time, with the error controlled within 1ms; next, associating the latitude and longitude coordinates uploaded by the data transmission terminal to add a spatial identifier to each data point; finally, compressing and encrypting the data to generate standardized real-time operating condition data and real-time environmental parameters, and uploading them to the cloud decision layer at fixed intervals, while simultaneously retaining backup data locally (retention time is 72 hours).

[0043] For example, the edge controller receives raw data uploaded from 10 acquisition devices in a certain area, including raw water pressure data from 3 water pressure sensors (0.78MPa, 0.76MPa, 0.81MPa), raw water level data from 2 water level sensors (3.3m, 3.4m), and raw temperature data (24℃, 25℃, 24.5℃, etc.) and raw vibration intensity data (0.18g, 0.19g, etc.) from 5 data transmission terminals. The edge controller first removes one abnormal water pressure data point (1.5MPa) caused by poor sensor contact. Then, it uses the NTP protocol to uniformly calibrate the timestamps of all valid raw data to "2024-05-20 10:30:00.000" and associates them with the latitude and longitude coordinates of each device (e.g., 116.35°E, 39.88°N). It then calculates the pipeline flow velocity to be 0.28m / s, confirms the pump set's operating status as "normal operation," and finally generates real-time operating condition data including water pressure of 0.78MPa, water level of 3.35m, pipeline flow velocity of 0.28m / s, and equipment operating status as "normal operation," as well as real-time environmental parameters such as temperature of 24.5℃, humidity of 58%, and pipeline vibration intensity of 0.185g. This data is then uploaded to the cloud platform.

[0044] This embodiment adopts a three-level global IoT architecture, which clearly defines the division of labor among the various levels of the architecture, and realizes the orderly connection of data collection, processing and management. At the same time, the edge terminal performs spatiotemporal synchronous processing, which gives the data spatiotemporal attributes, improves the data correlation and availability, and deploys multiple types of sensors at key nodes to comprehensively collect core data, ensuring data coverage and integrity, thereby breaking down data silos and meeting the precision requirements of modern fire safety management.

[0045] In one feasible implementation, determining the dynamic alarm threshold based on the real-time environmental parameters includes: Step S31: Extract the diurnal temperature difference and pipeline vibration intensity from the real-time environmental parameters, and establish a mapping relationship between environmental impact factors and dynamic alarm thresholds; It should be noted that diurnal temperature difference refers to the difference between the highest and lowest ambient temperatures within the same monitoring period, and is an important environmental factor affecting the stability of water pressure and level in fire protection pipe networks. Pipeline vibration intensity refers to the amplitude of vibration generated during the operation of the fire protection pipe network (usually measured in grams). Its variation may be caused by factors such as water flow impact and external construction interference, and is easily misjudged as abnormal by traditional fixed thresholds. Environmental impact factors refer to environmental parameters that can affect the operational status of fire protection water systems, specifically including diurnal temperature difference and pipeline vibration intensity. The mapping relationship refers to the correspondence between environmental impact factor values ​​and dynamic alarm thresholds established through mathematical models or rules, ensuring that the alarm thresholds accurately adapt to changes in environmental parameters.

[0046] The purpose of this step is to identify environmental factors that significantly affect the system's anomaly detection and establish their correlation logic with alarm thresholds, thereby fundamentally solving the problem that traditional fixed thresholds cannot adapt to environmental changes and are prone to false alarms.

[0047] Specifically, in one possible implementation, the system uses an edge controller to filter and calculate real-time environmental parameters. When extracting diurnal temperature differences, a 24-hour monitoring cycle is used to statistically analyze the highest and lowest temperatures within that cycle and calculate the difference. When extracting pipeline vibration intensity, the collected raw vibration data is filtered to remove instantaneous interference signals and retain effective vibration intensity values. The mapping relationship is established using a linear regression model. Historical environmental impact factor data and corresponding optimal alarm threshold data are used as training samples. The mapping function is obtained by fitting using the least squares method to ensure the accuracy of the linear correlation between environmental impact factors and dynamic alarm thresholds, while also supporting continuous optimization of the mapping relationship based on new data.

[0048] For example, the system collects environmental data of a fire protection pipe network in a certain area in real time, calculates that the highest temperature in 24 hours is 32℃ and the lowest temperature is 18℃, so the diurnal temperature difference is 14℃. After filtering the raw pipe vibration data, the effective pipe vibration intensity is obtained as 0.3g. Based on a pre-trained linear regression mapping model, the system uses the diurnal temperature difference of 14℃ and the pipe vibration intensity of 0.3g as inputs to establish a mapping relationship between them and the water pressure fluctuation alarm threshold and the water level fluctuation alarm threshold. That is, for every 1℃ increase in the diurnal temperature difference, the water pressure fluctuation alarm threshold is adjusted by 0.005MPa, and for every 0.1g increase in the pipe vibration intensity, the water level fluctuation alarm threshold is adjusted by 0.05m.

[0049] Step S32: When the diurnal temperature difference is detected to exceed the preset temperature difference threshold, the threshold adjustment coefficient is determined by combining historical data fitting and expert experience, and the dynamic alarm threshold corresponding to water pressure and water level fluctuations is determined. It should be noted that the preset temperature difference threshold refers to a critical value (e.g., 10℃) set in advance based on the operating characteristics of the fire protection water system, historical environmental data, and expert experience. This threshold is used to determine whether the diurnal temperature difference has a significant impact on the system. When the actual diurnal temperature difference exceeds this value, it indicates that changes in ambient temperature may cause normal fluctuations in the pipe network water pressure and level, requiring adjustment of the alarm threshold. Historical data fitting refers to using diurnal temperature difference data, corresponding pipe network water pressure / level fluctuation data, and alarm record data over a past period (e.g., 6 months) to analyze the changing patterns between the temperature difference and the optimal alarm threshold through mathematical algorithms (e.g., polynomial fitting). Expert experience refers to the judgment rules on the impact of temperature difference on alarm thresholds summarized by professionals with rich practical experience in fire protection engineering and equipment operation and maintenance, based on actual engineering cases. The threshold adjustment coefficient refers to a proportional coefficient (e.g., 1.1, 1.2, etc.) determined based on historical data fitting results and expert experience, used to adjust the initial alarm threshold. The dynamic alarm threshold corresponding to water pressure and level fluctuations refers to the new alarm threshold obtained by multiplying the initial alarm threshold by the threshold adjustment coefficient, adapting to the current diurnal temperature difference.

[0050] The purpose of this step is to adjust the alarm thresholds for water pressure and water level using a scientific coefficient determination method when the diurnal temperature difference exceeds the normal range. This avoids misjudging normal fluctuations caused by temperature changes as abnormalities, further reducing the false alarm rate, while ensuring the rationality of the alarm threshold adjustment.

[0051] Specifically, in one possible implementation, the preset temperature difference threshold can be flexibly configured according to the climatic characteristics of different regions. For example, the preset temperature difference threshold for cold northern regions can be set to 8℃, and for warm southern regions, it can be set to 12℃; no restrictions are placed here. Historical data fitting employs a cubic polynomial fitting algorithm, using the diurnal temperature difference as the independent variable and the corresponding optimal water pressure / level alarm threshold as the dependent variable, to fit a nonlinear relationship curve between the temperature difference and the threshold. The fitting results are then corrected based on expert experience. For instance, when the diurnal temperature difference exceeds 15℃, expert experience suggests an upper limit of 1.3 for the threshold adjustment coefficient to avoid excessive threshold adjustment leading to missed alarms. Furthermore, the dynamic alarm threshold is calculated as follows: Dynamic alarm threshold = Initial alarm threshold × Threshold adjustment coefficient, where the initial alarm threshold is the optimal alarm threshold of the system under standard ambient temperature (e.g., a diurnal temperature difference of 5℃).

[0052] For example, a fire-fighting water system has a preset temperature difference threshold of 10℃, an initial alarm threshold for water pressure fluctuation of 0.1MPa, and an initial alarm threshold for water level fluctuation of 0.2m. The system detects a current diurnal temperature difference of 13℃, exceeding the preset temperature difference threshold by 3℃. By performing a cubic polynomial fit on historical data from the past 6 months, the adjustment coefficients for the water pressure threshold and water level threshold corresponding to this temperature difference are determined to be 1.15 and 1.2, respectively. Based on expert experience, these adjustment coefficients are confirmed to be within a reasonable range, requiring no correction. Therefore, the dynamic alarm thresholds corresponding to the current water pressure fluctuation are determined to be 0.1MPa × 1.15 = 0.115MPa, and the dynamic alarm thresholds corresponding to the water level fluctuation are determined to be 0.2m × 1.2 = 0.24m. Using this as an example, other alarm thresholds requiring dynamic adjustment will not be elaborated upon here.

[0053] Step S33: When the vibration intensity of the pipeline is higher than the preset value, the threshold adjustment coefficient is determined by combining historical data fitting and expert experience, and the sensitivity threshold for abnormal judgment is adjusted to form a dynamically updated alarm threshold system to support real-time adaptation of the threshold to environmental changes.

[0054] It should be noted that the preset setting value refers to a critical value (e.g., 0.2g) pre-set based on factors such as pipe material, installation process, and service life, used to determine whether pipe vibration intensity will interfere with anomaly detection. When the actual vibration intensity is higher than this value, it indicates that pipe vibration may cause fluctuations in sensor data, and the sensitivity of anomaly detection needs to be adjusted. The anomaly detection sensitivity threshold refers to a threshold used to measure the system's sensitivity to abnormal events. The higher the sensitivity threshold, the higher the system's tolerance for data fluctuations and the less likely it is to trigger an alarm; conversely, the lower the sensitivity, the easier it is to trigger an alarm. The dynamically updated alarm threshold system refers to a complete set of alarm thresholds driven by environmental factors such as diurnal temperature difference and pipe vibration intensity, which can be adjusted in real time and work together, covering multiple dimensions of judgment standards such as water pressure, water level, and vibration. Real-time adaptation of thresholds to environmental changes means that the alarm thresholds can be adjusted in a short time (e.g., within seconds) according to the dynamic changes in environmental parameters to ensure consistency with the current environmental state.

[0055] The purpose of this step is to address situations where pipeline vibration intensity exceeds the standard by adjusting the sensitivity of anomaly detection to avoid false alarms caused by vibration interference. At the same time, it complements the temperature difference adaptation logic in step S32, constructing a comprehensive dynamic threshold system to improve the accuracy of anomaly detection in complex environments.

[0056] Specifically, in one possible implementation, the preset setting value can be set in stages according to the specific conditions of the pipeline. For example, the preset setting value for newly laid steel pipes is 0.15g, and the preset setting value for cast iron pipes that have been in operation for more than 5 years is 0.25g. The threshold adjustment coefficient is determined by using a method of "primarily historical data fitting and secondarily expert experience correction". Historical data fitting is used to determine the optimal adjustment coefficient by analyzing the relationship between sensitivity threshold adjustment and false alarm rate changes in past scenarios where pipeline vibration intensity exceeded the standard. Expert experience is mainly used to handle extreme vibration scenarios (such as a sudden doubling of vibration intensity), in which case the emergency adjustment coefficient preset by the expert (such as 1.5) is directly used. The dynamically updated alarm threshold system is uniformly managed through a cloud platform. The edge controller receives environmental parameters in real time and calculates the adjusted threshold, while uploading the adjustment record to the cloud for filing, supporting subsequent traceability and optimization.

[0057] For example, the preset vibration intensity of an old fire-fighting pipeline network is 0.25g, and the initial sensitivity threshold for anomaly detection is 0.8 (the higher the value, the lower the sensitivity). The system detects a current pipeline vibration intensity of 0.3g, which is higher than the preset value. Through fitting analysis of historical data, the threshold adjustment coefficient corresponding to this vibration intensity is found to be 1.2. Combining expert experience and considering that the pipeline has been in operation for a long time and the vibration decay is slow, the adjustment coefficient is corrected to 1.3. Finally, the sensitivity threshold for anomaly detection is adjusted to 0.8 × 1.3 = 1.04. At this point, the system's tolerance for data fluctuations caused by pipeline vibration is improved, avoiding misjudging slight fluctuations in water pressure caused by vibration as anomalies. At the same time, this adjustment result is synchronized in real time to the dynamic alarm threshold system, working in conjunction with the dynamic thresholds for water pressure and water level.

[0058] This embodiment dynamically adjusts the threshold in real time according to the environment to adapt to different operating scenarios, reduce the impact of environmental changes on alarm accuracy, avoid misjudgments caused by a single fixed threshold, and make alarms more accurate. At the same time, the adjustment coefficient is determined based on historical data and expert experience to ensure that the threshold adjustment is scientific and reasonable, guarantee the effectiveness of early warning, thereby reducing interference from invalid alarms, allowing operation and maintenance personnel to focus on real anomalies, improving emergency response efficiency, and coping with common environmental fluctuations such as day-night temperature differences and pipeline vibration, meeting the needs of complex operating environments of fire protection and water conservancy systems.

[0059] In one feasible implementation, the step of determining whether there are real abnormal events in the fire-fighting water conservancy system based on the real-time operating data, the real-time environmental parameters, and the dynamic alarm threshold through multi-source data cross-validation rules includes: Step S41: Calculate the water pressure deviation rate and water level drop rate based on the real-time operating data, and calculate the risk index using the risk index formula by combining the temperature fluctuation amplitude and equipment anomaly factor in the real-time environmental parameters. It should be noted that the water pressure deviation rate refers to the degree of deviation between the real-time water pressure value and the historical average. The calculation formula is (|real-time water pressure - historical average|) / historical average. A larger value indicates a more significant water pressure anomaly. The water level drop rate refers to the magnitude of the water level drop per unit time (unit: m / min), used to reflect the urgency of water level changes. The equipment anomaly factor refers to the equipment health score output by an AI model (such as random forest), used to quantify the current health status of the equipment (value range 0-1, the closer to 1, the higher the risk of equipment anomaly). The risk index refers to the result calculated by this formula, used to objectively quantify the overall anomaly level of the system. The risk index formula is a mathematical expression that comprehensively calculates the degree of system anomaly by considering multiple dimensions of parameters, specifically: R=α1 Pn+α2 Ld+α3 Tv+α4 Ef α1 to α4 are weight parameters, which are obtained by fitting historical data and adjusting with expert experience.

[0060] Specifically, in one possible implementation, the historical mean is calculated using water pressure / water level data from the same period over the past 7 days as a sample, and is obtained using a moving average method to ensure that the mean reflects the parameter level under normal operating conditions; the initial values ​​of the weight parameters α1~α4 are obtained by fitting historical alarm data and equipment fault records, where α1=0.4, α2=0.3, α3=0.1, and α4=0.2, and are dynamically optimized and adjusted every 30 days based on new data; the equipment anomaly factor is calculated in real time by a random forest model deployed on the edge controller, and the model input includes features such as equipment runtime, number of historical faults, and real-time operating parameters.

[0061] For example, the system obtains a real-time water pressure value of 0.6 MPa, and the historical average water pressure for the same period over the past 7 days is 0.8 MPa. The calculated water pressure deviation rate is Pn = |0.6 - 0.8| / 0.8 = 25%. The real-time water level drops from 3.5 m to 2.9 m within 10 minutes, with a water level drop rate Ld = (3.5 - 2.9) / 10 = 0.06 m / min. The real-time environmental parameters show a temperature fluctuation amplitude Tv = 0.3℃ / min. The equipment anomaly factor Ef = 0.75 is calculated using a random forest model. Given the weight parameters α1 = 0.4, α2 = 0.3, α3 = 0.1, and α4 = 0.2, substituting them into the risk index formula yields R = 0.4 × 25% + 0.3 × 0.06 + 0.1 × 0.3 + 0.2 × 0.75 = 0.1 + 0.018 + 0.03 + 0.15 = 0.298.

[0062] Step S42: When the risk index exceeds the dynamic alarm threshold, multi-source data cross-validation is triggered to determine the authenticity of the abnormal event through multi-source data cross-validation rules. The cross-validation rules are to detect whether two or more of the following features exist simultaneously: sudden drop in water pressure, drop in water level, sudden change in flow rate, and increase in vibration signal. It should be noted that multi-source data cross-validation refers to a verification method that checks whether multiple abnormal features from different dimensions exist simultaneously when the risk index exceeds the limit, in order to determine the authenticity of the abnormal event. The cross-validation rules refer to the specific criteria for determining whether an abnormal event is real, namely, detecting whether two or more of the following features are simultaneously present: sudden drop in water pressure, drop in water level, sudden change in flow rate, and increase in vibration signal. A sudden drop in water pressure refers to a water pressure deviation rate exceeding the dynamic threshold δp; a drop in water level refers to a water level drop rate exceeding the dynamic threshold δl; a sudden change in flow rate refers to a change in pipe flow velocity exceeding the dynamic threshold δv; and an increase in vibration signal refers to a pipe vibration intensity exceeding a preset critical value. The authenticity of the abnormal event refers to whether the abnormal signal is caused by an actual system fault (such as a pipe rupture or pump failure), rather than a false anomaly caused by environmental interference (such as temperature fluctuations or instantaneous vibration).

[0063] The purpose of this step is to filter out false alarms caused by single parameter anomalies through joint verification of multi-dimensional features, improve the accuracy of anomaly judgment, solve the problems of traditional single threshold alarms being easily affected by environmental interference and having a high false alarm rate, and ensure that the warning only applies to real faults.

[0064] Specifically, in one possible implementation, the dynamic thresholds δp, δl, and δv are linked to the dynamic alarm thresholds determined in steps S31-S33. For example, δp is the critical value of the water pressure deviation rate corresponding to the dynamically adjusted water pressure fluctuation threshold, and δl is the critical value of the water level drop rate corresponding to the dynamically adjusted water level fluctuation threshold. Multi-source data cross-validation is performed locally by the edge controller, and the temporal consistency of different features is verified based on the "spatiotemporal integration" data synchronization to ensure that multiple detected abnormal features occur in the same time period and the same area. If a certain feature data is temporarily missing (such as a flow sensor failure), the system automatically adjusts the verification rules to allow the detection of two or more anomalies in the remaining three features.

[0065] For example, the system determines the dynamic alarm threshold Rth=0.25, and the risk index R=0.298 calculated in step S41 exceeds the dynamic alarm threshold, triggering multi-source data cross-validation. The system extracts real-time data from the same time period and area: water pressure deviation rate Pn=25%>δp=20% (satisfying the water pressure drop characteristic), water level drop rate Ld=0.06m / min>δl=0.05m / min (satisfying the water level drop characteristic), pipeline flow velocity change Vs=0.1m / s<δv=0.2m / s (not satisfying the flow change characteristic), and pipeline vibration intensity=0.18g<preset critical value0.2g (not satisfying the vibration signal increase characteristic). According to the cross-validation rules, two characteristics, water pressure drop and water level drop, have been detected, meeting the judgment criterion of "two or more".

[0066] Step S43: If the judgment result is true, then it is determined that a real abnormal event exists.

[0067] It should be noted that a true result means that, after cross-validation of multi-source data, two or more of the following characteristics are confirmed simultaneously: sudden drop in water pressure, drop in water level, sudden change in flow rate, and increase in vibration signal. A genuine anomaly refers to a failure event that actually occurs in the fire-fighting water system and may affect the normal operation of the system or cause safety risks, such as pipe rupture, network leakage, or pump failure, distinguishing it from false anomalies caused by environmental interference or instantaneous sensor errors.

[0068] The core function of this step is to determine, based on the results of cross-validation, whether there are any real faults in the system that require intervention, ensuring that only real faults trigger subsequent response processes, avoiding invalid warnings from consuming maintenance resources, reducing the risk of missed reports, and ensuring the safe operation of the system.

[0069] Specifically, in one possible implementation, when the judgment result is true, the system automatically records the key information of the abnormal event, including the occurrence time, latitude and longitude coordinates, triggered abnormal features, risk index value, etc., and uploads it to the cloud platform for storage and filing, for subsequent equipment maintenance traceability and model optimization; if the judgment result is false (only one or zero abnormal features are detected), it is determined to be a false alarm, the system does not trigger subsequent response, only records the false alarm event and the cause, and provides data support for dynamic threshold adjustment and model optimization.

[0070] For example, the cross-validation results in step S42 show that both the sudden drop in water pressure and the drop in water level are satisfied, indicating a true anomaly. The system then determines that a real anomaly has occurred. At this point, the system automatically records the occurrence time of the anomaly as "2024-06-10 14:25:30", the latitude and longitude coordinates as (116.42°E, 39.91°N), the triggering characteristics as "sudden drop in water pressure + drop in water level", and the risk index as 0.298, and uploads this information to the cloud platform. Simultaneously, the system triggers subsequent processes to predict the water level change trend using a time series model, determine the cause of the anomaly, and generate a remediation plan.

[0071] This embodiment comprehensively covers potential abnormal scenarios of the system by integrating operating condition data and environmental parameters, and calculates a risk index through multi-dimensional parameters to objectively reflect the degree of system abnormality, avoid the one-sidedness of judgment by a single indicator, filter false anomalies caused by environmental interference, improve the accuracy of anomaly identification, and reduce the risk of missed reports.

[0072] In one feasible implementation, the step of predicting the water level change trend of the fire-fighting water system using a time series model, and combining the real-time operating data and the real-time environmental parameters to determine the causes of anomalies in the fire-fighting water system, in order to generate an anomaly repair plan and an anomaly early warning, includes: Step S51: Use a time series model to analyze the acquired historical water level data, water pump operating status and external ambient temperature to predict the water level change trend. It should be noted that time series models refer to mathematical models that, based on the temporal sequence characteristics of data, mine patterns of data change and predict trends. This step specifically uses the ARIMA time series model, which has accurate predictive capabilities for linear time series data, with prediction errors controllable within 3%. Historical water level data refers to the set of water level values ​​recorded by facilities such as water storage tanks and fire pools in the fire-fighting water system over a past period (e.g., the past 7 days, 30 days). It includes water level information at different points in time and is the core foundational data for predicting future trends. Pump operating status refers to the working status of fire pumps (including main pumps and standby pumps), such as running, stopped, or shut down due to malfunction. Changes in their status directly affect the rise and fall of water levels. External ambient temperature refers to the real-time temperature data of the environment in which the fire-fighting water system is located. Temperature changes affect the physical properties of water (e.g., density, volume), thus indirectly affecting water level monitoring results. Water level change trend refers to the predicted direction and rate of change of water levels over a future time period (e.g., the next 2 hours) after analysis using the time series model. By integrating historical data with real-time influencing factors, the system can predict water level changes in advance, solving the problem that traditional systems cannot predict water level changes and can only respond passively, thus enhancing the system's proactive intervention capabilities.

[0073] Specifically, in one possible implementation, historical water level data is obtained from standardized data preprocessed by an edge controller and stored on a cloud platform. The data sampling period is consistent with that of real-time data (1s~5s) to ensure data continuity and consistency. The parameters of the time series model are optimized and determined through a grid search algorithm. In the ARIMA model, p (number of autoregressive terms), d (number of differencing terms), and q (number of moving average terms) are set to 2, 1, and 2, respectively. The model is deployed on a cloud platform and uses the TensorFlow framework for parallel computation to improve prediction efficiency and ensure that prediction results are output within 10 seconds. For example, the system retrieves historical water level data (604,800 records, sampling period 1s) for a water storage tank over the past 7 days from the cloud platform. At the same time, it obtains the current pump operating status as "main pump running, standby pump on standby," the external ambient temperature as 26℃, and the fluctuation range as ±1℃ within the past hour. These data were input into the ARIMA time series model, and the model analysis yielded the water level change trend of the storage tank in the next 2 hours: the water level dropped slowly at a rate of 0.02 m / min in the first 30 minutes, and after 30 minutes, due to the stable water pump supply efficiency, the water level drop rate dropped to 0.005 m / min. Finally, the water level was predicted to be 3.2 m after 2 hours, with a prediction error of 2.5%.

[0074] Step S52: The real-time water pressure value and pipeline flow velocity in the real-time environmental parameters are detected by the isolated forest algorithm, and the leakage probability score of the sudden water pressure drop event in the fire water conservancy system is output. It should be noted that the Isolation Forest algorithm is a density-based anomaly detection algorithm. Its core idea is to randomly partition the data space to distinguish between anomalous data (isolated points) and normal data. It features fast detection speed and strong adaptability to high-dimensional data, making it suitable for detecting sudden anomalies. Real-time water pressure refers to the current pressure value (unit: MPa) within the fire protection pipe network, collected in real time by a water pressure sensor (using a MEMS pressure chip, accuracy ±0.5% FS). Pipe flow velocity refers to the flow velocity of water within the fire protection pipe network (unit: m / s), calculated based on water pressure changes and pipe network parameters. A sudden drop in water pressure refers to an abnormal situation where the water pressure value drops significantly within a short period (e.g., within 10 seconds), usually caused by pipe ruptures, pipe network leaks, or other faults. The leakage probability score is a numerical value calculated using the Isolation Forest algorithm, quantifying the likelihood of a leakage fault in the fire protection water system. The value ranges from 0-100%, with a higher score indicating a higher leakage risk. The purpose of this step is to specifically detect sudden water pressure anomalies that may be caused by leaks, solve the problem that traditional systems have difficulty quickly identifying the cause of sudden water pressure drops, and improve the efficiency and accuracy of leak fault detection.

[0075] Specifically, in one possible implementation, the training samples for the Isolation Forest algorithm are derived from the past year's fire protection pipeline network operation data, including water pressure and flow velocity data under normal operating conditions and abnormal data during known leaks, totaling 100,000 valid samples. The algorithm is deployed in an edge controller, utilizing the AI ​​inference capabilities of the ARM Cortex-A72 chip to process water pressure and flow velocity data in real time, ensuring a leak probability score is output within 3 seconds. When the leak probability score exceeds 80%, the system marks it as a high-risk leak and prioritizes triggering subsequent verification processes. For example, the edge controller obtains a real-time water pressure value of 0.5 MPa for a node in the fire protection pipeline network (the historical average normal water pressure for this node is 0.8 MPa), and a pipe flow velocity of 0.4 m / s (the normal flow velocity range is 0.2-0.3 m / s). The two datasets were input into a pre-trained Isolation Forest algorithm. The algorithm analyzed the characteristics of the sudden drop in water pressure (0.3 MPa) and the abnormal increase in flow velocity, and matched them with the leakage fault characteristics in the training samples. Finally, it output a leakage probability score of 92%, indicating that there is a high probability of pipeline leakage in the area.

[0076] Step S53: Based on the water level change trend and the leakage probability score, analyze the causes of the anomaly using a Bayesian network. The causes of the anomaly include pipe rupture, water pump failure, and pipeline leakage. It should be noted that Bayesian networks are graphical models based on probabilistic reasoning. Nodes represent variables, and edges represent dependencies between variables. They can fuse multi-source data for probability calculations and are suitable for diagnosing the causes of anomalies in complex systems. Pipeline rupture refers to damage to fire-fighting pipelines caused by corrosion, external impact, etc.; pump failure refers to abnormal operation of fire pumps (main or standby pumps) due to mechanical failure, electrical problems, etc.; pipeline leakage refers to minor or moderate leakage at interfaces, valves, etc., in the fire-fighting pipeline network. The purpose of this step is to utilize the multi-source data fusion capabilities of Bayesian networks, combined with water level change trends and leakage probability scores, to accurately pinpoint the specific causes of abnormal events, avoiding the problem of blindly troubleshooting in traditional systems.

[0077] Specifically, in one possible implementation, the nodes of the Bayesian network include "water level change trend (rising / stable / falling)," "leakage probability score (low / medium / high)," and "anomaly cause (pipe rupture / pump failure / network leakage)." The conditional probability table between nodes is constructed based on historical fault data and expert experience. The system first converts the prediction results of step S51 and the scoring results of step S52 into network inputs (e.g., a falling water level trend corresponds to the "falling" node state, and a leakage probability score of 92% corresponds to the "high" node state). Then, it calculates the posterior probability of each anomaly cause through Bayesian inference and finally selects the cause with the highest posterior probability as the final diagnosis result. For example, step S51 predicts that the water level change trend is "rapidly falling," and step S52 outputs a leakage probability score of 92% (high risk). Substituting these two input data into the Bayesian network, the network calculates the posterior probability of each anomaly cause: the posterior probability of pipe rupture is 85%, the posterior probability of pump failure is 10%, and the posterior probability of network leakage is 5%. The system ultimately determined that the cause of the abnormal event was a pipe rupture.

[0078] Step S54: Based on the cause of the anomaly, generate an anomaly repair plan and push an anomaly warning according to the anomaly level.

[0079] It should be noted that the anomaly level refers to a classification based on the severity of the anomaly's cause, the scope of its impact on system operation, and the urgency (e.g., general, moderate, severe). Pipeline rupture typically corresponds to a severe level, pump failure to a moderate level, and network leakage to a general level. Anomaly warning refers to a mechanism that pushes relevant information about the anomaly (including its cause, level, scope of impact, and handling suggestions) to relevant maintenance personnel or the management platform according to a preset method. The purpose of this step is to provide customized repair solutions based on accurately identified anomaly causes, while simultaneously pushing warnings according to levels to ensure that maintenance personnel can respond quickly and handle issues accurately. This addresses the problems of blind and inefficient response in traditional system fault handling, improving system fault recovery speed and security capabilities.

[0080] Specifically, in one possible implementation, the anomaly repair plan is automatically constructed by the cloud platform based on a preset fault handling rule base and expert experience base. The rule base contains information such as standard handling procedures, required tools, and personnel configurations corresponding to various anomaly causes. The anomaly level is determined by combining the scope of the fault impact (e.g., affecting a single node, multiple nodes, or an entire area) and the urgency (e.g., whether it leads to system paralysis or whether there are security risks). Anomaly warnings are pushed through various methods, including SMS, fire monitoring platform pop-ups, maintenance APP push notifications, and audible and visual alarms. Severe anomalies trigger all push notification methods, moderate anomalies trigger platform pop-ups, APP push notifications, and SMS, and general anomalies trigger only APP push notifications. For example, step S53 determines that the anomaly cause is "pipeline rupture," corresponding to an anomaly level of "severe." The cloud platform automatically generates an anomaly repair plan: 1. Immediately start two backup water pumps to maintain the basic pressure of the pipeline network; 2. Close the upstream and downstream valves of the pipeline rupture area to isolate the faulty pipe section; 3. Dispatch the maintenance team with pipeline repair tools and leak-sealing materials to the anomaly location (latitude and longitude coordinates 116.45°E, 39.93°N); 4. After the emergency repair is completed, gradually open the valves to restore normal water supply to the pipeline network and monitor water pressure and water level data. Simultaneously, the system pushes anomaly warnings according to severity levels: sending an SMS to the maintenance manager, triggering an audible and visual alarm at the fire monitoring center, and displaying pop-up warnings containing the cause of the anomaly, repair plan, and location information on the maintenance APP and monitoring platform.

[0081] This embodiment analyzes multi-dimensional data through time series models to anticipate water level changes and allow time for responding to anomalies. It also uses the isolated forest algorithm to specifically detect sudden drops in water pressure, quantify leakage probabilities, and quickly locate potential leakage risks. By fusing multi-source results through Bayesian networks, it accurately distinguishes the causes of pipe ruptures, pump failures, etc., avoiding blind handling and shifting from "post-event processing" to "advance prediction and precise handling," thereby strengthening the safety assurance capabilities of fire protection and water conservancy systems.

[0082] In one feasible implementation, after predicting the water level change trend of the fire-fighting water system using a time series model, and combining the real-time operating data and the real-time environmental parameters to determine the cause of the anomaly in the fire-fighting water system, in order to generate an anomaly repair plan and an anomaly early warning, the method further includes: Step S61: Based on the historical operation data, past alarm records and equipment maintenance logs of the fire-fighting water system, a time series prediction network structure is used to construct an equipment health model; It should be noted that historical operational data of the fire protection and water conservancy system refers to the collection of various data generated during the normal or abnormal operation of the system over a period of time (e.g., the past year), including water pressure values, water level values, pipeline flow velocity, equipment operating time, energy consumption data, etc. This data is the core foundational data reflecting the system's operational patterns and changes in equipment status. Past alarm records refer to relevant records of alarms triggered in the system's history, including alarm time, alarm location, alarm type (e.g., abnormal water pressure alarm, low water level alarm), alarm cause, and handling results. This data reflects high-incidence scenarios and patterns of equipment anomalies. Equipment maintenance logs refer to records of inspection, maintenance, and repair performed on system equipment by operations and maintenance personnel. This includes maintenance time, maintenance objects, maintenance content (e.g., replacing parts, cleaning equipment, adjusting parameters), maintenance results, and equipment health status assessments. This data is an important basis for labeling equipment health status. The time series prediction network structure refers to a neural network architecture capable of processing time-dimensional data, mining data temporal features, and making predictions. In this step, LSTM (Long Short-Term Memory) is specifically used, which has the ability to capture long-sequence data dependencies and is suitable for time-series prediction of equipment health status. Equipment health models are models trained on multi-source historical data that can predict equipment health, remaining lifespan, and pipeline leakage probability. Their core function is to predict potential risks to equipment and pipelines in advance. The purpose of this step is to integrate multi-dimensional historical data from the system and utilize a network structure adapted to time-series data to build an accurate equipment health model, thus addressing the problem that traditional systems lack risk prediction capabilities and can only handle issues after they occur.

[0083] Specifically, in one possible implementation, historical operating data, past alarm records, and equipment maintenance logs are all stored in a cloud platform database. After standardized preprocessing (such as missing value imputation, outlier removal, and data format standardization), the data is organized into a time-series dataset in chronological order. The time-series prediction network structure adopts an LSTM neural network. The input layer is a multi-dimensional feature vector [Pt, Lt, Tt, Et] (where Pt is historical water pressure data, Lt is historical water level data, Tt is historical ambient temperature data, and Et is historical equipment operating status data). The hidden layer contains two LSTM units (64 neurons per layer) and one fully connected layer. The output layer is an equipment health score (0~1, the closer to 1, the healthier the equipment). The model is built based on the TensorFlow framework.

[0084] For example, the system retrieves historical operational data (including 1 million time-series data points on water pressure and level) from a city's fire and water conservancy system over the past year from a cloud database, along with 500 past alarm records (200 abnormal water pressure alarms, 150 low water level alarms, and 150 equipment failure alarms) and 300 equipment maintenance logs (including 120 fire pump maintenance logs, 100 pipeline maintenance logs, and 80 sensor calibration logs). After preprocessing this data, it is divided into training and validation sets in a 7:3 ratio and input into a pre-built LSTM neural network for training. The final result is an equipment health model capable of outputting equipment health status, achieving a 92% accuracy rate in predicting health status on the validation set.

[0085] Step S62: An incremental learning strategy is adopted to conduct supervised training on the equipment health model. The model weights are dynamically updated according to the newly added operating data at fixed intervals. When the model prediction error is detected to continuously exceed the preset model error range, the model retraining process is automatically triggered to correct the equipment health model by combining the latest data and external environmental parameters. It's important to note that incremental learning refers to a learning method where the model is not retrained on all historical data during training. Instead, it dynamically updates the model parameters based solely on new data, using existing training results. Its advantages include saving computational resources, shortening training time, and allowing the model to adapt to changes in data distribution promptly. Supervised training refers to a training method that uses labeled training data (i.e., time-series data annotated with device health status) to backpropagate and optimize model parameters by calculating the error between the model's predictions and the true labels. A fixed period refers to a preset model weight update interval, specifically set to 24 hours in this step, ensuring the model can promptly absorb effective information from new data. New operational data refers to real-time operating data, environmental parameter data, alarm records, maintenance logs, etc., generated by the system within the new operational cycle after the last model update. Model weights refer to the parameters connecting neurons in each layer of a neural network; their values ​​determine the model's emphasis on input features. Dynamic updates to the weights allow the model to continuously optimize its predictive performance. The preset model error range refers to the pre-set critical value used to judge whether the model's prediction effect is acceptable (e.g., prediction error <5%). When the model's prediction error continues to exceed this range, it indicates that the model can no longer accurately adapt to the current data distribution and needs to be retrained.

[0086] Furthermore, the automatic model retraining process refers to the system automatically initiating a full-data retraining process when the model's prediction error is continuously exceeded (e.g., prediction error > 10% for three consecutive times) by monitoring the model's prediction error in real time, without requiring manual intervention. External environmental parameters refer to environmental factors that affect the equipment's operating status, including temperature, humidity, diurnal temperature range, and pipeline vibration intensity. Incorporating these into the retraining process improves the model's adaptability to environmental changes. The purpose of this step is to achieve continuous self-optimization of the model through incremental learning, while simultaneously addressing the model performance degradation problem through the automatic retraining mechanism, ensuring that the equipment health model maintains high prediction accuracy over the long term.

[0087] Specifically, in one possible implementation, the incremental learning process is as follows: At 0:00 every day, the system automatically extracts the newly added running data from the previous 24 hours, preprocesses the data, and combines it with a portion of the original training data of the model (randomly selecting 10% of historical data) to fine-tune and update the weights of the LSTM model. The loss function is still optimized using mean squared error (MSE). The preset model error range is determined to be <5% through 5-fold cross-validation. After each output of the prediction result, the system calculates the error between the predicted value and the actual equipment health status (based on maintenance log annotations). When the error is >10% for 3 consecutive times, the retraining process is automatically triggered. At this time, the model is retrained and the model parameters are updated using the full historical data (including newly added data) of the past 6 months and the latest external environment parameter data.

[0088] For example, the equipment health model undergoes incremental learning on a fixed cycle (24 hours). Every day at midnight, it extracts the previous day's newly added operational data (including 10,000 time-series operating condition data, 5 alarm records, and 3 maintenance logs), combines it with 10% of historical training data, and fine-tunes the model weights. After one month of incremental updates, the model prediction error stabilizes between 3% and 4%. On day 32, the system detected model prediction errors of 11%, 12%, and 13% three times consecutively, exceeding the preset model error range (<5%), and automatically triggers the retraining process. The system retrieves all historical data from the past 6 months (including 3 million time-series data) and the latest external environmental parameters (such as recent diurnal temperature range and pipeline vibration intensity data), retrains the LSTM model, and after training, the model prediction error drops to 4%, returning to the acceptable range.

[0089] Step S63: Using the trained equipment health model, predict the remaining lifespan of each key piece of equipment in the fire protection and water conservancy system and the probability of leakage in the pipeline network facilities, generate targeted maintenance plans and equipment maintenance work orders, and push them to relevant operation and maintenance personnel.

[0090] It should be noted that the trained equipment health model refers to an LSTM model that, after initial training, incremental learning updates, or retraining optimization, meets the preset requirement of prediction error (<5%) and possesses stable prediction performance. Critical equipment refers to equipment that plays a core supporting role in the normal operation of the fire protection and water conservancy system, including fire pumps (main pumps, standby pumps), makeup water pumps, valves, and sensors (water pressure sensors, water level sensors), etc. Remaining life refers to the remaining service life of critical equipment predicted by the model from the current time until the expected failure or malfunction (e.g., 3 months remaining, 1 year remaining), providing a time reference for equipment replacement or major overhaul. The leakage probability of the pipeline network refers to the probability predicted by the model that the fire protection pipeline network (including pipes, interfaces, valves, etc.) will leak within a future period (e.g., the next month), with a value ranging from 0-100%, and a higher score indicating a higher leakage risk. Targeted maintenance plans refer to maintenance schemes tailored to actual needs, based on the remaining lifespan of equipment and the probability of pipeline leaks. These plans include maintenance priorities, maintenance time windows, and maintenance content (such as preventative maintenance, parts replacement, and pipeline inspection), which differs from traditional, undifferentiated, and periodic maintenance plans.

[0091] Furthermore, an equipment maintenance work order refers to a standardized execution document that visualizes the maintenance plan. It includes information such as work order number, maintenance object (equipment name, location), maintenance task, required tools, responsible person, and completion deadline, facilitating execution by maintenance personnel. Relevant maintenance personnel refer to professionals responsible for the maintenance and troubleshooting of fire protection and water conservancy system equipment, including on-site maintenance engineers and technical management personnel. The purpose of this step is to accurately predict potential risks to equipment and pipelines through models, generate customized maintenance plans and executable work orders, and achieve a shift from "passive maintenance" to "proactive operation and maintenance," reducing the incidence of equipment failures and pipeline leaks, and improving the stability and reliability of system operation.

[0092] Specifically, in one possible implementation, when the model makes predictions, it takes current equipment operating data and recent environmental parameter data as input, and outputs the remaining lifespan of each key piece of equipment and the leakage probability of each area of ​​the pipeline network. The system classifies maintenance priorities based on the remaining lifespan (e.g., <6 months for high priority, 6-12 months for medium priority, >12 months for low priority) and leakage probability (e.g., >80% for high risk, 50%-80% for medium risk, <50% for low risk), with high-priority tasks being included in the maintenance plan first. Equipment maintenance work orders are automatically generated by the cloud platform and pushed to the corresponding maintenance personnel through the maintenance APP. At the same time, work order records are stored on the cloud platform to support progress tracking and result feedback. After the maintenance personnel complete the maintenance, they enter the maintenance results into the system as training data for the model's subsequent incremental learning.

[0093] For example, the trained equipment health model, inputting current fire pump operating data (2000 hours of operation, recent water pressure fluctuation of 0.05 MPa) and ambient temperature data (25°C), predicts the remaining lifespan of the fire pump to be 5 months (high priority); inputting recent operating data and vibration intensity data (0.2g) of a certain area's pipe network, predicts the leakage probability of that area's pipe network to be 85% (high risk). The system generates a targeted maintenance plan: 1. Conduct a comprehensive overhaul of the fire pump within one week, replacing aged bearings and seals; 2. Conduct ultrasonic testing on the high-risk area's pipe network within three days to identify and repair leaks. Simultaneously, two equipment maintenance work orders are generated, each labeled with a work order number, maintenance object (fire pump A, location: XX pump station; pipeline B, location: XX road fire pipeline), maintenance task, required tools (wrench, ultrasonic detector, sealing material), responsible persons (maintenance engineers Zhang and Li), and completion deadline (within 3 days and within 1 week), and pushed to Zhang and Li via the maintenance APP. The cloud platform tracks the progress of the work orders in real time.

[0094] This embodiment shifts from "post-event repair" to "proactive prediction," avoiding system paralysis caused by sudden equipment failures. Through incremental learning and automatic retraining mechanisms, it adapts to equipment aging and environmental changes, ensuring the long-term reliability of the model, generating targeted maintenance plans, avoiding blind inspections, reducing ineffective maintenance investment, and lowering maintenance costs. This allows for proactive handling of potential faults, reducing equipment failure rates and pipeline leakage risks, ensuring the continuous and stable operation of the fire protection and water conservancy system, and enabling precise delivery of maintenance work orders, clarifying maintenance priorities, and reducing maintenance response time and communication costs.

[0095] In one feasible implementation, after predicting the water level change trend of the fire-fighting water system using a time series model, and combining the real-time operating data and the real-time environmental parameters to determine the cause of the anomaly in the fire-fighting water system, in order to generate an anomaly repair plan and an anomaly early warning, the method further includes: Step S71: Obtain the geographic coordinate data uploaded by each terminal in the global Internet of Things network, and construct a digital topology map of fire protection facilities in combination with the actual layout of the pipeline network; It should be noted that each terminal refers to fire protection data transmission terminals, water pressure sensors, water level sensors, and other equipment deployed at key nodes such as fire protection pipe networks, water storage tanks, and pumping stations. These terminals all have embedded GPS modules, capable of generating location information including latitude and longitude coordinates. Geographic coordinate data refers to the latitude and longitude data (e.g., 116.4°E, 39.9°N) obtained by each terminal through its GPS module to identify the physical location of the equipment, possessing high-precision positioning capabilities. Actual pipe network layout refers to the actual laying route, pipe segment connection relationships, equipment installation locations, and other real engineering layout information of the fire protection pipe network, including the direction of main and branch pipes, node connection methods, and spatial distribution of water storage tanks and pumping stations. The digital topology map of fire protection facilities refers to a digital and visualized graphic model built on a cloud platform based on geographic coordinate data and the actual pipe network layout. It can intuitively present the spatial distribution of fire protection facilities, pipe segment relationships, and equipment connection logic, realizing visualized management of fire protection facilities across the entire area.

[0096] The purpose of this step is to break through the limitations of traditional fire protection systems, such as the dispersed locations of equipment and unclear relationships between them. By establishing spatial relationships between equipment and pipelines through digital topology maps, it solves the problems of poor linkage and insufficient overall coordination capabilities in traditional systems.

[0097] Specifically, in one possible implementation, each terminal uploads geographic coordinate data and operating condition data at fixed intervals of 1 to 5 seconds via the LoRaWAN protocol. After receiving the data, the cloud platform first calibrates the geographic coordinate data (correcting positioning deviations using GIS maps), then imports preset vector data of the actual pipeline layout. Through a topology algorithm, it automatically associates the correspondence between equipment and pipe segments, and between pipe segments and pump stations / storage tanks, constructing a digital topology map containing equipment identifiers, location information, pipe segment parameters, and associated paths. This topology map supports zooming, panning, and click-to-query operations. Maintenance personnel can view the location of any device and associated pipeline information in real time through the cloud platform. Simultaneously, the system supports dynamic updates to the topology map; when equipment is added, moved, or the pipeline is modified, the topology relationships are automatically updated synchronously.

[0098] For example, a city's fire protection and water conservancy system deployed 500 data transmission terminals, 600 water pressure sensors, and 200 water level sensors. All terminals upload geographic coordinate data in real time via GPS modules. After receiving this data, the cloud platform combines it with the actual layout vector map of the city's fire protection pipeline network (including the route of 300km of pipeline, the location information of 15 pumping stations and 20 water storage tanks). Using a topology algorithm, it associates each sensor with its corresponding pipe segment, and each pipe segment with its upstream and downstream associated pipe segments and pumping stations, constructing a digital topology map of the entire fire protection facility. This topology map clearly shows the distribution of 10 water pressure sensors on the XX Road fire protection main pipe, their connection paths with the XX pumping station, and the water supply range of the surrounding 3 water storage tanks. Clicking on any sensor icon allows users to view its latitude and longitude coordinates (e.g., 116.35°E, 39.88°N), the diameter of the associated pipe segment, material, and other information.

[0099] Step S72: When an abnormal event is detected in the equipment deployment area, the abnormal event is determined based on the digital topology map of the fire protection facilities to determine whether it is caused by the failure of the upstream pipe section or related equipment. The cross-regional linkage node corresponding to the equipment deployment area is matched, and a collaborative control command is sent to the edge controller of the cross-regional linkage node to complete the cross-regional abnormal correction collaborative operation. It should be noted that the equipment deployment area refers to the installation area of ​​fire protection data transmission terminals, sensors, and other equipment, typically corresponding to a specific pipe section of the fire protection network, the area around a water storage tank, or the coverage area of ​​a pumping station. An abnormal event refers to a genuine anomaly confirmed by the system through cross-verification of multi-source data, such as pipe rupture, network leakage, or pump failure. An upstream pipe section refers to a pipe section located upstream of the area where the abnormal event occurred in the direction of water flow in the fire protection network; its failure (such as leakage or blockage) may cause abnormal water pressure or level in the downstream area. Associated equipment refers to equipment that has a functional association or network connection with the equipment in the abnormal area, such as fire pumps supplying water to the network in the abnormal area, and associated valves. Cross-regional linkage nodes refer to cross-regional equipment or pipe section control nodes that have a network connection with the abnormal equipment deployment area and can participate in anomaly correction, such as water supply pumps in adjacent areas, cross-regional connection valves, and backup pumping stations. An edge controller refers to a core processing device deployed at the edge computing layer, based on an ARM Cortex-A72 chip, possessing protocol parsing and AI inference capabilities, and responsible for executing local collaborative control commands.

[0100] Furthermore, collaborative control commands refer to the operational instructions generated by the cloud platform based on topological relationships, used to schedule cross-regional collaborative nodes to collaboratively handle anomalies, such as starting a water supply pump, switching valve states, and adjusting the pump group operating frequency. Cross-regional anomaly correction collaborative operation refers to the process where multiple cross-regional collaborative nodes synchronously execute operations according to collaborative control commands to jointly correct anomalies and restore normal system operation. The purpose of this step is to quickly trace the root cause of anomalies using the relationships in the digital topology map, achieve rapid anomaly correction through cross-regional node collaborative operation, avoid the spread of local anomalies, solve the problems of isolated anomaly handling, delayed response, and poor linkage in traditional systems, and improve the system's overall collaborative handling capabilities.

[0101] Specifically, in one possible implementation, after detecting an abnormal event, the system first locates the abnormal area in the digital topology map. It then traverses the operating data of upstream pipe sections and related equipment (such as water pressure in upstream pipe sections and the operating status of related pump sets) through topological relationships to determine whether the abnormality was caused by an upstream fault (e.g., a sudden drop in water pressure in the upstream pipe section occurs simultaneously with the downstream abnormality, indicating an upstream fault). If a cross-regional impact is confirmed, the system automatically matches linked nodes (such as upstream pumping stations, adjacent area water supply pumps, and cross-regional isolation valves). The cloud platform generates targeted collaborative control commands (such as starting two adjacent area water supply pumps, closing the upstream faulty pipe section isolation valve, and adjusting the water supply frequency of the target area pump set), which are then sent to the edge controllers of each linked node via the 4G network. The edge controllers receive the commands and execute the operations in real time, while simultaneously feeding back the execution status to the cloud platform, ensuring the synchronicity and effectiveness of the collaborative operation.

[0102] For example, the system detects a sudden drop in water pressure in the XX Road fire protection pipeline area (equipment deployment area). After locating the area using the fire protection facility digital topology map, it traverses the operational data of its upstream pipeline section (3km upstream of the XX Road main pipeline) and finds that the water pressure in the upstream pipeline section drops simultaneously, determining that the abnormal event is caused by a leak in the upstream pipeline section. The system automatically matches cross-regional linkage nodes: the isolation valve of the upstream pipeline section, two make-up water pumps in the adjacent area, and the standby pump of the XX pumping station. The cloud platform generates collaborative control commands: 1. Close the isolation valve of the upstream pipeline section to isolate the faulty area; 2. Start the two make-up water pumps in the adjacent area to replenish water to the pipeline network in the abnormal area; 3. Start the standby pump of the XX pumping station to increase the overall water supply pressure of the pipeline network. After the commands are sent to the edge controllers of each linkage node, the edge controllers execute the operations synchronously within 30 seconds, completing the cross-regional abnormality correction collaborative operation, and the water pressure in the abnormal area gradually returns to normal.

[0103] Step S73: After the anomaly correction collaborative operation is completed, the repair status data of the fire-fighting water conservancy system after restoration is collected in real time, and an anomaly correction report is generated.

[0104] It should be noted that the repair status data refers to the real-time data collected by the system through the global IoT network after the anomaly correction operation, reflecting the recovery status of the fire protection and water conservancy system. This includes water pressure, water level, pipeline flow velocity, and equipment operating status in the abnormal area and related areas, and is the core data for evaluating the effectiveness of the anomaly correction. The anomaly correction report is a standardized report generated by the system based on anomaly event information, collaborative operation content, and repair status data. It includes basic anomaly information, handling process, repair effect, data comparison, and optimization suggestions. The purpose of this step is to verify the anomaly handling effect by collecting repair status data, forming a closed-loop management of "anomaly detection - collaborative handling - effect evaluation." At the same time, the report retains handling experience, providing a basis for subsequent optimization of linkage strategies and improvement of handling efficiency, solving the problem of lack of effect evaluation and data traceability after anomaly handling in traditional systems.

[0105] Specifically, in one possible implementation, after the collaborative operation for anomaly correction is completed, the system automatically switches to the repair condition data acquisition mode. It collects data such as water pressure, water level, and flow rate at a high frequency of 1 second using sensors in the anomaly area and related areas, continuously collecting data for 30 minutes to ensure data integrity and stability. After collection, the system analyzes and processes the data, comparing key indicators before, during, and after the anomaly (e.g., water pressure in the anomaly area recovering from 0.5 MPa to 0.8 MPa), and assesses whether the correction effect meets preset standards (e.g., water pressure recovering to within ±5% of the normal range). The anomaly correction report is automatically generated by the cloud platform, including details of the anomaly event (occurrence time, location, cause), details of collaborative operation instructions, execution status of linked nodes, comparison charts of repair condition data, assessment conclusions of the correction effect, and optimization suggestions (e.g., adjusting the response priority of a linked node). The report can be exported as a PDF and simultaneously pushed to the operation and maintenance management platform for relevant personnel to view.

[0106] For example, after the cross-regional anomaly correction collaborative operation in step S72 is completed, the system collects repair condition data at high frequency from 20 water pressure sensors and 10 water level sensors in the anomaly area and related areas for 30 minutes. Data shows that the water pressure in the anomaly area gradually recovered from 0.5MPa before treatment to 0.82MPa (normal range 0.75MPa~0.85MPa), the water level recovered from 2.8m to 3.5m, and the pipeline flow velocity stabilized at 0.3m / s. The system compares the data before the anomaly (water pressure 0.8MPa, water level 3.6m) with the data after repair, confirming that the correction effect is satisfactory. The cloud platform automatically generates an anomaly correction report, clearly recording the anomaly event as "leakage in the upstream pipe section of XX road." The collaborative operation includes closing the isolation valve, starting two water supply pumps and a standby pump. After repair, key indicators all returned to normal ranges, and optimization suggestions are proposed: "Regularly check the corrosion of the upstream pipe section and improve the response speed of the linkage valves." The report is pushed to the operation and maintenance management platform and archived.

[0107] This embodiment constructs a digital topology map of fire protection facilities, which intuitively presents the pipeline layout and equipment association, facilitating global control of the system status. Based on the topology relationship, it can quickly determine whether anomalies are caused by upstream pipe sections or related equipment, avoiding blind investigation. At the same time, it automatically matches linkage nodes and issues collaborative control commands to achieve rapid response and joint handling in multiple areas, shortening the anomaly correction time, improving cross-regional collaboration efficiency, thereby preventing the spread of local anomalies, reducing the impact on the overall operation of the fire protection and water conservancy system, breaking down regional barriers, solving the problem of poor linkage in traditional systems, and achieving efficient allocation of resources across the entire region.

[0108] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0109] This application also provides a safety early warning device for a fire protection and water conservancy system; please refer to... Figure 3 The fire-fighting water conservancy system safety early warning device includes: The acquisition module 31 is used to acquire real-time operating data and real-time environmental parameters of the fire protection and water conservancy system through the global Internet of Things network, and determine the dynamic alarm threshold based on the real-time environmental parameters. The judgment module 32 is used to determine whether there is a real abnormal event in the fire protection and water conservancy system based on the real-time operating data, the real-time environmental parameters and the dynamic alarm threshold, through multi-source data cross-validation rules. The early warning module 33 is used to predict the water level change trend of the fire-fighting water system through a time series model if the condition is met, and to determine the cause of the abnormality of the fire-fighting water system by combining the real-time operating data and the real-time environmental parameters, so as to generate an abnormality repair plan and an abnormality early warning.

[0110] The fire-fighting water conservancy system safety early warning device provided in this application, employing the fire-fighting water conservancy system safety early warning method in the above embodiments, can solve the technical problems in the background art. Compared with the prior art, the beneficial effects of the fire-fighting water conservancy system safety early warning device provided in this application are the same as the beneficial effects of the fire-fighting water conservancy system safety early warning method provided in the above embodiments, and other technical features in the fire-fighting water conservancy system safety early warning device are the same as the features disclosed in the methods of the above embodiments, and will not be repeated here.

[0111] This application provides a fire protection and water conservancy system safety early warning device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the fire protection and water conservancy system safety early warning method in the above embodiment 1.

[0112] The following is for reference. Figure 4 The diagram illustrates a structural schematic suitable for implementing the fire and water system safety early warning device in the embodiments of this application. The fire and water system safety early warning device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), vehicle terminals (e.g., vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 4 The fire protection and water conservancy system safety early warning device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0113] like Figure 4As shown, the fire and water system safety early warning equipment may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the fire and water system safety early warning equipment. The processing unit 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: input devices 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. Communication device 1009 allows the fire and water system safety early warning equipment to communicate wirelessly or wiredly with other equipment to exchange data. Although the figure shows fire and water system safety early warning equipment with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems may be implemented alternatively.

[0114] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0115] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0116] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the fire protection and water conservancy system safety early warning method provided by the above methods.

[0117] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0118] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0119] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A safety early warning method for a fire-fighting water conservancy system, characterized in that, include: The system acquires real-time operating data and environmental parameters of the fire protection and water conservancy system through a global Internet of Things network, and determines dynamic alarm thresholds based on the real-time environmental parameters. Based on the real-time operating data, the real-time environmental parameters, and the dynamic alarm threshold, the existence of a real abnormal event in the fire-fighting water conservancy system is determined by multi-source data cross-validation rules. If so, the water level change trend of the fire-fighting water system is predicted by time series model, and the causes of the anomaly of the fire-fighting water system are determined by combining the real-time operating data and the real-time environmental parameters, so as to generate an anomaly repair plan and an anomaly early warning.

2. The fire protection and water conservancy system safety early warning method as described in claim 1, characterized in that, The acquisition of real-time operating data and real-time environmental parameters of the fire protection and water conservancy system through a global Internet of Things network includes: The system adopts a full-domain IoT architecture, which collects raw operating data and raw environmental parameters through data transmission terminals, water pressure sensors and water level sensors deployed in fire protection pipe networks, water storage tanks and pumping stations. The full-domain IoT architecture includes a terminal acquisition layer, an edge computing layer and a cloud decision layer. The original operating condition data and the original environmental parameters are synchronized in time and space by the edge controller to generate the real-time operating condition data and the real-time environmental parameters. The real-time operating condition data includes water pressure value, water level value, pipeline flow velocity, and equipment operating status. The real-time environmental parameters include temperature, humidity, and pipeline vibration intensity.

3. The fire protection and water conservancy system safety early warning method as described in claim 1, characterized in that, The process of determining the dynamic alarm threshold based on the real-time environmental parameters includes: Extract the diurnal temperature difference and pipeline vibration intensity from the real-time environmental parameters, and establish a mapping relationship between environmental impact factors and dynamic alarm thresholds; When the diurnal temperature difference is detected to exceed the preset temperature difference threshold, the threshold adjustment coefficient is determined by combining historical data fitting and expert experience, and the dynamic alarm threshold corresponding to water pressure and water level fluctuations is determined. When the vibration intensity of the pipeline exceeds the preset value, the threshold adjustment coefficient is determined by combining historical data fitting and expert experience, and the sensitivity threshold for anomaly judgment is adjusted to form a dynamically updated alarm threshold system that supports real-time adaptation of thresholds to environmental changes.

4. The fire protection and water conservancy system safety early warning method as described in claim 1, characterized in that, The step of determining whether there are real abnormal events in the fire protection and water conservancy system based on the real-time operating data, the real-time environmental parameters, and the dynamic alarm threshold, through multi-source data cross-validation rules, includes: Based on the real-time operating data, the water pressure deviation rate and water level drop rate are calculated. Combined with the temperature fluctuation amplitude and equipment malfunction factors in the real-time environmental parameters, the risk index is calculated using the risk index formula. When the risk index exceeds the dynamic alarm threshold, multi-source data cross-validation is triggered to determine the authenticity of the abnormal event through multi-source data cross-validation rules. The cross-validation rules are to detect whether two or more of the following features exist simultaneously: sudden drop in water pressure, drop in water level, sudden change in flow rate, and increase in vibration signal. If the judgment result is true, then a real abnormal event is confirmed to exist.

5. The fire protection and water conservancy system safety early warning method as described in claim 1, characterized in that, The method involves predicting the water level change trend of the fire-fighting water system using a time series model, and combining this with real-time operating data and real-time environmental parameters to determine the causes of anomalies in the fire-fighting water system, thereby generating anomaly repair plans and anomaly early warnings, including: A time series model was used to analyze the acquired historical water level data, water pump operating status, and external ambient temperature to predict the trend of water level changes. The isolated forest algorithm is used to detect the real-time water pressure value and pipeline flow velocity in the real-time environmental parameters, and outputs the leakage probability score of the sudden water pressure drop event in the fire water conservancy system. Based on the water level change trend and the leakage probability score, the causes of anomalies are analyzed using Bayesian networks. The causes of anomalies include pipe rupture, water pump failure, and pipeline leakage. Based on the causes of the anomalies, an anomaly repair plan is generated, and anomaly warnings are pushed out according to the anomaly level.

6. The fire protection and water conservancy system safety early warning method as described in claim 1, characterized in that, After predicting the water level change trend of the fire-fighting water system using a time series model, and combining the real-time operating data and real-time environmental parameters to determine the causes of anomalies in the fire-fighting water system, thereby generating an anomaly repair plan and an anomaly early warning, the method further includes: Based on the historical operation data, past alarm records, and equipment maintenance logs of the fire-fighting water conservancy system, a time series prediction network structure is used to construct an equipment health model. An incremental learning strategy is adopted to conduct supervised training on the equipment health model. The model weights are dynamically updated according to the newly added operating data at fixed intervals. When the model prediction error is detected to continuously exceed the preset model error range, the model retraining process is automatically triggered to correct the equipment health model by combining the latest data and external environmental parameters. The trained equipment health model predicts the remaining lifespan of key equipment in the fire protection and water conservancy system and the probability of leakage in the pipeline network, generates targeted maintenance plans and equipment maintenance work orders, and pushes them to relevant operation and maintenance personnel.

7. The fire protection and water conservancy system safety early warning method as described in claim 1, characterized in that, After predicting the water level change trend of the fire-fighting water system using a time series model, and combining the real-time operating data and real-time environmental parameters to determine the causes of anomalies in the fire-fighting water system, thereby generating an anomaly repair plan and an anomaly early warning, the method further includes: Obtain the geographic coordinate data uploaded by each terminal in the global Internet of Things network, and construct a digital topology map of fire protection facilities in combination with the actual layout of the pipeline network; When an abnormal event is detected in the equipment deployment area, the system determines whether the abnormal event is caused by a fault in the upstream pipe section or related equipment based on the digital topology map of the fire protection facilities. The system then matches the cross-regional linkage node corresponding to the equipment deployment area and sends a collaborative control command to the edge controller of the cross-regional linkage node to complete the cross-regional abnormal correction collaborative operation. After the anomaly correction collaborative operation is completed, the repair status data of the fire-fighting water conservancy system after restoration is collected in real time, and an anomaly correction report is generated.

8. A safety early warning device for a fire-fighting water conservancy system, characterized in that, include: The acquisition module is used to acquire real-time operating data and real-time environmental parameters of the fire protection and water conservancy system through the global Internet of Things network, and determine the dynamic alarm threshold based on the real-time environmental parameters. The judgment module is used to determine whether there is a real abnormal event in the fire-fighting water conservancy system based on the real-time operating data, the real-time environmental parameters and the dynamic alarm threshold, and through multi-source data cross-validation rules. The early warning module is used to predict the water level change trend of the fire-fighting water system through a time series model if the condition is met, and to determine the cause of the abnormality of the fire-fighting water system by combining the real-time operating data and the real-time environmental parameters, so as to generate an abnormality repair plan and an abnormality early warning.

9. A safety early warning device for a fire-fighting and water conservancy system, characterized in that, The fire and water conservancy system safety early warning device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the fire and water conservancy system safety early warning method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the fire protection and water conservancy system safety early warning method as described in any one of claims 1 to 7.

Images