A hydrological monitoring method and device, computer equipment and storage medium

Abstract

The application relates to the computer technical field and discloses a hydrological monitoring method and device, computer equipment and a storage medium, the method comprises the following steps: receiving hydrological observation data of a hydrological station at each sampling time reported from a sensor cluster, and receiving video stream information of the hydrological station reported from a video monitoring system, processing the hydrological observation data to obtain observation data of each sensor and water level data of a water gauge in the hydrological station, analyzing the video stream information, determining the corresponding relationship between a photo set in the video stream information and the observation data and the water level data, and calculating water level information of the water gauge, inputting the water level data and the water level information of the water gauge into a trained neural network model, outputting a target water level, and monitoring the water level condition of the hydrological station according to the target water level. The method adopts an integrated mode combining sensors and video monitoring for intelligent management, and effectively solves the problems of weak information technology equipment and insufficient perception of the hydrological station.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a hydrological monitoring method, device, computer equipment and storage medium. Background Art

[0002] Water resources are fundamental to the development and survival of human resources and equipment, making their management and utilization crucial. Hydrological stations, as the primary vehicle and fundamental unit for hydrological monitoring, collect, analyze, and transmit hydrological data, providing a crucial basis for water resource management. Hydrological elements observed at hydrological stations include water level, flow velocity, flow direction, waves, sediment content, water temperature, ice conditions, groundwater, and water quality. Meteorological elements include precipitation, evaporation, temperature, humidity, air pressure, and wind.

[0003] Currently, hydrological stations mainly collect hydrological data through rain gauges, radar water level gauges, float water level gauges, radar wave flow meters, and Doppler current profilers (ADCP). These instruments suffer from problems such as aging equipment, single monitoring methods, inconsistent data collection protocols, inaccurate data collection, and a high risk of data leakage.

[0004] In addition, managers check hydrological information through on-site inspections and video surveillance, which is highly dependent on manpower, high cost, and cannot identify changes in hydrological data in real time, quickly, and accurately. Summary of the Invention

[0005] In view of this, the present invention provides a hydrological monitoring method, device, computer equipment and storage medium to solve the above-mentioned series of problems in the current hydrological station's collection of hydrological data, as well as the problems of high dependence on manpower and high cost when viewing hydrological information, and the inability to identify changes in hydrological data in real time, quickly and accurately.

[0006] In a first aspect, the present invention provides a hydrological monitoring method, the method comprising:

[0007] Receive hydrological observation data of the hydrological station at each sampling moment reported by the sensor cluster, and receive video stream information of the hydrological station reported by the video monitoring system, the video stream information including a continuous set of photos taken of the hydrological station environment;

[0008] Processing the hydrological observation data to obtain observation data of each sensor and water level data of the water gauge in the measured hydrological station;

[0009] parsing the video stream information, determining a correspondence between a photo set in the video stream information and the observation data and the water level data, and calculating the water level information of the water gauge based on the correspondence, the observation data and the water level data;

[0010] The water level data and the water level information of the water gauge are input into a trained neural network model, a target water level is output, and the water level condition of the hydrological station is monitored according to the target water level.

[0011] The hydrological monitoring method provided by the present invention has the following advantages:

[0012] By integrating sensor data and video streams, the system provides multi-dimensional hydrological data analysis, enabling a more comprehensive understanding of hydrological phenomena. Combining video surveillance photo collections with sensor data allows for more precise determination of water level changes, reducing the potential for errors associated with a single data source.

[0013] Furthermore, the data collection and monitoring approach can mitigate issues such as aging equipment, single monitoring methods, inconsistent data collection protocols, and inaccurate data collection. The system can calculate water level information in real time and predict target water levels using a neural network model, helping to detect anomalies and issue early warnings.

[0014] Automated data processing and water level forecasting reduce manual intervention and improve decision-making efficiency. The integration of multi-source data enhances system robustness, ensuring continued operation even in the event of sensor failure. Intuitive video streams and precise water level data provide users with a more intuitive understanding of hydrological conditions, enhancing the user experience.

[0015] Automated monitoring and early warning systems can reduce the frequency of manual inspections, thereby lowering operating costs. At the same time, intelligent identification through video surveillance can enhance safety management capabilities around hydrological stations and achieve intelligent upgrades to hydrological stations.

[0016] In an optional implementation, the receiving of hydrological observation data of the hydrological station at each sampling moment reported by the sensor cluster includes:

[0017] receiving the hydrological observation data reported by the sensor cluster via a wired manner, or,

[0018] Receive the hydrological observation data reported by the sensor cluster via wireless LoRa, or,

[0019] The hydrological observation data is received through Beidou communication access. Specifically, in the above method of the present application, by supporting a variety of wired or wireless communication methods, the deployment of hydrological stations in various environments can be met. In this way, the problems of weak information technology equipment, insufficient perception, insufficient data security protection capabilities and insufficient intelligence levels of hydrological stations in related technologies can be solved. For example, in an environment with a wired network, a wired network can be used to collect sensor data to improve data security and the stability of data transmission. In an environment without a wired network, a wireless LoRa method can be used to collect sensor data, reduce the risk of data exposure to the public network, improve data security and save costs.

[0020] Furthermore, the high-precision positioning capabilities of Beidou communication access can enhance system reliability, ensure accurate and real-time data collection, and potentially reduce reliance on traditional communication networks, lowering communication costs. The use of wireless LoRa technology can reduce wiring costs. Wireless LoRa and Beidou communication access methods are not restricted by terrain and are suitable for complex and changing natural environments, expanding the system's applicability. Wired methods offer relatively low maintenance costs and more straightforward troubleshooting. In the event of a wired connection interruption, wireless connections can serve as a backup, ensuring the continuity and integrity of data transmission. Furthermore, in emergencies, wireless communication can quickly restore data transmission, enhancing the system's emergency response capabilities. In summary, the above multi-mode data reception strategy significantly improves the performance, adaptability, and practicality of the hydrological monitoring system.

[0021] In an optional implementation, the receiving of video stream information from a hydrological station reported by a video surveillance system includes:

[0022] In the case of wired communication, receiving the video stream information reported by the video surveillance system through a wired network;

[0023] Alternatively, in the case of wireless communication, the video stream information reported by the video surveillance system is received via a wireless bridge;

[0024] Alternatively, when wireless communication is available but wireless bridge communication is not available, the video stream information is received via WLAN.

[0025] Specifically, similar to the above, in the above method of the present application, by supporting a variety of wired or wireless communication methods, the deployment of hydrological stations in various environments can be met. In this way, the problems in the related art of hydrological stations, such as weak information technology equipment, insufficient perception, insufficient data security protection capabilities, and insufficient intelligence levels, can be solved. For example, in an environment with a wired network, the video stream information reported by the video surveillance system can be received using a wired network to improve data security and data transmission stability. In an environment without a wired network, the video stream information reported by the video surveillance system is received through a wireless bridge to reduce the risk of data exposure to the public network, improve data security, and save costs.

[0026] In the case of wireless communication but no wireless bridge communication, the video stream information is received via WLAN. In the case of poor or unavailable wired communication conditions, wireless bridges and WLAN provide alternative solutions, allowing video surveillance systems to be deployed in a wider range of environments.

[0027] Wireless communication, particularly WLAN, is generally easier to deploy and more cost-effective than wired networks, making it particularly suitable for areas with complex terrain or difficult-to-lay cabling. Wireless transmission typically offers faster response times, enabling rapid reception of video stream information in emergency situations, facilitating quick response and decision-making. Multiple communication methods provide system redundancy; if one communication method fails, other methods can ensure continuous data transmission. Wireless communication reduces the need for physical connection maintenance, lowering system maintenance costs and complexity. Wired communication typically offers more stable and higher data rates, helping to ensure the quality and integrity of video streams. In unstable network conditions, wireless methods such as WLAN provide a more reliable connection, preventing data loss caused by network fluctuations. Wireless communication facilitates remote monitoring, allowing users to access the video surveillance system from anywhere via the internet. The system can select the appropriate communication method based on specific site conditions, ensuring continuous video monitoring, whether wired or wireless. Furthermore, support for multiple communication methods facilitates future system expansion, making it easy to add new monitoring points or upgrade the system. Flexible video stream information reception methods improve the overall performance of the hydrological station monitoring system, enhancing its adaptability, reliability, and efficiency.

[0028] In an optional implementation manner, before processing the hydrological observation data, the method further includes:

[0029] Aligning the hydrological observation data with the plurality of photos in the photo set according to each sampling moment and the time point of the continuous photo set;

[0030] The processing of the hydrological observation data to obtain the observation data of each sensor and the water level data of the water gauge in the measured hydrological station includes:

[0031] The time-aligned hydrological observation data are processed to obtain the observation data of each sensor and the water level data of the measured water gauge, wherein the observation data includes the sampling data measured by each sensor and the sensor ID.

[0032] Specifically, this approach ensures precise alignment between hydrological observation data and photographic time points, improving data consistency and accuracy. Through time alignment, researchers and engineers can save time on data synchronization and matching, enabling more efficient data analysis. Furthermore, the enhanced correlation between observation data and photographs facilitates more accurate interpretation and verification of observation data, particularly when analyzing water level fluctuations and environmental conditions. Real-time aligned data can also facilitate rapid response and appropriate measures in the event of water level anomalies.

[0033] Furthermore, by processing observational data and water gauge water level data, noise and inaccuracies can be removed, improving data quality. Combining photographs with hydrological data provides more intuitive data visualization, helping to understand and interpret trends and patterns within the data. The combination of observational data and photographs allows for analysis from multiple perspectives, such as water level changes, flow velocity, and water quality. More accurate water level data helps optimize water resource management decisions, including irrigation, drainage, and flood prevention. Furthermore, ensuring data consistency improves the reliability of the entire monitoring system and reduces errors and false alarms.

[0034] In an optional implementation, after parsing the video stream information, the method further includes:

[0035] parsing the video stream information to obtain a plurality of photos taken continuously in the photo collection;

[0036] Inputting the plurality of photos into a self-trained neural network model to obtain training results;

[0037] Determining whether the plurality of photos contain a water gauge target according to the training result;

[0038] If included, get the frame height of the water gauge target and the water gauge area image;

[0039] Detecting whether the water gauge area image contains the target character E through the self-trained neural network model;

[0040] If the target character E is included, then the average height EH of the target character E is calculated;

[0041] Calculating the water level information of the water gauge according to the corresponding relationship, the observation data, and the water level data includes:

[0042] Obtain the actual length eH of the target character E;

[0043] According to the corresponding relationship, the average height of the target character E and the actual length eH, the actual length of the water gauge is calculated, and the water level information including the actual length of the water gauge is generated.

[0044] Specifically, automated detection of water level information from water gauge photographs reduces the burden of manual monitoring and improves the efficiency and accuracy of water level monitoring. The real-time analysis capabilities of the neural network model enable real-time updates of water level information, which is crucial for hydrological events requiring rapid response. Automated object detection and water level calculation reduce data bias caused by human error. The neural network model enables more accurate recognition of water gauge objects and characters, thereby improving the accuracy of water level information. Automated processing reduces the need for human resources and lowers long-term operating costs.

[0045] This method can be easily expanded to multiple monitoring points and is applicable to large hydrological monitoring networks. By combining water level information with observational and water level data, it improves data availability and comprehensiveness. Accurate water level information provides strong support for decision-making in water resource management, flood prevention, and response. The neural network model can adapt to varying lighting conditions and water gauge variations, enhancing the robustness of the system. The application of this method promotes the application and development of computer vision and machine learning technologies in hydrological monitoring.

[0046] In an optional embodiment, the method further includes: obtaining the trained neural network model;

[0047] The step of obtaining a trained neural network model includes:

[0048] Obtain the initial neural network model, as well as the predicted value and auxiliary value of the previous moment;

[0049] Obtain the current water level value t from the water level detection system or video monitoring system;

[0050] Merge the current moment input value and the auxiliary value at the previous moment to obtain the current moment merged output gate;

[0051] The following processing is performed on the current moment merge output gate:

[0052] Passing the current moment merge output gate through the first sigmoid neural network to obtain the current moment forget gate;

[0053] Passing the current moment merged output gate through a second sigmoid neural network to obtain the current moment input gate;

[0054] Passing the current moment merge output gate through the first tanh neural network to obtain the current moment state update value;

[0055] Perform a dot product of the current moment forget gate and the previous moment prediction value to obtain;

[0056] Perform dot multiplication on the current state update value and the current input gate to obtain;

[0057] Adding the above and the above point by point to obtain the current moment prediction value;

[0058] The current moment prediction value is calculated by the second tanh neural network, and the current moment combined output gate is multiplied to obtain the current moment auxiliary value;

[0059] The current moment prediction value and the current moment auxiliary value are used as input loop for training, and are detected by the cross entropy loss function until the difference between the measured true value and the predicted value reaches a preset value, then the loop is exited and the neural network model training is completed.

[0060] In an optional embodiment, monitoring the water level condition of the hydrological station according to the target water level includes:

[0061] Determining whether the target water level is within a safety line;

[0062] If yes, it is determined that the water level condition at the hydrological station is normal;

[0063] If not, it is determined that the water level condition is abnormal, an alarm message is generated, and the alarm message is sent to the alarm system.

[0064] Specifically, the system monitors water levels at hydrological stations in real time, ensuring timely detection of potential water level anomalies, thereby protecting people's lives and property. By promptly determining whether water levels exceed safety limits, early warnings can be provided, reducing the risk of disasters such as floods and waterlogging. When water level anomalies occur, the system quickly generates and sends alerts to the warning system, enabling relevant departments to respond quickly and take necessary preventive measures. Timely monitoring and alerting can reduce property losses and casualties caused by excessively high or low water levels.

[0065] In a second aspect, the present invention provides a hydrological monitoring device, comprising:

[0066] A receiving module, configured to receive hydrological observation data of the hydrological station at each sampling moment reported by the sensor cluster, and receive video stream information of the hydrological station reported by the video monitoring system, wherein the video stream information includes a continuous set of photos taken of the hydrological station environment;

[0067] A processing module is used to process the hydrological observation data to obtain the observation data of each sensor and the water level data of the water gauge in the measured hydrological station;

[0068] a determination module, configured to parse the video stream information, determine a correspondence between a set of photos in the video stream information and the observation data and the water level data, and calculate the water level information of the water gauge based on the correspondence, the observation data, and the water level data;

[0069] The processing module is further used to input the water level data and the water level information of the water gauge into the trained neural network model, output the target water level, and monitor the water level status of the hydrological station according to the target water level.

[0070] In a third aspect, the present invention provides a computer device comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, computer instructions being stored in the memory, and the processor executing the hydrological monitoring method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0071] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the hydrological monitoring method of the first aspect or any corresponding embodiment thereof.

[0072] In addition, the present invention provides a computer program product, including computer instructions, which are used to enable a computer to execute the hydrological monitoring method of the above-mentioned first aspect or any corresponding embodiment thereof.

[0073] In a fifth aspect, the present invention provides a hydrological monitoring system comprising at least one edge device, a sensor cluster, a video monitoring system, and a server, wherein the sensor cluster, the video monitoring system, and the edge device may be connected by wire or wirelessly. Each edge device is configured to execute the hydrological monitoring method of the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0075] Figure 1 is a schematic structural diagram of a hydrological monitoring system according to an embodiment of the present invention;

[0076] Figure 2 is a flow chart of a hydrological monitoring method according to an embodiment of the present invention;

[0077] Figure 3 is a schematic diagram of input device information according to an embodiment of the present invention;

[0078] Figure 4 is a flow chart of another hydrological monitoring method according to an embodiment of the present invention;

[0079] Figure 5 1 is a flow chart of a method for calculating the actual length of a water gauge according to an embodiment of the present invention;

[0080] Figure 6 is a schematic diagram of a neural network model according to an embodiment of the present invention;

[0081] Figure 7 is a flow chart of an alarm method according to an embodiment of the present invention;

[0082] Figure 8 is a structural block diagram of a hydrological monitoring device according to an embodiment of the present invention;

[0083] Figure 9 1 is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0084] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are 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 those skilled in the art without making creative efforts shall fall within the scope of protection of the present invention.

[0085] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed, detachable, or integral connections; mechanical or electrical connections; direct or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0086] In addition, the terms “first” and “second” are used for descriptive purposes only and should not be understood as indicating or implying relative importance.

[0087] The technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0088] The technical solution of the present invention is applied to the detection environment of a hydrological station or a hydrological system, see Figure 1 As shown, it is a structural diagram of a hydrological monitoring system according to an embodiment of the present invention. The system includes: a sensor set, a video monitoring system, an edge computing device, a server or a data center.

[0089] The edge computing device includes a data management system that has functions such as data collection, data storage, and data compilation. Optionally, the data management system can be a functional module or functional software.

[0090] Furthermore, the aforementioned edge computing device may be an edge device or edge node, including but not limited to a client, a PC (personal computer), or a terminal device. Optionally, the terminal device may also be a smartphone, a tablet computer, a foldable terminal, a wearable device with wireless communication capabilities, a user device, or a user equipment (UE). This embodiment does not limit the specific device form of the terminal device.

[0091] In addition, the server can also be other network devices, such as a server cluster, a data center, a service node, etc.

[0092] Optionally, the above system may also include an alarm device or an alarm system, such as a speaker, a horn, etc., which is connected to the edge computing device and is used to issue an alarm when the water level at the hydrological station exceeds the safety line.

[0093] The server or data center includes a hydrological management platform, which can be a software platform or a hardware and software platform for analyzing and coordinating data reported by edge computing devices.

[0094] Currently, sensor clusters report sampled observation data to edge computing devices via remote terminal units (RTUs), transmitting the corresponding sensor data directly to the hydrological management platform via 4G / 5G. This solution requires a corresponding RTU device for each sensor, resulting in a lack of unified management and the inability to monitor operating conditions in real time. Furthermore, data transmission via 4G / 5G carries a high risk of data leakage.

[0095] In addition, the data of the water level gauge in the plan needs to be read and recorded by staff on site or through video surveillance. When the corresponding data is entered into the provincial management platform, data entry errors are prone to occur.

[0096] In order to solve the above technical problems, an embodiment of the present invention provides an embodiment of a hydrological monitoring method. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0097] In this embodiment, a hydrological monitoring method is provided, which can be used in the above-mentioned edge computing device. Figure 2 Flowchart of a hydrological monitoring method according to an embodiment of the present invention, the process includes:

[0098] Step S101: receiving hydrological observation data of a hydrological station at each sampling moment reported by a sensor cluster, and receiving video stream information of the hydrological station reported by a video monitoring system.

[0099] The video stream information includes a continuous set of photos taken of the hydrological station environment.

[0100] The sensor cluster includes rainfall sensors or rain gauges, water level sensors, radar wave flow measurement, Doppler current profiler (ADCP) flow measurement, and other sensors such as marker float water level gauges. Hydrological observation data is a general term for the observation data collected by all sensors. It can be sampled data collected at different times, such as water level information.

[0101] The video monitoring system may include at least one camera or camera module for collecting information about the surrounding environment of the hydrological station and taking a set of continuous photos or pictures to form a photo collection.

[0102] See also Figure 3As shown, in this embodiment of the hydrological monitoring method based on the integration of sensors and video surveillance, the input device information includes not only information about at least one sensor device, such as the device ID, parameters, and data information corresponding to each sensor. Similarly, the video surveillance device also includes the corresponding IP address and parameter information.

[0103] In this step, the edge device can receive the hydrological observation data reported by the sensor cluster or the video stream information of the hydrological station reported by the video monitoring system in a wired or wireless manner.

[0104] Step S102: Process the hydrological observation data to obtain the observation data of each sensor and the water level data of the water gauge in the measured hydrological station.

[0105] The water level gauges, flow meters, rain gauges, and other sensors in the sensor cluster are used to monitor water levels, flow rates, rainfall, and other data in real time. These sensors are typically deployed at key locations in reservoirs or hydrological stations to ensure comprehensive and accurate data.

[0106] Each sensor collects water gauge observation data and sends it to the edge device. The edge device uses a water level gauge to observe changes in the water level and obtain water level data. For example, the observed data can be stored on the edge device using digital water gauge storage or digitally coded water gauge storage. Alternatively, the automatic recording and archiving functions of hydrological software can be used to organize all data into a single file.

[0107] Data preprocessing: During the transmission of observation data, it may be necessary to perform preliminary processing on the transmitted data, such as data compression, format conversion, deduplication or interference removal, etc., to improve the transmission efficiency and compatibility of the observation data.

[0108] In addition, the water level data of the water gauge in the measured hydrological station, including information such as the water gauge height or water gauge length, can be obtained through a neural network model or a preset algorithm.

[0109] Step S103: parsing the video stream information, determining the correspondence between the photo set in the video stream information and the observation data and water level data, and calculating the water level information of the water gauge based on the correspondence, the observation data and the water level data.

[0110] The water level information of the water gauge includes water gauge height information, which is real water level data obtained by correcting the water gauge height measured by at least one sensor in the sensor set.

[0111] In this step, each photo in the video stream information is associated with the observation data and water level data collected by the sensor at the same time, and the associated observation data and water level data are used to calculate the water level information of the current water gauge under the data collected by the two systems, such as the water gauge length.

[0112] Step S104: input the water level data and the water level information of the water gauge into the trained neural network model, output the target water level, and monitor the water level status of the hydrological station according to the target water level.

[0113] The trained neural network model can be a trained LSTM model. The LSTM (Long Short-Term Memory) model is a special recurrent neural network (RNN) architecture designed to address the vanishing and exploding gradient problems of traditional RNNs when processing long sequences of data. It captures long-term dependencies of information by introducing a memory cell (Cell State) and three key gating mechanisms (input gate, forget gate, and output gate).

[0114] Monitor the water level status of the hydrological station according to the target water level, specifically including judging whether the target water level exceeds the safety warning line based on the target water level. If it exceeds, make corresponding alarm processing.

[0115] The hydrological monitoring method provided by the present invention has the following advantages:

[0116] By integrating sensor data and video streams, the system provides multi-dimensional hydrological data analysis, enabling a more comprehensive understanding of hydrological phenomena. Combining video surveillance photo collections with sensor data allows for more precise determination of water level changes, reducing the potential for errors associated with a single data source.

[0117] Furthermore, the data collection and monitoring approach can mitigate issues such as aging equipment, single monitoring methods, inconsistent data collection protocols, and inaccurate data collection. The system can calculate water level information in real time and predict target water levels using a neural network model, helping to detect anomalies and issue early warnings.

[0118] Automated data processing and water level prediction functions can reduce manual intervention and improve decision-making efficiency.

[0119] The fusion of multi-source data improves the robustness of the system, allowing it to maintain normal operation even if some sensors fail. Through intuitive video streaming information and accurate water level data, users can more intuitively understand the hydrological conditions, improving the user experience.

[0120] Automated monitoring and early warning systems can reduce the frequency of manual inspections, thereby lowering operating costs. At the same time, intelligent identification through video surveillance can enhance safety management capabilities around hydrological stations and achieve intelligent upgrades to hydrological stations.

[0121] In this embodiment, in the above step S101: during the data collection process, all hydrological data of the hydrological station can be collected simultaneously through wired, wireless bridge, 4G / 5G, LoRa, Beidou and other communication methods, and the data of the hydrological station can be analyzed and processed, and uploaded to the provincial management platform through the hydrological dedicated line at the same time, thereby reducing the risk of data exposure and improving the security of hydrological data.

[0122] Data security is primarily improved through two approaches: first, establishing a local area network (LAN) at the hydrological station using wired or wireless LoRa and wireless bridges to collect and aggregate data, preventing data transmission over public networks and reducing the risk of leakage. Second, after data is collected and aggregated on edge computing devices, it is encrypted using the asymmetric RSA encryption algorithm and decrypted on the server's hydrological platform. This embodiment does not limit the specific encryption and decryption processes.

[0123] Intelligent equipment supervision can simultaneously monitor the operating status of sensors and video surveillance equipment in real time, effectively identify the operating status of the equipment, and intelligently manage the management area of ​​the hydrological station. If there are abnormalities such as strangers breaking in or floating objects in the river, real-time alarms will be issued to improve the safety management level of the hydrological station, realize intelligent supervision of the hydrological station, and ultimately achieve "one station, one measurement, one station, one system".

[0124] In a possible implementation of this embodiment, the above-mentioned step S101: receiving the hydrological observation data of the hydrological station at each sampling moment reported by the sensor cluster, includes: receiving the hydrological observation data reported by the sensor cluster via a wired method, or, receiving the hydrological observation data reported by the sensor cluster via a wireless LoRa method, or, receiving the hydrological observation data via Beidou communication access.

[0125] Similarly, the above step S101, receiving the video stream information of the hydrological station reported by the video surveillance system, includes: in the case of wired communication, receiving the video stream information reported by the video surveillance system through a wired network.

[0126] Alternatively, when wireless communication is available, the video stream information reported by the video surveillance system is received through a wireless bridge; or, when wireless communication is available but wireless bridge communication is not available, the video stream information is received through a WLAN (Wireless Local Area Network).

[0127] Further, see Figure 3 and Figure 4 , the method comprising:

[0128] Obtain device information and classify devices into sensor devices and video surveillance devices.

[0129] For video surveillance equipment, determine whether a wired network environment is available. If so, directly access the system through the wired network.

[0130] If there is no wired network environment, determine whether there are conditions for wireless bridge access;

[0131] If the conditions for wireless bridge transmission are met, the system is accessed through the wireless bridge, and video stream information is sent to the edge computing device through the wireless bridge.

[0132] If a wireless bridge environment is not available, the system determines whether WLAN network transmission is available, such as a 4G or 5G environment. If WLAN transmission is available, the system is connected via 4 / 5G to send video stream information to the edge computing device.

[0133] Optionally, in this embodiment, the video surveillance system supports collecting video streams in two ways: standard video protocol and RTSP (Real Time Streaming Protocol).

[0134] In addition, the video stream information is divided into continuous picture sets according to time periods; optionally, in one example, the audio segment uses a time period of 10 seconds, and the video is divided into picture sets of continuous pictures corresponding to 10 seconds.

[0135] For sensor devices, determine whether each sensor has wired communication conditions;

[0136] If wired communication conditions are available, access the system via a wired method, such as connecting to a USB interface, and transmit hydrological observation data through the USB cable of the USB interface.

[0137] If the wired communication conditions are not met, it is determined whether the wireless LoRa conditions are met. If so, the edge computing device is connected via wireless LoRa.

[0138] LoRa (Long Range Radio) is a low-power, long-range wireless transmission technology, also known as Low Power Wide Area Network (LPWAN) wireless communication technology. LoRa technology utilizes spread spectrum modulation, specifically Chirp Spread Spectrum (CSS) modulation. This technology improves signal immunity and transmission distance by spreading the data signal across a wider spectrum. LoRa uses unlicensed frequency bands, enabling communication without the need to acquire spectrum resources.

[0139] If the wireless LoRa conditions cannot be met, determine whether WLAN network transmission is available, that is, 4G / 5G environment. If so, access the system through 4G / 5G.

[0140] In addition, if none of the above transmission conditions are met, but the system supports access via Beidou communication, in this embodiment, the system supports access via wired network, wireless bridge, WLAN network, such as 4G / 5G, Beidou, etc., and has good compatibility.

[0141] It should be noted that, in special circumstances, Beidou communication transmission is used, and the setting of Beidou communication unidirectional and bidirectional modes, collection cycles and collection frequencies is supported to save equipment resources and Beidou communication costs.

[0142] In a possible implementation of this embodiment, before processing the hydrological observation data, the above step S102 further includes: aligning the hydrological observation data with multiple photos in the photo set according to each sampling moment and the time point of the continuous photo set taken.

[0143] Based on the time information, the hydrological observation data sampled by the sensor is temporally aligned with at least one photo in the photo set to obtain the aligned hydrological observation data and photo set. For example, a binding relationship is established between the hydrological observation data at each sampling moment and the corresponding photo taken at that moment, thereby establishing a binding relationship between a set of photo sets and a set of hydrological observation data at N moments.

[0144] Step S103: Processing the hydrological observation data to obtain the observation data of each sensor and the water level data of the water gauge in the measured hydrological station, including: processing the time-aligned hydrological observation data to obtain the observation data of each sensor and the water level data of the measured water gauge, wherein the observation data includes the sampling data measured by each sensor and the sensor ID.

[0145] Specifically, the edge node inputs the aligned hydrological observation data, such as hydrological observation data collected by at least one sensor at N moments, into the data management system to obtain the data corresponding to each sensor device ID, namely the water level data observed by the sensor.

[0146] Separate all files corresponding to water gauges (such as upstream and downstream water gauges) in the video image set based on the sensor ID data.

[0147] For example, the table below shows the sampling data of water gauges 1 to 3 observed by sensors 1 to 3, that is, hydrological observation data.

[0148] Sensor 1 Sensor 2 Sensor 3 Water gauge 1 A1 B1 C1 Water gauge 2 A2 B2 C2 Water gauge 3 A3 B3 C3

[0149] Among them, A1 represents the observation data of sensor 1 at water gauge 1 of the hydrological station, A2 represents the observation data of sensor 1 at water gauge 2 of the hydrological station, and A3 represents the observation data of sensor 1 at water gauge 3 of the hydrological station. Similarly, B1 represents the observation data of sensor 2 at water gauge 1 of the hydrological station, B2 represents the observation data of sensor 2 at water gauge 2 of the hydrological station, and B3 represents the observation data of sensor 2 at water gauge 3 of the hydrological station.

[0150] In this step, the observation data of different water gauges collected by different sensors 1 to 3 can be separated through the data management system, and the sampling observation data of each water gauge 1 to 3 in the image set can be separated to pave the way for subsequent processing of the neural network model.

[0151] In addition, a specific implementation of the above step S103 includes: using a video intelligent recognition system to analyze the observation data separated in the above table (such as A1 to A3, B1 to B3, C1 to C3), combined with Figure 5 The method flow shown is used to obtain the corresponding water level information of the water gauge.

[0152] Specifically, see Figure 5 As shown, specifically including:

[0153] Step S201: parsing the video stream information to obtain a plurality of photos taken continuously in a photo collection.

[0154] Step S202: Input multiple photos into the self-trained neural network model to obtain training results.

[0155] Specifically, the water gauge target is detected by autonomously training the yolov5 model.

[0156] Step S203: determining whether the plurality of photos contain a water gauge target according to the training result.

[0157] Step S204: If included, obtain the frame height WH of the water gauge target and the water gauge area image.

[0158] If not, return to step S201.

[0159] Step S205: Detect the target character E in the water gauge area image through the self-trained neural network model.

[0160] The autonomously trained neural network model may be a yolov5 neural network model, and the yolov5 neural network model is used to detect the character E in the water gauge area image.

[0161] Step S206: Determine whether the target character E exists.

[0162] Step S207: If the target character E is included, the average height EH of the target character E is calculated.

[0163] The above step S103: calculating the water level information of the water gauge according to the corresponding relationship, the observation data and the water level data, specifically includes:

[0164] Step S208: Obtain the actual length eH of the target character E.

[0165] Step S209: Calculate the actual length of the water gauge according to the corresponding relationship, the average height of the target character E, and the actual length eH, and generate water level information including the actual length of the water gauge.

[0166] The calculation formula is: wH=eH*WH / EH, where wH is the actual length of the water gauge.

[0167] The method provided in this embodiment uses a self-trained neural network model to achieve real-time detection of the target character E in the water gauge area image and calculate the corresponding actual length of the water gauge based on the detected target character E. The above process can be measured and calculated in real time by the edge computing node.

[0168] Furthermore, this method automatically detects water level information from water gauge photographs, reducing the burden of manual monitoring and improving the efficiency and accuracy of water level monitoring. The real-time analysis capabilities of the neural network model enable real-time updates of water level information, which is crucial for hydrological events requiring rapid response. Automated target detection and water level calculations reduce data bias due to human error. The use of the neural network model allows for more accurate identification of water gauge targets and characters, thereby improving the accuracy of water level information. Automated processing reduces the need for human resources and lowers long-term operating costs.

[0169] This method improves data availability and comprehensiveness by combining water level information with observational and water level data. Accurate water level information provides strong support for decision-making in water resource management, flood prevention, and response. The neural network model can adapt to varying lighting conditions and water gauge variations, enhancing the robustness of the system. This method promotes the application and development of computer vision and machine learning technologies in hydrological monitoring.

[0170] Furthermore, before the above step S104, it also includes obtaining a trained neural network model. Specifically, while retaining all the original monitoring data collected by the hydrological station, the Kalman filter is used to consider the uncertainty factors of various errors, and the water level measured data and the video water gauge AI real-time recognition data are combined. Based on the multi-sensor data fusion method of DS (Dempster-Shafer) evidence theory, the evidence grouping and weighted correction of the grouped evidence synthesis are used to eliminate the disturbance of uncertainty factors and the interference of some abnormal data. After the interference is eliminated, data fusion is performed to achieve optimal perception of the station status.

[0171] like Figure 6 Specifically, a specific implementation method of obtaining a trained neural network model includes:

[0172] First, the parameters of the relevant hydrological stations are configured, which include: [water level tx: water level t, flow tx: flow t, flow velocity tx: flow velocity t, rainfall t+1-x: rainfall t+1, evaporation t+1-x: evaporation t+1] [water level t+1, flow t+1, flow velocity t+1]; and [rainfall tx: rainfall t, evaporation tx: evaporation t, temperature tx: temperature t, wind speed tx: wind speed t] [rainfall t+1, evaporation t+1].

[0173] In this embodiment, for the input parameters [rainfall t+1, evaporation t+1] or [water level t+1, flow t+1, flow velocity t+1], the following water level prediction results are obtained.

[0174] The initial neural network model can be an LSTM model. The process of training the LSTM model is as follows:

[0175] Step 1: Get the initial neural network model and the predicted value C at the previous moment t-1 and the auxiliary value h at the previous moment t-1 ;

[0176] Step 2: Obtain the current water level value t from the water level detection system or video monitoring system;

[0177] Step 3: Input the current time value x t and the auxiliary value h at the previous moment t-1 Merge and get the current merge output gate o t ;

[0178] Step 4: Merge the output gate o at the current moment t Perform the following processing:

[0179] Step 4-1: Merge the current moment into the output gate o tThrough the first sigmoid neural network (for example, including FC+ activation function sigmoid), the current forget gate f is obtained t ; The current moment forget gate f t Used to control or abandon certain characteristics;

[0180] Step 4-2: Merge the current moment into the output gate o t Through the second sigmoid neural network (for example, including FC+ activation function sigmoid), the current input gate i is obtained t ;

[0181] Step 4-3: Merge the current moment into the output gate o t Through the first tanh neural network (for example, including FC + activation function tanh1), the current state update value is obtained

[0182] Step 5: Forget the current moment f t and the previous moment prediction value C t-1 Perform dot multiplication to get f t ×C t-1 ;

[0183] Step 6: Update the current state value and the current input gate i t Perform dot multiplication and get

[0184] Step 7: Place the f t ×C t-1 and Add point by point to get the current moment prediction value C t ;

[0185] Step 8: The current prediction value C t Through the second tanh neural network operation (for example, including FC + activation function tanh2), and merged with the current moment output gate o t Perform dot multiplication to get the current auxiliary value h t ;

[0186] Among them, C is the predicted value and h is the auxiliary value.

[0187] The current moment prediction value C t and the current auxiliary value h t The training is performed as an input loop and detected by the cross entropy loss function until the difference between the measured true value and the predicted value reaches a preset value. For example, when the preset value is 0.01, the loop is exited, the training of the neural network model is completed, and the corresponding water level value is output.

[0188] The expression of the above cross entropy loss function is:

[0189]

[0190] Among them, H(p,q) represents the cross entropy loss function, p(x i ) represents the true value of the measurement, q(x i ) represents the predicted value, corresponding to the above C t , n represents the total amount of data, i is a constant, and its value range is 1 to n. t It is C t Auxiliary value, which does not participate in the calculation of the loss function.

[0191] In this embodiment, the trained deep neural network LSTM is used to reduce the amount of data of an output Gate, and is directly multiplied by tanh to obtain the water level value. Analysis of a small amount of data or short-period data is realized at the hydrological station, and the water level data is obtained. Compared with the original hydrological data, the accuracy is improved by about 5cm, and the overall deviation is controlled within the range of 5cm, providing more complete data support for the management of hydrological resources and disaster prevention and reduction.

[0192] The data analysis at the pilot hydrological station is as follows Figure 6 ,Through the trained LSTM, compared with the original detection results, the error range has been significantly narrowed by reducing the model parameters (such as reducing the sigma operation), and the pressure on the use of computing resources has also been reduced.

[0193] See also Figure 7 As shown, the method provided in this embodiment also includes: using a video information intelligent recognition system to analyze and identify acquired video data in real time, and determining alarm information for abnormal on-site behavior. The identified alarm information is then broadcasted via a loudspeaker to alert on-site staff, and simultaneously sent to supervisors and the higher-level hydrological monitoring platform.

[0194] The above-mentioned alarm information includes but is not limited to: abnormal situations such as intrusion of personnel at the hydrological station, falls, fireworks, floating objects, etc. At the same time, the intruder is warned immediately through the loudspeaker and pushed to the management staff in real time.

[0195] The above step S104 monitors the water level status of the hydrological station according to the target water level, including: determining whether the target water level is within the safety line; if so, determining that the water level status of the hydrological station is normal; if not, determining that the water level status is abnormal, generating an alarm message, and sending the alarm message to the alarm system, such as broadcasting through a loudspeaker, or reporting the abnormal water level status information to the hydrological management platform of the server.

[0196] An embodiment of the present invention also provides another alarm method, comprising: obtaining at least one image from a video and preprocessing the image; using an object detection model, invoking a deep learning algorithm to determine whether the value exceeds a set threshold; if not, returning to the first step; otherwise, executing the next step. Alarm information corresponding to the threshold of each model is output, and corresponding voice alarm information is invoked for each alarm type and pushed to an external device, such as a speaker.

[0197] The method provided by the present invention realizes rapid deployment of applications. By deploying a hydrological data management platform and corresponding algorithm models through edge computing devices, it realizes functions such as full-factor perception access of hydrological data, data monitoring, data compilation, equipment condition management, and equipment operation and maintenance. On-duty personnel at the hydrological station can use large screens, PCs, and other means to view and display on-site visualization, realizing "one station, one measurement, one station, one system."

[0198] In addition, this method also supports a variety of wired and wireless communication methods, which can meet the deployment of hydrological stations in various environments, and solve the problems of weak information technology equipment, insufficient perception, insufficient data security protection capabilities and insufficient intelligence level in existing hydrological stations. At the same time, through intelligent identification of video surveillance, it improves the security management capabilities around the hydrological stations and realizes the intelligent improvement of the hydrological stations.

[0199] Specifically, in environments with a wired network, you can use a wired network and wireless LoRa to collect video surveillance data and sensor data, respectively, to improve data security and data transmission stability. In environments without a wired network, you can use a wireless bridge and wireless LoRa to collect video surveillance data and sensor data, respectively, to reduce the risk of data exposure to the public network, improve data security, and save costs. In environments without a wired network, you can use 4G / 5G and wireless LoRa to collect video surveillance data and sensor data, respectively, to improve data security and data parameter stability. In environments without any network, you can use Beidou communication to collect hydrological station sensor data in real time and improve data transmission stability.

[0200] The method provided in this embodiment also supports the setting of the frequency of device data collection, which can be customized according to specific regulatory requirements, thereby reducing the use of unnecessary resources and improving overall utilization efficiency.

[0201] It supports normal communication in disaster environments, automatically identifies and determines the communication environment, automatically switches to Beidou communication, collects hydrological data in real time, and provides data support for handling harsh environments.

[0202] In addition, the use of three systems for independent data collection and analysis can also link information to achieve intelligent management of the hydrological station and improve the management efficiency of the hydrological station.

[0203] This embodiment provides a hydrological monitoring device, such as Figure 8 As shown, the device includes: a receiving module 810, a processing module 820 and a determining module 830. In addition, the device may also include other more or fewer modules, such as a sending module 840, which is not limited in this embodiment.

[0204] The receiving module 810 is configured to receive hydrological observation data of the hydrological station at each sampling moment reported by the sensor cluster, and receive video stream information of the hydrological station reported by the video monitoring system, wherein the video stream information includes a continuous set of photos taken of the hydrological station environment;

[0205] The processing module 820 is used to process the hydrological observation data to obtain the observation data of each sensor and the water level data of the water gauge in the measured hydrological station;

[0206] a determination module 830 configured to parse the video stream information, determine a correspondence between a set of photos in the video stream information and the observation data and the water level data, and calculate the water level information of the water gauge based on the correspondence, the observation data, and the water level data;

[0207] The processing module 820 is further used to input the water level data and the water level information of the water gauge into the trained neural network model, output the target water level, and monitor the water level status of the hydrological station according to the target water level.

[0208] In some optional implementations, the receiving module 810 is specifically configured to receive the hydrological observation data reported by the sensor cluster via a wired manner, or,

[0209] Receive the hydrological observation data reported by the sensor cluster via wireless LoRa, or,

[0210] The hydrological observation data is received via Beidou communication access.

[0211] In some optional implementations, the receiving module 810 is specifically configured to receive the video stream information reported by the video surveillance system through a wired network when wired communication is available;

[0212] Alternatively, in the case of wireless communication, the video stream information reported by the video surveillance system is received via a wireless bridge;

[0213] Alternatively, when wireless communication is available but wireless bridge communication is not available, the video stream information is received via WLAN.

[0214] In some optional embodiments, the processing module 820 is further used to align the hydrological observation data and the multiple photos in the photo set according to each sampling moment and the time point of the continuous photo set taken before processing the hydrological observation data; process the time-aligned hydrological observation data to obtain the observation data of each sensor and the water level data of the measured water gauge, wherein the observation data includes the sampling data measured by each sensor and the sensor ID.

[0215] In some optional embodiments, the processing module 820 is further configured to parse the video stream information to obtain a plurality of photos taken continuously in the photo collection; input the plurality of photos into the autonomously trained neural network model to obtain training results;

[0216] Determine whether the multiple photos contain a water gauge target based on the training results; if so, obtain the frame height of the water gauge target and the water gauge area image; detect whether the water gauge area image contains a target character E through the self-trained neural network model; if the target character E is contained, calculate the average height of the target character E.

[0217] The determination module 830 is further configured to obtain the actual length of the target character E; calculate the actual length of the water gauge according to the corresponding relationship, the average height and the actual length of the target character E, and generate the water level information including the actual length of the water gauge.

[0218] In some optional implementations, the processing module 820 is further configured to obtain the trained neural network model;

[0219] Processing module 820 is specifically used to obtain the initial neural network model and obtain the prediction value C at the previous moment t-1 and the auxiliary value h at the previous moment t-1 ;

[0220] Obtain the current water level value t from the water level detection system or video monitoring system;

[0221] Input the current time value x t and the auxiliary value h at the previous moment t-1 Merge and get the current merge output gate o t ;

[0222] Merge the output gate o at the current moment t Perform the following processing:

[0223] Merge the current moment into the output gate o t Through the first sigmoid neural network, the current forget gate f is obtained t ;

[0224] Merge the current moment into the output gate o t Through the second sigmoid neural network, the current input gate i is obtained t ;

[0225] Merge the current moment into the output gate o t Through the first tanh neural network, the current state update value is obtained

[0226] The current moment forget gate f t and the predicted value C at the previous moment t-1 Perform dot multiplication to get f t ×C t-1 ;

[0227] Update the current state value and the current input gate i t Perform dot multiplication and get

[0228] The f t ×C t-1 and stated Add point by point to get the current moment prediction value C t ;

[0229] The current moment prediction value C t Through the second tanh neural network operation, and merged with the current moment output gate o t Perform dot multiplication to get the current auxiliary value h t ;

[0230] The current moment prediction value C t and the current auxiliary value h t The training is performed as an input loop and detected by the cross entropy loss function until the difference between the measured true value and the predicted value reaches a preset value. Then, the loop is exited and the training of the neural network model is completed to obtain a trained neural network module, such as a trained LSTM model.

[0231] In some optional embodiments, the processing module 820 is specifically used to determine whether the target water level is within the safety line; if so, it is determined that the water level condition of the hydrological station is normal; if not, it is determined that the water level condition is abnormal, an alarm message is generated, and the alarm message is sent to the alarm system through the sending module 840.

[0232] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0233] The hydrological monitoring device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0234] The present invention provides a hydrological monitoring device that integrates sensor data and video stream information to provide multi-dimensional hydrological data analysis, thereby providing a more comprehensive understanding of hydrological phenomena. Combining video surveillance photos with sensor data allows for more accurate determination of water level changes, reducing the potential for errors associated with a single data source.

[0235] Furthermore, data collection and monitoring methods can mitigate issues such as aging equipment, single monitoring methods, inconsistent data collection protocols, and inaccurate data collection. The system can calculate water level information in real time and predict target water levels using a neural network model, helping to promptly detect anomalies and issue early warnings. Automated data processing and water level prediction reduce manual intervention and improve decision-making efficiency. The integration of multi-source data enhances system robustness, ensuring continued operation even in the event of sensor failure. Through intuitive video streaming and precise water level data, users can gain a more intuitive understanding of hydrological conditions, enhancing the user experience.

[0236] Automated monitoring and early warning systems can reduce the frequency of manual inspections, thereby lowering operating costs. At the same time, intelligent identification through video surveillance can enhance safety management capabilities around hydrological stations and achieve intelligent upgrades to hydrological stations.

[0237] The embodiment of the present invention also provides a hydrological monitoring system, the structure of which is as described above. Figure 1 As shown, the system includes at least one edge device, a sensor cluster, a video surveillance system and a server, wherein the sensor cluster, the video surveillance system and the edge device can be connected by wire or wirelessly. Each edge device is used to perform the above Figures 2 to 5 The hydrological monitoring method flow is shown.

[0238] The specific implementation process is described in the above embodiment and will not be repeated here.

[0239] The embodiment of the present invention also provides a computer device having the above Figure 8 The hydrological monitoring device shown.

[0240] See also Figure 9, is a schematic structural diagram of a computer device provided by an optional embodiment of the present invention, the computer device comprising: one or more processors 10, a memory 20, and interfaces for connecting the various components, including high-speed interfaces and low-speed interfaces. The various components are interconnected using different buses and can be installed on a common motherboard or in other ways as needed. The processor can process instructions executed within the computer device, including instructions stored in or on the memory to display graphical information of a GUI on an external input / output device (such as a display device coupled to the interface).

[0241] In some optional embodiments, multiple processors and / or multiple buses can be used with multiple memories and multiple storages if desired. Similarly, multiple computer devices can be connected, with each device providing part of the necessary operations (e.g., as a server array, a group of blade servers, or a multi-processor system). Figure 9 A processor 10 is taken as an example.

[0242] The processor 10 may be a central processing unit, a network processor, or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.

[0243] The memory 20 stores instructions that can be executed by at least one processor 10, so that the at least one processor 10 can implement the hydrological monitoring method shown in the above embodiment.

[0244] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and application programs required for at least one function; the data storage area may store data created based on the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may include a memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0245] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0246] The computer device further includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 9 The bus connection is taken as an example.

[0247] The input device 30 can receive input digital or character information and generate signal input related to user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a trackpad, a touch pad, a pointer, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 can include a display device, an auxiliary lighting device (e.g., an LED), and a tactile feedback device (e.g., a vibration motor). The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display, and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0248] In addition, the computer device also includes a communication interface for the computer device to communicate with other devices or communication networks, such as a sensor cluster or a video surveillance system, or a server via the communication interface.

[0249] Optionally, the above-mentioned computer device can be an edge device or an edge computing node, or an edge computing device.

[0250] It should be understood that the structure of the above server or data center can be different from Figure 9 The computer devices shown have the same structure, and this embodiment does not limit this.

[0251] An embodiment of the present invention also provides a computer-readable storage medium, and the above-mentioned method according to the embodiment of the present invention can be implemented in hardware, firmware, or implemented as a computer code that can be recorded on a storage medium, or downloaded via a network and originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware.

[0252] The storage medium may be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, or a solid-state drive, etc.; further, the storage medium may include a combination of the aforementioned types of memory. It is understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the hydrological monitoring method shown in the above embodiment is implemented.

[0253] The embodiments of the present application may also provide a computer program product, including computer program instructions, which, when executed by a processor, cause the processor to perform the steps in the above method. The computer program product may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the embodiments of the present disclosure, wherein the programming languages ​​include object-oriented programming languages ​​such as Java, C++, etc., and also include conventional procedural programming languages ​​such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0254] The above embodiments are only used to illustrate the technical solutions of the embodiments of the present invention, rather than to limit them. Although the embodiments of the present invention have been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A hydrological monitoring method, characterized in that: Applied to an edge device, the method includes: Receive hydrological observation data of the hydrological station at each sampling moment reported by the sensor cluster, and receive video stream information of the hydrological station reported by the video monitoring system, wherein the video stream information includes a continuous set of photos taken of the hydrological station environment; Aligning the hydrological observation data with the plurality of photos in the photo set according to each sampling moment and the time point of the continuous photo set; Processing the time-aligned hydrological observation data to obtain the hydrological observation data of each sensor and the water level data of the measured water gauge, wherein the observation data includes the sampling data measured by each sensor and the sensor ID; parsing the video stream information, determining a correspondence between a photo set in the video stream information and the observation data and the water level data, and calculating the water level information of the water gauge based on the correspondence, the observation data and the water level data; Input the water level data and the water level information of the water gauge into the trained neural network model, output the target water level, and monitor the water level status of the hydrological station according to the target water level; The method for obtaining the trained neural network model includes the following steps: Get the initial neural network model and the predicted value of the previous moment and the auxiliary value at the previous moment ; Obtain the current water level value t from the water level detection system or video monitoring system; Enter the current time value and the auxiliary value at the previous moment Merge to get the merge output gate at the current moment ; Merge the output gate at the current moment Perform the following processing: Merge the current moment into the output gate Through the first sigmoid neural network, the current moment forget gate is obtained ; Merge the current moment into the output gate Through the second sigmoid neural network, the current input gate is obtained ; Merge the current moment into the output gate Through the first tanh neural network, the current state update value is obtained ; The current moment is forgotten and the predicted value at the previous moment Perform dot multiplication and get ; Update the current state value and the current input gate Perform dot multiplication and get ; The and stated Add point by point to get the current moment prediction value ; The current moment prediction value Through the second tanh neural network operation, and merge the output gate with the current moment Perform dot multiplication to get the current auxiliary value ; The current moment prediction value and the current auxiliary value The training is performed as an input loop and detected by the cross entropy loss function until the difference between the measured true value and the predicted value reaches a preset value. Then the loop is exited and the neural network model training is completed.

2. The method according to claim 1, characterized in that The receiving of hydrological observation data of the hydrological station at each sampling moment reported by the sensor cluster includes: receiving the hydrological observation data reported by the sensor cluster via a wired manner, or, Receive the hydrological observation data reported by the sensor cluster via wireless LoRa, or, The hydrological observation data is received via Beidou communication access.

3. The method according to claim 1, characterized in that The receiving of video stream information from the hydrological station reported by the video monitoring system includes: In the case of wired communication, receiving the video stream information reported by the video surveillance system through a wired network; Alternatively, in the case of wireless communication, the video stream information reported by the video surveillance system is received via a wireless bridge; Alternatively, when wireless communication is available but wireless bridge communication is not available, the video stream information is received via WLAN.

4. The method according to any one of claims 1 to 3, characterized in that After parsing the video stream information, the method includes: parsing the video stream information to obtain a plurality of photos taken continuously in the photo collection; Inputting the plurality of photos into a self-trained neural network model to obtain training results; Determining whether the plurality of photos contain a water gauge target according to the training result; If included, get the frame height of the water gauge target and the water gauge area image; Detecting whether the water gauge area image contains the target character E through the self-trained neural network model; If the target character E is included, then the average height of the target character E is calculated; Calculating the water level information of the water gauge according to the corresponding relationship, the observation data, and the water level data includes: Obtain the actual length of the target character E; The actual length of the water gauge is calculated according to the corresponding relationship, the average height of the target character E and the actual length, and the water level information including the actual length of the water gauge is generated.

5. The method according to claim 1, wherein The step of monitoring the water level of the hydrological station according to the target water level includes: Determining whether the target water level is within a safety line; If yes, it is determined that the water level condition at the hydrological station is normal; If not, it is determined that the water level condition is abnormal, an alarm message is generated, and the alarm message is sent to an alarm system.

6. A hydrological monitoring device, characterized in that: The device comprises: A receiving module, configured to receive hydrological observation data of the hydrological station at each sampling moment reported by the sensor cluster, and receive video stream information of the hydrological station reported by the video monitoring system, wherein the video stream information includes a continuous set of photos taken of the hydrological station environment; a processing module configured to align the hydrological observation data with a plurality of photos in the photo set according to each sampling moment and a time point of the continuous photo set before processing the hydrological observation data; and process the time-aligned hydrological observation data to obtain the hydrological observation data of each sensor and the water level data of the measured water gauge, wherein the observation data includes the sampling data measured by each sensor and the sensor ID; a determination module, configured to parse the video stream information, determine a correspondence between a set of photos in the video stream information and the observation data and the water level data, and calculate the water level information of the water gauge based on the correspondence, the observation data, and the water level data; The processing module is further configured to input the water level data and the water level information of the water gauge into a trained neural network model, output a target water level, and monitor the water level condition of the hydrological station according to the target water level; The processing module is specifically configured to obtain the trained neural network model by the following method: Get the initial neural network model and the predicted value of the previous moment and the auxiliary value at the previous moment ; Obtain the current water level value t from the water level detection system or video monitoring system; Enter the current time value and the auxiliary value at the previous moment Merge to get the merge output gate at the current moment ; Merge the output gate at the current moment Perform the following processing: Merge the current moment into the output gate Through the first sigmoid neural network, the current moment forget gate is obtained ; Merge the current moment into the output gate Through the second sigmoid neural network, the current input gate is obtained ; Merge the current moment into the output gate Through the first tanh neural network, the current state update value is obtained ; The current moment is forgotten and the predicted value at the previous moment Perform dot multiplication and get ; Update the current state value and the current input gate Perform dot multiplication and get ; The and stated Add point by point to get the current moment prediction value ; The current moment prediction value Through the second tanh neural network operation, and merge the output gate with the current moment Perform dot multiplication to get the current auxiliary value ; The current moment prediction value and the current auxiliary value The training is performed as an input loop and detected by the cross entropy loss function until the difference between the measured true value and the predicted value reaches a preset value. Then the loop is exited and the neural network model training is completed.

7. A computer device, characterized in that: comprising a memory and a processor, wherein the memory and the processor are connected; The memory stores computer instructions, and the processor executes the hydrological monitoring method according to any one of claims 1 to 5 by executing the computer instructions.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the hydrological monitoring method according to any one of claims 1 to 5.

Images