Water conservancy safety detection method based on AI edge calculation

Abstract

The invention provides a water conservancy safety detection method based on AI edge calculation, and the method comprises the steps: deploying a sensor in a water conservancy system according to the installation and coding of the sensor, coding the sensor, and recording the geographic position information and measurement physical quantity of the sensor; setting measurement parameters, and setting a measurement interval value and a threshold value of each sensor; data linkage setting: defining data linkage selection data, and triggering pump IO output when the data reaches a specific interval threshold value; the method comprises the following steps: monitoring sensor data in real time, if the data is abnormal, starting a video AI to check, comparing whether the result is consistent, if the video AI confirms that the sensor data is abnormal, intercepting a real-time video as sample data, sending the sample data to a cloud server, triggering a corresponding alarm according to the abnormal condition, and notifying related personnel to process. The accuracy and timeliness of abnormal condition detection are improved, and a powerful guarantee is provided for safe operation of a water conservancy system.

Description

Technical Field

[0001] The present invention belongs to the field of computer technology, and specifically relates to a water conservancy safety detection method based on AI edge computing. Background Art

[0002] Traditional water safety inspection methods often rely on manual inspections and simple sensor monitoring. Manual inspections are limited by time, personnel experience, and work intensity, and may result in problems such as untimely inspections and missed inspection points. For example, manual inspections are generally carried out according to a fixed schedule. If an emergency occurs between two inspections, such as a sudden rupture of a pipeline or a sharp rise in water levels, it is difficult to detect it in time.

[0003] When relying solely on sensor monitoring, if the sensor fails or is disturbed, the lack of other auxiliary verification methods may lead to erroneous monitoring results. Moreover, a single type of sensor can only monitor specific physical quantities, and cannot fully reflect the actual status of the water conservancy system, and cannot use advanced artificial intelligence technology to perform intelligent diagnosis and prediction of the water conservancy system. It is difficult to identify complex abnormal patterns and potential safety hazards through learning historical data and real-time data, and it is difficult to achieve intelligent safety management. Summary of the invention

[0004] The present invention solves the technical problem of not being able to perform adaptive adjustments according to the actual operation conditions and environmental changes of the water conservancy system by proposing a water conservancy safety detection method based on AI edge computing.

[0005] One of the embodiments of the present invention provides a water conservancy safety detection method based on AI edge computing, including the following steps: sensor installation and encoding, deploying the sensor in the water conservancy system, encoding the sensor, recording the geographical location information and measured physical quantity of the sensor; setting measurement parameters, setting the measurement interval value and threshold of each sensor; data linkage setting, including defining data linkage selection data, triggering the pump IO output when the data reaches a specific interval threshold; real-time monitoring data of the sensor according to the AI ​​algorithm linkage, when data abnormality occurs, starting the video AI to check whether the comparison results are consistent, if the video AI confirms that the sensor data is abnormal, intercepting the real-time video as sample data and sending it to the cloud server, triggering the corresponding alarm according to the abnormal situation, and notifying the operator to handle it; according to the input pump control instruction, combined with the sensor detection result and the video AI judgment result, confirm the working status of the pump, if the instruction execution result is inconsistent with the expectation, according to the linkage setting, trigger the fault code alarm, notify the operator to handle it, and send the information to the cloud operation and maintenance application background.

[0006] In one embodiment, the sensor installation and encoding process steps also include: the sensor includes a water level sensor, a water pump on / off sensor and a water gauge identification sensor, the water level sensor is arranged in the physical field of the water conservancy system, and is used to monitor the water level height in real time; the water pump on / off sensor is arranged in the water pump, and is used to monitor the working status of the water pump, including opening, closing, and running time data; the water gauge sensor is used to accurately read the water level scale on the water gauge.

[0007] In one of the embodiments, the step of setting the measurement parameters includes: setting the sensor database, and setting the measurement interval value and threshold value of the sensor according to the historical data model analysis.

[0008] In one of the embodiments, the data linkage setting step also includes: setting a PLC control system to read sensor measurement data in real time; determining whether the sensor data reaches or exceeds the measurement interval value and threshold; if the above steps meet the conditions, triggering the pump IO output to control the start and stop of the water pump.

[0009] In one of the embodiments, the real-time monitoring of sensor data also includes: setting up a video AI system, and establishing an association mechanism between the video AI system and the sensor data based on AI edge computing; setting a comparison mechanism, and comparing the real-time sensor data with the video monitoring data according to the comparison mechanism, and judging whether the sensor data is consistent with the video AI result based on the comparison result; when the video AI system has abnormal sensor data, it automatically captures real-time video clips as sample data and uploads them to the cloud server for storage.

[0010] In one of the embodiments, the association mechanism between the video AI system and the sensor data includes the following units: an edge computing unit, used to run the video AI system and process sensor data; a video AI unit, located on an edge computing device, processes the real-time video stream of the water conservancy system and identifies abnormal situations; a sensor data acquisition unit, connected to the sensors in the water conservancy system, collects sensor data in real time and performs preprocessing; an association mechanism unit, which realizes time synchronization, spatial position association and data fusion between the video AI system and the sensor data.

[0011] In one of the embodiments, the edge computing element includes: a processor subunit, which is used to run the video AI algorithm, has a number of CPU cores, and has a suitable hardware configuration and software environment for running the video AI system and processing sensor data; a network interface subunit, which is used to establish a stable connection between the sensor and the cloud and configure network security settings.

[0012] In one of the embodiments, the video AI unit also includes: selecting a video AI model suitable for the needs of identifying abnormal situations in the water conservancy system, and converting it into a format suitable for operation on an edge computing device; fine-tuning parameters of the video AI model on the edge device and optimizing hyperparameters; and pre-processing the real-time video stream and inputting it into the video AI model for processing.

[0013] In one embodiment, the association mechanism unit also includes a time synchronization service for synchronizing the time data of the video time data and the time data of the sensor data; establishing a mapping relationship between the physical position of the sensor in the water conservancy system and the spatial position of the video monitoring area; formulating data fusion rules to fuse the video AI recognition results and the sensor data; and implementing the association engine program to associate the video AI system with the sensor data in real time. Further, according to the operation logic of the water conservancy system and the abnormal situation judgment criteria, the video AI recognition results and the sensor data are fused, and the association engine program is implemented to associate the video AI system with the sensor data in real time according to time synchronization, spatial position association and data fusion rules. The spatial position association establishes a mapping relationship by determining the physical position of the sensor and the spatial range of the video monitoring area, and adds a location information tag or a monitoring area identifier to the video AI model or sensor data. The data fusion rules consider the weight and credibility of different types of data and perform reasonable weighted processing. When the video AI detects an abnormal situation and the sensor data also shows an abnormal situation, it is determined that an abnormal situation has occurred.

[0014] In one embodiment, the time synchronization service uses a network time protocol to record time stamps for video data and sensor data.

[0015] The water conservancy safety detection method provided by the above embodiment combined with AI edge computing has the following features:

[0016] Beneficial effects:

[0017] By using sensors to monitor various parameters of the water conservancy system in real time and combining video AI analysis, abnormal situations can be detected in time, alarms can be triggered, and operators can be notified to handle them. This greatly improves the safety of the water conservancy system and reduces the risk of potential safety accidents. The linkage comparison of video AI and sensor data improves the accuracy of abnormal situation judgment. Compared with a single monitoring method, this multi-dimensional detection method can more reliably identify real abnormal situations and avoid false alarms and missed reports. When a fault occurs, the fault code alarm can be quickly triggered and the operator can be notified to handle it. At the same time, the information is sent to the cloud operation and maintenance application background, which is convenient for managers to understand the situation in time and coordinate resources for repair, shortening the fault handling time and improving the reliability of the water conservancy system. When the data reaches a specific interval threshold, the pump IO output is triggered to realize automatic control and adjustment. This can automatically adjust the operating status of the equipment according to the actual operation of the water conservancy system, improve the operating efficiency of the system, and also reduce energy consumption. When the pump control command is input, the working status of the pump is confirmed by combining the sensor detection results and the video AI judgment results to ensure that the pump is operated accurately. This avoids system operation abnormalities caused by incorrect operations and improves the use efficiency and life of the pump. The cloud-based operation and maintenance application backend can analyze the collected data and generate a report on the safety status of the water conservancy system. This provides managers with a basis for decision-making, helps optimize the operation strategy of the water conservancy system, and further improves operation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0019] Figure 1 A working step diagram of a water conservancy safety detection method based on AI edge computing provided in the first embodiment of the present invention

[0020] Figure 2 A diagram showing the working steps of the AI ​​algorithm linked sensor real-time monitoring data provided by the first embodiment of the present invention;

[0021] Figure 3 This is a diagram of the video AI working unit of the present invention. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] It should be noted that if the embodiments of the present invention involve directional indications (such as up, down, left, right, front, back, etc.), the directional indications are only used to explain the relative position relationship, movement status, etc. between the components in a certain specific posture. If the specific posture changes, the directional indication will also change accordingly.

[0024] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features limited to "first" and "second" may explicitly or implicitly include at least one of the features. In addition, if "and / or" or "and / or" appears in the full text, its meaning includes three parallel solutions. Taking "A and / or B" as an example, it includes solution A, solution B, or solutions that satisfy both A and B. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in this field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0025] Example 1

[0026] refer to Figure 1 to Figure 3One of the embodiments of the present invention provides a water conservancy safety detection method based on AI edge computing, including the following steps: sensor installation and encoding, deploying the sensor in the water conservancy system, encoding the sensor, recording the geographical location information and measured physical quantity of the sensor; setting measurement parameters, setting the measurement interval value and threshold of each sensor; data linkage setting, including defining data linkage selection data, triggering the pump IO output when the data reaches a specific interval threshold; according to the AI ​​algorithm linkage sensor real-time monitoring data, when data abnormalities occur, starting the video AI to check whether the comparison results are consistent, if the video AI confirms that the sensor data is abnormal, intercepting the real-time video as sample data and sending it to the cloud server, triggering the corresponding alarm according to the abnormal situation, and notifying the operator to handle it; according to the input pump control instruction, combined with the sensor detection result and the video AI judgment result, confirm the working status of the pump, if the instruction execution result is inconsistent with the expectation, according to the linkage setting, trigger the fault code alarm, notify the operator to handle it, and send the information to the cloud operation and maintenance application background.

[0027] The sensor part includes: a water level sensor, which is deployed in a water storage and water diversion physical field such as a water conservancy system, and is used to monitor the water level in real time. The water level sensor can be of different types such as pressure type, ultrasonic type, radar type, etc., and the appropriate sensor type can be selected according to the actual situation. The water pump opening and closing sensor is installed on the water pump to monitor the working status of the water pump, including information such as opening, closing, and running time. The water gauge recognition sensor is used to accurately read the water level scale on the water gauge, and image recognition technology can be used to improve the accuracy of water level measurement. The control part includes: a water pump control switch, which is used to receive instructions to control the start and stop of the water pump. The water pump control switch can be controlled manually, automatically or remotely, and the appropriate control method can be selected according to the actual situation. In some embodiments, a number of relays are also included to amplify the control signal and drive the water pump. The selection of relays should be based on the power of the water pump and the requirements of the control signal. The cloud server is used to collect data, process alarm information, and provide manual disposal procedures. The cloud server can use cloud computing technology to realize data storage, analysis and processing, and improve the reliability and scalability of the system.

[0028] In one embodiment, the sensor installation and coding process steps also include: the sensor includes a water level sensor, a water pump on / off sensor and a water gauge identification sensor, the water level sensor is arranged in the physical field of the water conservancy system, and is used to monitor the water level in real time; the water pump on / off sensor is arranged in the water pump, and is used to monitor the working state of the water pump, including opening, closing, and running time data; the water gauge sensor is used to accurately read the water level scale on the water gauge. The coding rule uses a unified coding rule to encode various sensors to ensure the uniqueness and identifiability of the coding. The coding can be a combination of numbers, letters or symbols, and is selected according to the actual situation. The coding should include information such as the type, installation location, and number of the sensor to facilitate subsequent data query and management. The physical quantity to be measured of each sensor is accurately defined, including information such as water level, water pump working status, and water gauge scale. The definition of the physical quantity to be measured should be consistent with the type and function of the sensor to facilitate subsequent data processing and analysis. The definition of the physical quantity data to be measured should adopt standard units and specifications to improve the comparability and accuracy of the data.

[0029] In one embodiment, the step of setting the measurement parameters includes: setting the sensor database, and setting the measurement interval value and threshold value of the sensor according to the historical data model analysis. In this embodiment, the measurement interval value and threshold value are automatically adjusted and set through the historical data model analysis. The automatic adjustment setting can improve the intelligence level of the system, reduce manual intervention, and improve the reliability and stability of the system.

[0030] The historical data model analysis step includes: collecting historical data of water conservancy system sensors, including data in different time periods and different operating states. Sorting and cleaning the historical data, removing noise and outliers, and ensuring the quality and reliability of the data. Selecting a suitable data analysis method according to the characteristics of the historical data and the analysis requirements. Statistical analysis, machine learning, data mining and other methods can be used to analyze and model the historical data. Using machine learning algorithms, such as regression analysis and classification algorithms, a relationship model between sensor data and the operating state of the water conservancy system is established, and the measurement interval value and threshold value are set according to the model prediction results. The selected data analysis method is used to establish a historical data model. Model training and optimization are performed according to historical data, and model parameters are adjusted to improve the accuracy and reliability of the model. The established model is verified, and the model is tested using an independent test data set to evaluate the performance and prediction accuracy of the model. According to the historical data model analysis results, the measurement interval value and threshold value of the sensor are set. The measurement interval value should be able to cover the range of changes in sensor data under the normal operating state of the water conservancy system, while leaving a certain margin to avoid false alarms. The threshold value should be set according to the safety requirements and risk tolerance of the water conservancy system. When the sensor data exceeds the threshold value, an alarm is triggered or corresponding control measures are taken. By setting up a sensor database and setting measurement interval values ​​and thresholds based on historical data model analysis, the accuracy and reliability of water conservancy system safety detection can be improved, abnormal situations can be discovered in time and corresponding measures can be taken to ensure the safe operation of the water conservancy system.

[0031] In one of the embodiments, the data linkage setting step also includes: setting a PLC control system to read sensor measurement data in real time; determining whether the sensor data reaches or exceeds the measurement interval value and threshold; if the above steps meet the conditions, triggering the pump IO output to control the start and stop of the water pump.

[0032] The PLC reads the sensor's measurement data in real time through the input module. Preprocess the read data, including noise removal, filtering and other operations to improve the data quality. Convert the sensor data into actual physical quantities, such as converting analog input into physical quantities such as water level height and pressure. At the same time, perform unit conversion and standardization on the data for subsequent comparison and judgment. Design a suitable comparison algorithm to determine whether the sensor data reaches or exceeds the measurement interval value and threshold. Simple comparison operators such as greater than, less than, equal to, etc. can be used, or complex logical judgment statements can be used. Consider the fluctuations and errors of the sensor data and set a certain tolerance range. When the sensor data is close to the measurement interval value and threshold within the tolerance range, the pump IO output is not triggered immediately, but further judgment and confirmation is performed.

[0033] The design of the water pump sensor is related to the working logic of the water pump. The control logic of the pump is designed according to the control requirements of the water conservancy system. When the sensor data meets the trigger condition, the PLC controls the start and stop of the pump through the output module. Direct control can be used, that is, the output signal of the PLC directly controls the power switch of the pump; indirect control can also be used, that is, the start and stop of the pump can be controlled by controlling intermediate devices such as relays and contactors.

[0034] Consider the safety issues during the start and stop of the pump and set up corresponding protection measures. For example, pre-lubrication and check the operating status of the pump before starting the pump; when the pump stops, gradually reduce the speed of the pump to avoid water hammer.

[0035] In one of the embodiments, the real-time monitoring of sensor data also includes: setting up a video AI system, and establishing an association mechanism between the video AI system and the sensor data based on AI edge computing; setting a comparison mechanism, and comparing the real-time sensor data with the video monitoring data according to the comparison mechanism, and judging whether the sensor data is consistent with the video AI result based on the comparison result; if the video AI system confirms that the sensor data is abnormal, it automatically captures the real-time video clip as sample data.

[0036] In one of the embodiments, the association mechanism between the video AI system and the sensor data includes the following units: an edge computing unit, used to run the video AI system and process sensor data; a video AI unit, located on an edge computing device, processes the real-time video stream of the water conservancy system and identifies abnormal situations; a sensor data acquisition unit, connected to the sensors in the water conservancy system, collects sensor data in real time and performs preprocessing; an association mechanism unit, which realizes time synchronization, spatial position association and data fusion between the video AI system and the sensor data.

[0037] In one of the embodiments, the edge computing element includes: a processor subunit, which is used to run the video AI algorithm, has a number of CPU cores, and has a suitable hardware configuration and software environment for running the video AI system and processing sensor data; a network interface subunit, which is used to establish a stable connection between the sensor and the cloud and configure network security settings.

[0038] In one of the embodiments, the video AI unit also includes: selecting a video AI model suitable for the needs of identifying abnormal situations in the water conservancy system, and converting it into a format suitable for operation on an edge computing device; fine-tuning parameters of the video AI model on the edge device and optimizing hyperparameters; and pre-processing the real-time video stream and inputting it into the video AI model for processing.

[0039] In one embodiment, the association mechanism unit also includes a time synchronization service for synchronizing the time data of the video time data and the time data of the sensor data; establishing a mapping relationship between the physical position of the sensor in the water conservancy system and the spatial position of the video monitoring area; formulating data fusion rules to fuse the video AI recognition results and the sensor data; and implementing the association engine program to associate the video AI system with the sensor data in real time. Further, according to the operation logic of the water conservancy system and the abnormal situation judgment criteria, the video AI recognition results and the sensor data are fused, and the association engine program is implemented to associate the video AI system with the sensor data in real time according to time synchronization, spatial position association and data fusion rules. The spatial position association establishes a mapping relationship by determining the physical position of the sensor and the spatial range of the video monitoring area, and adds a location information tag or a monitoring area identifier to the video AI model or sensor data. The data fusion rules consider the weight and credibility of different types of data and perform reasonable weighted processing. When the video AI detects an abnormal situation and the sensor data also shows an abnormal situation, it is determined that an abnormal situation has occurred.

[0040] In one embodiment, the time synchronization service uses the Network Time Protocol to record accurate time stamps for video data and sensor data.

[0041] The National Time Protocol (NTP) is able to achieve microsecond-level time synchronization accuracy, ensuring highly accurate timestamps for video data and sensor data. This is critical for accurately correlating video footage and sensor data, especially when analyzing rapidly changing water system states. NTP is a mature and widely used time synchronization protocol with high reliability and stability. It can work in different network environments and can automatically adjust time deviations to ensure continued accuracy of time synchronization. NTP can be easily extended to large-scale water systems and support time synchronization for multiple video surveillance devices and sensors. Whether it is a small water facility or a large water project, time synchronization can be effectively achieved. Configure an NTP server: Set up one or more NTP servers in the water system to provide an accurate time source for video surveillance devices and sensors. An NTP server can be an independent hardware device or a software program running on a server.

[0042] Video surveillance devices and sensors, as NTP clients, need to be configured accordingly to connect to the NTP server and obtain time synchronization. This usually includes setting parameters such as the IP address and port number of the NTP server.

[0043] In order to ensure the accuracy of time synchronization, it is necessary to monitor the time synchronization status of video surveillance equipment and sensors. You can use a dedicated time synchronization monitoring tool to regularly check whether the time deviation is within an acceptable range. If the time deviation is found to be too large, timely measures should be taken to adjust it.

[0044] Methods for determining the physical location of sensors and the spatial scope of video surveillance areas include: Geographic Information System (GIS) positioning. Using GIS technology, the geographical location of sensors in the water conservancy system can be accurately determined. By combining the coordinate information of the sensor with the map data of the water conservancy system, the location distribution of the sensors can be clearly understood. For the video surveillance area, accurate calibration is required to determine its spatial scope in the water conservancy system. The boundaries and scope of the video surveillance area can be determined by setting reference points in the video screen, measuring distances, and other methods. For complex water conservancy systems, three-dimensional modeling technology can be used to build a three-dimensional model of the water conservancy system. In the model, the location of the sensor and the scope of the video surveillance area can be accurately marked to achieve a more intuitive spatial position mapping.

[0045] Add location information tags to the video AI model: You can add location information tags to each video frame during the training process of the video AI model. These tags can include information such as the sensor number and location coordinates, so that the video AI system can identify the abnormal situation while determining the location where the abnormal situation occurred. When collecting data, the sensor can also record the identification of the monitoring area. In this way, when the sensor data is transmitted to the association engine program, it can be associated with the corresponding video monitoring area. As the water conservancy system operates and changes, the location of the sensor and the scope of the video monitoring area may change. Therefore, it is necessary to establish a dynamic update mechanism to update the location information tags and monitoring area identification in a timely manner to ensure the accuracy of spatial position mapping.

[0046] Conduct in-depth analysis of video AI recognition results and sensor data to evaluate their accuracy and reliability under different circumstances. For example, by analyzing historical data, determine the recognition accuracy of the video AI system under different lighting conditions and weather conditions, as well as the stability and accuracy of the sensor under different working environments. According to the results of data analysis, formulate reasonable weight distribution principles. For data with higher accuracy and reliability, higher weights can be assigned; for data with greater uncertainty, lower weights can be assigned. At the same time, consider the correlation between different types of data to avoid repeated calculations and contradictory results. In addition to weight allocation, the credibility of the data also needs to be evaluated. In some embodiments, the credibility of the data is determined by setting a confidence interval method. When the credibility of the data is low, further verification measures can be taken or its impact in the fusion result can be reduced. When the video AI detects an abnormality and the sensor data also shows an abnormality, it is determined that the decision-making process of the abnormality has occurred.

[0047] In some embodiments, the definition of abnormal conditions in the water conservancy system includes various possible abnormal types and characteristics. For example, too high a water level, abnormal water flow velocity, equipment failure, etc. can all be defined as abnormal conditions.

[0048] In this embodiment, multimodal data fusion is used. When the video AI detects an abnormal situation, the correlation engine program will simultaneously check whether the sensor data also shows an abnormality. If both the video AI recognition result and the sensor data meet the definition of an abnormal situation, then it can be determined that an abnormal situation has occurred. In this process, multimodal data fusion is required to comprehensively consider information from multiple aspects such as video images and sensor values.

[0049] Once an abnormal situation is determined to have occurred, the correlation engine program will immediately trigger the alarm mechanism and notify relevant personnel to handle it. At the same time, according to the severity of the abnormal situation, the corresponding emergency plan can be automatically activated and emergency measures can be taken to ensure the safe operation of the water conservancy system.

[0050] In some embodiments, the association engine program can adopt an event-driven architecture, and trigger corresponding events when the video AI system detects an abnormal situation or a change in sensor data. The association engine program will perform real-time association and analysis based on the event type and related data. In order to improve the efficiency of association, the association engine program can set up a data caching mechanism. Cache video data and sensor data for quick access and processing when needed. At the same time, adopt efficient data processing algorithms to ensure that data fusion and abnormality judgment can be completed in a short time. For large-scale water conservancy systems, a distributed deployment of the association engine program can be considered. Deploy the association engine program on multiple nodes to achieve parallel processing and load balancing. This can improve the system's processing power and response speed, and ensure that a large amount of video and sensor data can be processed in a timely manner.

[0051] Monitor the performance of the correlation engine program, including processing speed, response time, resource usage and other indicators. According to the monitoring results, optimize the performance, such as adjusting algorithm parameters, optimizing data structure, increasing hardware resources, etc. By expanding the time synchronization service, spatial location mapping relationship, data fusion rules and correlation engine program, the accuracy and reliability of the water conservancy safety detection method based on AI edge computing can be further improved, providing a more powerful guarantee for the safe operation of the water conservancy system.

[0052] The present invention provides a water conservancy safety detection method based on AI edge computing. Through sensors, various parameters of the water conservancy system are monitored in real time. Combined with video AI analysis, abnormal situations can be discovered in time and alarms can be triggered, and operators can be notified to handle them. This greatly improves the safety of the water conservancy system and reduces the potential risk of safety accidents. The linkage comparison of video AI and sensor data improves the accuracy of abnormal situation judgment. Compared with a single monitoring method, this multi-dimensional detection method can more reliably identify real abnormal situations and avoid false alarms and missed reports. When a fault occurs, the fault code alarm can be quickly triggered and the operator can be notified to handle it. At the same time, the information is sent to the cloud operation and maintenance application background, which is convenient for managers to understand the situation in time and coordinate resources for repair, shortening the fault handling time and improving the reliability of the water conservancy system. When the data reaches a specific interval threshold, the pump IO output is triggered to achieve automatic control and regulation. This can automatically adjust the operating status of the equipment according to the actual operation of the water conservancy system, improve the operating efficiency of the system, and also reduce energy consumption. When the pump control instruction is input, the working status of the pump is confirmed by combining the sensor detection results and the video AI judgment results to ensure that the pump is operated accurately. This avoids system operation abnormalities caused by incorrect operations and improves the efficiency and life of the pump. The cloud-based operation and maintenance application background can analyze the collected data and generate a water conservancy system safety status report. This provides managers with a basis for decision-making, helps optimize the operation strategy of the water conservancy system, and further improves operation efficiency.

[0053] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. All equivalent structural changes made by using the contents of the present invention specification and drawings under the inventive concept of the present invention, or directly / indirectly applied in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A water conservancy safety detection method based on AI edge computing, characterized in that: The following steps are involved: Sensor installation and coding: deploying sensors in the water conservancy system, coding the sensors, and recording the geographical location information and measured physical quantities of the sensors; Set measurement parameters, set the measurement interval value and threshold value of each sensor; Data linkage settings, including defining data linkage selection data, triggering pump IO output when the data reaches a specific interval threshold; According to the AI ​​algorithm, the sensor monitors the data in real time. When abnormal data occurs, the video AI is started to check whether the comparison results are consistent. If the video AI confirms that the sensor data is abnormal, the real-time video is intercepted as sample data and sent to the cloud server. According to the abnormal situation, the corresponding alarm is triggered to notify the operator to handle it. According to the input pump control instructions, combined with the sensor detection results and video AI judgment results, the working status of the pump is confirmed. If the instruction execution result is inconsistent with the expectation, the fault code alarm is triggered according to the linkage setting, the operating personnel are notified to handle it, and the information is sent to the cloud operation and maintenance application background.

2. The water conservancy safety detection method based on AI edge computing according to claim 1 is characterized in that: The sensor installation and encoding process steps also include: The water conservancy system includes a water pump and a water gauge; The sensors include a water level sensor, a water pump on / off sensor and a water gauge identification sensor. The water level sensor is arranged in the physical field of the water conservancy system to monitor the water level height in real time; the water pump on / off sensor is arranged in the water pump to monitor the working status of the water pump; the water gauge sensor is used to obtain the water level scale data on the water gauge in real time.

3. The water conservancy safety detection method based on AI edge computing according to claim 1 is characterized in that: The steps to set the measurement parameters include: A sensor database is set, and the measurement interval value and threshold value of the sensor are set according to the historical data model analysis.

4. The water conservancy safety detection method based on AI edge computing according to claim 2 is characterized in that: The steps for setting up data linkage also include: Set up a PLC control system to read sensor measurement data in real time; Determining whether the sensor data reaches or exceeds the measurement interval value and threshold value; If the above conditions are met, the pump IO output is triggered to control the start and stop of the water pump.

5. The water conservancy safety detection method based on AI edge computing according to claim 1 is characterized in that: The step of real-time monitoring data by linking sensors with the AI ​​algorithm also includes: Set up a video AI system and establish a mechanism to associate the video AI system with sensor data based on AI edge computing; Set up a comparison mechanism, and compare the real-time sensor data with the video monitoring data according to the comparison mechanism. Based on the comparison results, determine whether the sensor data is consistent with the video AI results. When the video AI system detects abnormal sensor data, it automatically captures real-time video clips as sample data and uploads them to the cloud server for storage.

6. The water conservancy safety detection method based on AI edge computing according to claim 5 is characterized in that: The video AI system is built by including the following units: An edge computing unit for running the video AI system and processing sensor data; The video AI unit, located on the edge computing device, processes the real-time video stream of the water system and identifies abnormal situations; The sensor data acquisition unit is connected to the sensors in the water conservancy system to collect sensor data in real time and perform preprocessing; The association mechanism unit realizes the time synchronization, spatial position association and data fusion of the video AI system and sensor data.

7. The water conservancy safety detection method based on AI edge computing according to claim 6 is characterized in that: The elements of edge computing include: The processor subunit is used to run the video AI algorithm and has a certain number of CPU cores; The network interface subunit is used to establish a stable connection between the sensor and the cloud and configure network security settings.

8. The water conservancy safety detection method based on AI edge computing according to claim 6 is characterized in that: The video AI unit working steps also include: Select a video AI model that is suitable for identifying abnormal conditions in the water conservancy system and convert it into a format suitable for edge computing devices. Fine-tune the parameters of the video AI model on edge devices and optimize hyperparameters; The real-time video stream is preprocessed and then input into the video AI model for processing.

9. The water conservancy safety detection method based on AI edge computing according to claim 1 is characterized in that: The association mechanism unit step also includes, Time synchronization service, used to synchronize the video time data and sensor data time data; Establish a mapping relationship between the physical location of the sensor in the water conservancy system and the spatial location of the video surveillance area; Formulate data fusion rules to fuse video AI recognition results and sensor data; Implement a correlation engine program to correlate the video AI system with sensor data in real time.

10. The water conservancy safety detection method based on AI edge computing according to claim 8, characterized in that: The time synchronization service uses the Network Time Protocol to record time stamps for video data and sensor data.

Images