Edge computing terminal and system and gate control method

Abstract

The invention discloses an edge computing terminal and system and a gate control method, and relates to the technical field of automation control, and the terminal comprises a pluggable I / O part which is used for connecting a sensor group and an external execution mechanism; a heterogeneous calculation module; the heterogeneous computing module comprises an application processor unit and a real-time microcontroller unit; the hybrid energy management module is used for supplying power by utilizing self-powered equipment and triggering an external power-off instruction to control the real-time microcontroller unit to control the dam gate to be closed within a preset time period through the dam gate control component when the external power supply is detected to be powered off; high-precision local water level prediction is achieved through the application processor unit, and quick response of gate control is achieved through the real-time microcontroller unit; power supply during external power failure is guaranteed through the hybrid energy management module, so that the real-time microcontroller unit can take over control in time during external power failure, and automation and safety of hydropower station operation are improved.

Description

Technical Field

[0001] This invention relates to the field of automation control technology, and in particular to an edge computing terminal, system, and gate control method. Background Technology

[0002] In the operation and management of the forebay dams of small and medium-sized unmanned hydropower stations, accurate prediction of the relationship between water level and reservoir capacity is crucial for ensuring flood control safety, optimizing water storage and power generation, and realizing intelligent gate control. However, the water level monitoring methods in the centralized control centers of related technologies mostly rely on a small amount of telemetry data from a single sensor (such as a float level gauge), lacking the ability to predict future water level trends. This results in delayed response in sudden events such as rainstorms and floods, and a lack of foresight in control decisions. Furthermore, in the event of a power outage and a sudden flood upstream, it is difficult for the local area to drive a 10kW gate motor to complete a full-stroke closure, posing a significant safety hazard.

[0003] Therefore, how to achieve high-precision and low-latency local water level prediction and gate control, and improve the automation and safety of hydropower station operation, is an urgent problem to be solved. Summary of the Invention

[0004] The purpose of this invention is to provide an edge computing terminal, system, and gate control method to improve the automation and safety of hydropower station operation through high-precision and low-latency local water level prediction and gate control.

[0005] To address the aforementioned technical problems, the present invention provides an edge computing terminal, comprising:

[0006] Pluggable I / O components for connecting sensor arrays and external actuators; wherein the sensor arrays include level gauges and flow meters; and the external actuators include dam gate control components.

[0007] A heterogeneous computing module; the heterogeneous computing module includes an application processor unit and a real-time microcontroller unit. The application processor unit is used to generate a gate opening sequence corresponding to water level prediction information within a preset future time period based on sensor data collected by the sensor group, and to generate an opening control command corresponding to the gate opening sequence. The real-time microcontroller unit is used to control the dam gate control components to adjust the opening of the dam gate according to the opening control command.

[0008] The hybrid energy management module is used to connect external power supply equipment and self-powered equipment; when an external power outage is detected, power is supplied using the self-powered equipment, and an external power outage command is triggered to control the real-time microcontroller unit to control the dam gate to close within a preset time period through the dam gate control component.

[0009] On the other hand, the heterogeneous computing module also includes: shared memory to which both the application processor unit and the real-time microcontroller unit are connected;

[0010] The application processor unit is configured to generate an emergency control instruction based on the external power failure instruction; write the emergency control instruction into the shared memory and trigger a high-priority interrupt;

[0011] The real-time microcontroller unit is used to read the emergency control instruction from the shared memory according to the high-priority interrupt; and according to the emergency control instruction, to control the dam gate to close within the preset time period through the dam gate control component.

[0012] On the other hand, the application processor unit is specifically used to generate water level prediction information for the preset future time period using a long short-term memory network model based on the sensor data; generate the gate opening sequence using a gate-hydraulic coupling model based on the water level prediction information; and generate the opening control command based on the gate opening sequence.

[0013] On the other hand, the long short-term memory network model includes a high-precision prediction model and a lightweight prediction model; the step of generating water level prediction information for the preset future time period using the long short-term memory network model based on the sensor data includes:

[0014] When the preset triggering conditions are not met, the lightweight prediction model generates water level prediction information for the preset future time period based on the target sensor data; wherein, the target sensor data is data collected by some sensors in the sensor group;

[0015] When a preset trigger condition is met, water level prediction information for the preset future time period is generated using the high-precision prediction model based on the sensor data; wherein, the preset trigger condition includes at least one of the following: water level change rate exceeds a change rate threshold, water level prediction error reaches an error threshold, a preset external control command is received, and data collected by key sensors in the sensor group is abnormal.

[0016] On the other hand, the external actuator also includes cleaning equipment;

[0017] The application processor unit is further configured to detect whether the target control strategy is met based on the water level prediction information and the sensor data; if so, generate a preset opening control sequence and / or a preset cleaning instruction corresponding to the target control strategy to control the real-time microcontroller unit to drive the dam gate control component and / or the cleaning equipment; wherein, the target control strategy is any preset control strategy.

[0018] On the other hand, the self-powered equipment includes a supercapacitor, a lithium battery, and a diesel generator;

[0019] The hybrid energy management module is specifically used to switch to the supercapacitor, the lithium battery, and the diesel generator for power supply when an external power outage is detected.

[0020] On the other hand, the terminal also includes: a communication unit; the communication unit includes a mobile communication module and a satellite communication module;

[0021] The heterogeneous computing module is used to send alarm information through the satellite communication module when the signal quality of the mobile communication module does not meet the communication requirements.

[0022] On the other hand, the sensor group also includes at least one of a rain gauge, a weather instrument, and a sediment concentration meter; the sensor group is mounted on the same mast;

[0023] The pluggable I / O component is also used to connect to the AI ​​vision PTZ camera; the external actuator also includes a cleaning device; the application processor unit is also used to generate a cleaning control command based on the cleaning event information output by the AI ​​vision PTZ camera, so as to control the real-time microcontroller unit to drive the cleaning device to perform cleaning.

[0024] The present invention also provides an edge computing system, comprising: an edge computing terminal as described above and a sensor group connected to the edge computing terminal.

[0025] Furthermore, the present invention also provides a gate control method, applied to an edge computing terminal as described above, comprising:

[0026] The system utilizes pluggable I / O components to receive sensor data collected by a sensor array, which includes a water level gauge and a flow rate gauge.

[0027] Using the application processor unit, based on the sensor data, a gate opening sequence corresponding to the water level prediction information within a preset future time period is generated, and an opening control command corresponding to the gate opening sequence is generated.

[0028] Using a real-time microcontroller unit, the dam gate control components are controlled to adjust the opening of the dam gate according to the opening control command;

[0029] Using the real-time microcontroller unit, based on the external power outage command triggered by the hybrid energy management module, the dam gate control component controls the dam gate to close within a preset time period; wherein, the hybrid energy management module is used to provide power using its self-powered equipment when an external power outage is detected, and to trigger the external power outage command.

[0030] The present invention provides an edge computing terminal, comprising: a pluggable I / O component for connecting a sensor group and an external actuator; wherein the sensor group includes a water level gauge and a flow rate gauge; the external actuator includes a dam gate control component; a heterogeneous computing module; the heterogeneous computing module includes an application processor unit and a real-time microcontroller unit, the application processor unit being used to generate a gate opening sequence corresponding to water level prediction information within a preset future time period based on sensor data collected by the sensor group, and to generate an opening control command corresponding to the gate opening sequence; the real-time microcontroller unit being used to control the dam gate control component to adjust the opening of the dam gate according to the opening control command; and a hybrid energy management module being used to connect an external power supply device and a self-powered device; when an external power supply failure is detected, power is supplied using the self-powered device, and an external power failure command is triggered to control the real-time microcontroller unit to control the dam gate to close within a preset time period through the dam gate control component.

[0031] As can be seen, this invention, through the setting of heterogeneous computing modules in the edge computing terminal, utilizes the application processor unit to achieve high-precision local water level prediction and the real-time microcontroller unit to achieve rapid response in gate control; a hybrid energy management module ensures power supply during external power outages, enabling the real-time microcontroller unit to promptly take over control and drive the dam gates to close within a preset time period, thus improving the automation and safety of hydropower station operation. Furthermore, this invention also provides an edge computing system and gate control method, which similarly possess the aforementioned beneficial effects. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0033] Figure 1 This is a structural block diagram of an edge computing terminal provided in an embodiment of the present invention;

[0034] Figure 2 This is a schematic diagram of an edge computing system provided in an embodiment of the present invention;

[0035] Figure 3 This is a diagram illustrating an integrated multi-parameter sensor mast provided in an embodiment of the present invention.

[0036] Figure 4 This is an exploded hardware view of an edge computing terminal provided in an embodiment of the present invention;

[0037] Figure 5 This is a schematic diagram of the software flow of an edge computing terminal provided in an embodiment of the present invention;

[0038] Figure 6 This is a flowchart of a gate control method provided in an embodiment of the present invention. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] Please refer to Figure 1 , Figure 1 This is a structural block diagram of an edge computing terminal provided in an embodiment of the present invention. The edge computing terminal may include:

[0041] Pluggable I / O component 11 is used to connect sensor group and external actuator; wherein, sensor group includes water level gauge and flow rate gauge; external actuator includes dam gate control component;

[0042] Heterogeneous computing module 12; Heterogeneous computing module 12 includes an application processor unit and a real-time microcontroller unit. The application processor unit is used to generate a gate opening sequence corresponding to the water level prediction information within a preset future time period based on the sensor data collected by the sensor group, and to generate an opening control command corresponding to the gate opening sequence. The real-time microcontroller unit is used to control the dam gate control components to adjust the opening of the dam gate according to the opening control command.

[0043] The hybrid energy management module 13 is used to connect external power supply equipment and self-powered equipment; when an external power supply failure is detected, power is supplied using the self-powered equipment, and an external power failure command is triggered to control the real-time microcontroller unit to control the dam gate to close within a preset time period through the dam gate control component.

[0044] It is understood that the pluggable I / O (input / output) component 11 in this embodiment can provide various industrial interface slots for connecting sensor groups and external actuators. The specific structure and type of the pluggable I / O component 11 in this embodiment can be customized by the designer according to the practical scenario and user needs. For example, the pluggable I / O component 11 can not only connect sensor groups and external actuators, but also other devices, such as AI (artificial intelligence) vision PTZ cameras and monitoring devices for gate opening, motor current, battery voltage, etc. Figure 2As shown, the pluggable I / O component 11 (pluggable I / O backplane) may include: 16 analog inputs (AI, for connecting various sensor signals), 32 digital inputs / outputs (DI / DO, for status monitoring and relay control), 4 CAN-FD (a communication bus protocol) buses, 2 EtherCAT (Ethernet Control Automation Technology) master interfaces (for controlling permanent magnet synchronous gate motors, etc.), and 2 isolated RS485 (a serial bus standard) interfaces; all interfaces adopt optocoupler isolation and π-type filtering to improve electromagnetic compatibility.

[0045] Correspondingly, the specific number and type of devices in the sensor group and external actuator connected to the pluggable I / O component 11 in this embodiment can be set by the designer according to the practical scenario and user needs. For example, the sensor group can include not only water level gauges (such as millimeter-wave radar water level gauges) and flow meters (such as ultrasonic flow meters), but also at least one of rain gauges (such as tipping bucket rain gauges), meteorological instruments (such as six-element meteorological instruments), and sediment concentration meters (such as laser sediment concentration meters). For example, the sensor group can include water level gauges, flow meters, rain gauges, meteorological instruments, and sediment concentration meters, and all sensors in the sensor group can be installed on the same mast (such as...). Figure 3 As shown, the sensors can be integrated onto a 1.5-meter mast to avoid the difficulties in installation and maintenance and the challenges in data fusion caused by the dispersed arrangement of sensors in related technologies. This enables multi-parameter collaborative sensing, reduces wiring, and improves installation efficiency and data consistency. The sensor group on the mast can adopt a foldable quick-release structure and a unified power and data interface, supporting RS485-Modbus-RTU (a type of serial communication) cascaded communication. External actuators may include dam gate control components, such as gate motors (e.g., Figure 4 The permanent magnet synchronous gate motor in the middle); may also include cleaning equipment (such as Figure 4 (The hydraulic grab bucket cleaning machine in this embodiment). This implementation makes no restrictions on this.

[0046] Correspondingly, when the pluggable I / O component 11 is connected to the AI ​​vision PTZ camera, the AI ​​vision PTZ camera can be integrated with the sensor group on the same mast, or it can be installed independently in other locations. The AI ​​vision PTZ camera can run target detection algorithms, such as the YOLOv8-nano (a target detection algorithm) model, to identify in real time relevant information (i.e., information on events to be cleaned) of events that require cleaning, such as floating objects, siltation, and gate jamming.

[0047] For example, such as Figure 4As shown, the integrated multi-parameter sensor mast can be connected to the pluggable I / O backplane of the edge computing terminal (edge ​​computing smart terminal) via a hybrid cable (integrating power and RS485 data cables). The mast, approximately 1.5 meters high, can be fixed to the forebay bank or near the dam using a quick-release flange base and can be manually folded down for maintenance. From top to bottom, the mast integrates a millimeter-wave radar level gauge, an ultrasonic Doppler current meter, a tipping bucket rain gauge, a six-element meteorological instrument, and a laser sediment concentration meter. All sensors are cascaded via RS485 interfaces using the Modbus-RTU protocol and ultimately connected to a single RS485 interface on the pluggable I / O backplane. The AI ​​vision PTZ camera can be independently installed on a pole aligned with the forebay inlet and gate area. It is powered via PoE (Power Over Ethernet) and connected to an Ethernet port on a pluggable I / O backplane. The camera has a built-in lightweight AI chip that directly runs the trained YOLOv8-nano model. It reports the identified "floating objects," "siltation," and "abnormal gate displacement" cleaning events, along with their coordinates and confidence levels, as structured data (such as JSON format) to the edge computing terminal via Ethernet. In other words, the cleaning event information can include the cleaning event, coordinates, and confidence level in a preset data structure.

[0048] Correspondingly, the application processor unit can also be used to generate cleaning control commands based on the cleaning event information output by the AI-powered vision PTZ camera, so as to control the real-time microcontroller unit to drive the cleaning equipment to perform cleaning.

[0049] It should be noted that the hybrid energy management module 13 in this embodiment can be used to connect external power supply equipment and self-powered equipment. When the external power supply fails, the self-powered equipment can provide power and trigger an external power failure command to control the real-time microcontroller unit to control the dam gate to close within a preset time period (e.g., 80 seconds) via the dam gate control component. The specific number and type of external power supply equipment and self-powered equipment can be set by the designer according to the practical scenario and user needs. For example, external power supply equipment may include photovoltaic power generation equipment and / or wind power generation equipment; self-powered equipment may include supercapacitors and lithium batteries (such as lithium iron phosphate batteries), and may also include diesel generators. For example, external power supply equipment may include photovoltaic power generation equipment and wind power generation equipment, while self-powered equipment may include supercapacitors, lithium batteries, and diesel generators, realizing a hybrid energy system of "photovoltaic + wind turbine + lithium battery + supercapacitor + diesel generator". When the external power supply (photovoltaic and wind turbine) is detected to be interrupted, the hybrid energy management module 13 can switch to use supercapacitors, lithium batteries, and diesel generators for power supply, realizing a three-level seamless switching of "supercapacitor → lithium battery → diesel generator" with zero milliseconds. That is, supercapacitors are used to cope with millisecond-level power surges, lithium batteries are used to support minute-level endurance, and diesel generators are used to ensure long-term power supply, completely solving the safety hazard of the gate not being able to close in an emergency under power failure, and improving the system's energy self-governance capability.

[0050] Correspondingly, the specific structure of the hybrid energy management module 13 in this embodiment can be customized by the designer according to the practical scenario and user needs. For example, the external power supply equipment may include photovoltaic power generation equipment (such as...). Figure 4 1.5 kW photovoltaic panels) and wind power generation equipment (such as Figure 4 The 500 W vertical axis fan in the middle), self-powered equipment includes supercapacitors (such as Figure 4 The 200 Wh supercapacitor module), lithium battery (such as...) Figure 4 When the system includes a 6 kWh lithium battery pack and a diesel generator, the hybrid energy management module 13 may include a photovoltaic controller, a wind power controller, a bidirectional DC-DC converter (Buck-Boost), a battery management system (BMS), and a generator self-starting controller. The battery management system can continuously monitor the states of various energy sources (battery state of charge (SOC) and state of health (SOH)), load power demand, and grid status through energy dispatching algorithms. Under normal conditions, the photovoltaic and wind power controllers supply power to the load and charge the lithium battery using photovoltaic and wind power generation equipment. In the event of an external power outage, the bidirectional DC-DC converter utilizes a supercapacitor to take over the entire load within milliseconds, while simultaneously triggering an external power outage command to close the dam gate. The lithium battery takes over within seconds to ensure continuous power supply. If a longer power outage is predicted, the diesel generator is automatically started by the generator self-starting controller.

[0051] Understandably, due to the response delay between water level prediction inference and real-time gate control in related technologies, it is difficult to achieve millisecond-level linkage in emergency situations such as power outages. In this embodiment, a heterogeneous computing module 12 (such as a dual-core heterogeneous computing module) can be used. The application processor unit in the heterogeneous computing module 12 is responsible for water level prediction inference, and the real-time microcontroller unit (such as a microcontroller unit MCU) in the heterogeneous computing module 12 is responsible for real-time interlocking control. This allows the opening control command obtained by the application processor unit through water level prediction inference to be read and executed by the real-time microcontroller unit within a preset response time (such as 1ms). Furthermore, when the external power supply fails, the real-time microcontroller unit can immediately take over control and drive the dam gate to close within a preset time period. This achieves seamless integration of AI (artificial intelligence) decision-making and real-time control, meeting the high real-time and high reliability control requirements in unattended scenarios.

[0052] Correspondingly, the specific types of the application processor unit and real-time microcontroller unit in this embodiment can be set by the designer according to the practical scenario and user needs. For example, the application processor unit can be a GPU (Graphics Processing Unit). Figure 4 The Cortex-A78+ GPU (one type of GPU) is used; the real-time microcontroller unit can be an MCU, such as... Figure 4 The Cortex-M7 MCU (one type of MCU) is used; the application processor unit or real-time microcontroller unit may also be a CPU (central processing unit) or other processing devices, and this embodiment does not impose any restrictions on this.

[0053] Furthermore, to reduce the response latency between the application processor unit and the real-time microcontroller unit, the heterogeneous computing module 12 in this embodiment may further include: shared memory (e.g., where both the application processor unit and the real-time microcontroller unit are connected) Figure 2 The system utilizes shared memory (DDR RAM). The application processor unit generates emergency control instructions based on external power-off commands; writes these instructions to the shared memory and triggers a high-priority interrupt; the real-time microcontroller unit reads the emergency control instructions from the shared memory based on the high-priority interrupt; and controls the dam gate to close within a preset time period using the dam gate control components based on the emergency control instructions. For example, a circular buffer can be set within the shared memory. The application processor unit can write control instructions (such as opening control instructions and emergency control instructions) to the circular buffer and trigger an interrupt. Through a dynamic adjustment algorithm of the circular buffer in the shared memory and the interrupt priority, the control instructions from the application processor unit can be read and executed by the real-time microcontroller unit within <1 ms.

[0054] For example, the edge computing terminal may also include a heat dissipation housing, which employs a finned structure (such as...). Figure 2 The all-aluminum finned heat sink shown has interface openings, an IP67 protection rating, and an operating temperature range of -40℃ to +70℃, making it suitable for the humid, high and low temperature, and dusty environments of the gate hoisting room in the forebay of hydropower stations. The terminal has a standard 2U chassis size, facilitating wall mounting or cabinet installation in space-constrained gate hoisting rooms. The heterogeneous computing module 12 can be housed on a motherboard within the heat sink; the application processor unit can be a SoC (System-on-a-Chip) equipped with an 8-core Cortex-A78 CPU and a Mali-G610 GPU, providing 6 TOPS (processor power unit) of computing power and running the Ubuntu Core operating system; the real-time microcontroller unit can use a Cortex-M7 core, independently running a real-time operating system (RTOS) to handle emergency interlock control logic at the <1ms level, such as emergency control commands during power outages. The application processor unit and the real-time microcontroller unit share data through a shared memory and utilize a priority-based hardware interrupt channel. When a "gate jamming" or "over-alert" event is detected by the water level prediction module, the application processor unit writes the result to the circular buffer of shared memory and notifies the real-time microcontroller unit by triggering a high-priority interrupt. The real-time microcontroller unit can read the data and execute the preset safety interlock action within 1 ms.

[0055] In this embodiment, the specific method by which the application processor unit generates a gate opening sequence corresponding to the water level prediction information within a preset future time period based on the sensor data collected by the sensor group, and generates the opening control command corresponding to the gate opening sequence, can be set by the designer according to the practical scenario and user needs. For example, the application processor unit may specifically generate water level prediction information (such as a water level prediction curve) within a preset future time period using a Long Short-Term Memory (LSTM) network model based on the sensor data, so as to use the LSTM model for water level prediction; generate a gate opening sequence using a gate-hydraulic coupling model based on the water level prediction information; and generate an opening control command based on the gate opening sequence, so as to use the opening control command to control the real-time microcontroller unit to adjust the opening of the dam gate.

[0056] Furthermore, to reduce the consumption of computing resources and avoid resource waste, this embodiment can adopt an event-triggered model reduction mechanism. For example, the Long Short-Term Memory (LSTM) network model can include a high-precision prediction model (such as a 6-layer LSTM unit) and a lightweight prediction model (such as a 3-layer LSTM unit, a linear regression model, or an autoregressive moving average (ARIMA) model). The above-mentioned generation of water level prediction information for a preset future time period using the LSTM network model based on sensor data can include: when the preset triggering conditions are not met, generating water level prediction information for a preset future time period using a lightweight prediction model based on target sensor data; wherein, the target sensor data is data collected by some sensors in the sensor group; when the preset triggering conditions are met, generating water level prediction information for a preset future time period using a high-precision prediction model based on sensor data; wherein, the preset triggering conditions include at least one of the following: water level change rate exceeding a change rate threshold, water level prediction error reaching an error threshold, receiving a preset external control command, and abnormal data collected by key sensors in the sensor group.

[0057] In other words, through the collaborative design of a high-precision prediction model and a lightweight prediction model, for example, the high-precision prediction model can include a complete LSTM model and a multi-source fusion model (all sensor inputs, 6-layer LSTM units); the lightweight prediction model can adopt a simplified linear regression / ARIMA model (only 2-3 sensor inputs such as water level and rainfall), which can dynamically adjust the computational complexity according to the system state and requirements. The lightweight prediction model is used in a stable state, and the high-precision prediction model is only activated when the preset trigger conditions are triggered.

[0058] Correspondingly, the specific settings for the aforementioned preset trigger conditions can be configured by the designers. For example, switching to the high-precision prediction model when any preset trigger condition is met can be implemented. These trigger conditions might include: the water level change rate exceeding a threshold (e.g., 0.5 cm / min); the water level prediction error (i.e., the error between the water level predicted by the lightweight prediction model and the actual value) reaching an error threshold (e.g., 5 cm); receiving preset external control commands (e.g., commands triggered by external events such as "rainstorm warning" and "flood discharge notice," as well as gate control commands or mode switching commands); and abnormal data collected by key sensors in the sensor group (e.g., sudden changes or failures). For instance, in a typical scenario, the system is in a stable state 95% of the time → using the lightweight prediction model; and in the state triggered by preset trigger conditions 5% of the time → using the high-precision prediction model. The average CPU utilization of the application processor unit is 0.95 × 5% + 0.05 × 30% = 6.25%, which reduces CPU load by approximately 80% compared to continuously running the high-precision prediction model (30% utilization).

[0059] Correspondingly, if the preset triggering conditions are not met within a consecutive preset downgrade time (e.g., 10 minutes), the system can switch from using a high-precision prediction model to using a lightweight prediction model to generate water level prediction information for a preset future time period. For example, in a scenario of a sudden rainstorm event: 1. 00:00-06:00 (calm period): Use a lightweight ARIMA model, predicting once every 5 minutes; CPU usage: 4%; only water level and rainfall are monitored. 2. 06:05 (event trigger): Rainfall sensor: rainfall > 10 mm within 5 minutes; water level change rate: ΔH > 2 cm / 5 min; trigger event: switch to a high-precision LSTM model. 3. 06:05-08:00 (rainstorm period): activate all sensors; LSTM model predicts once every 1 minute; CPU usage rises to 28%; real-time calculation of gate opening sequence. 4. 08:05 (Event ends): Rainfall drops to 0, water level change rate <0.2cm / min; after continuous monitoring for 10 minutes, switch back to ARIMA model; CPU usage recovers to 5%.

[0060] Correspondingly, in other embodiments, a multi-level triggering mechanism with more models can be used. For example, the long short-term memory network model may include a high-precision prediction model (such as multiple model integration), a medium-precision model (such as a 6-layer LSTM), and a lightweight prediction model (such as a 3-layer LSTM). The application processor unit can adopt a similar approach to the above dual-model collaboration to perform collaborative switching of these three models. This embodiment does not impose any restrictions on this.

[0061] Furthermore, in this embodiment, the application processor unit can also control the real-time microcontroller unit to drive the dam gate control components to adjust the dam gate according to a preset control strategy (i.e., preset control strategy) and a corresponding preset opening control sequence. For example, the external actuator can also include a cleaning device; the application processor unit can also be used to detect whether the target control strategy is met based on water level prediction information and sensor data; if so, it generates a preset opening control sequence and / or a preset cleaning instruction corresponding to the target control strategy to control the real-time microcontroller unit to drive the dam gate control components and / or the cleaning device; wherein, the target control strategy is any preset control strategy. For example, when the preset control strategy of "predicted water level exceeds warning water level by 0.3 meters" and "sediment concentration > 5%" is triggered, a preset "pre-discharge + flushing" opening control sequence is automatically generated so that the real-time microcontroller unit controls the dam gate to open according to a specific curve through the EtherCAT bus and starts the hydraulic grab cleaning machine for cleaning.

[0062] It should be noted that the edge computing terminal provided in this embodiment may further include: a communication unit; the communication unit includes a mobile communication module (such as a 4G and / or 5G communication module) and a satellite communication module (such as a BeiDou short message module); wherein, the heterogeneous computing module 12 (such as an application processor unit) is used to send alarm information through the satellite communication module when the signal quality of the mobile communication module does not meet the communication requirements. Figure 4 As shown, the edge computing terminal can have a built-in communication unit, including a 4G or 5G communication module as the primary remote channel, a BeiDou short message module as a backup channel, and an eSIM (digital SIM card). The software layer enables intelligent switching: when the signal quality of 4G / 5G continuously monitors and fails to meet communication requirements (e.g., when the packet loss rate consistently exceeds 1% or the connection is completely lost), it automatically sends the highest-level alarm information via BeiDou short message. The communication unit can also include two fiber optic Ethernet ports (e.g., gigabit fiber optic Ethernet ports) for accessing a dual-redundant fiber optic ring network constructed within the station, such as between points like the gate hoist room, gate motor, cleaning machine, and sensor mast. The edge computing terminal connects as a node in the ring network, ensuring that any single-point fiber optic cable failure does not affect communication.

[0063] For example, the software of an edge computing terminal can adopt a layered containerized architecture, meaning that the application processor unit can run a containerized operating system, such as... Figure 5 As shown, the underlying operating system runs a customized Ubuntu Core system on the application processor unit, and deploys the lightweight Kubernetes (an open-source container orchestration platform) distribution K3s to achieve containerized orchestration and management of services. Middleware and data bus: A real-time data bus based on DDS (Data Distribution Service) is deployed as the system hub. QoS (Quality of Service) policies are defined for different types of data; for example, sensor data is set to "Best-Effort," while emergency control commands are set to "Highest Reliability + Lowest Latency," ensuring that the end-to-end latency of critical commands is <5 ms. Data acquisition and driving containers are responsible for protocol parsing (such as Modbus and EtherCAT) with all sensors and external actuators, and publishing standardized data to the DDS bus.

[0064] Digital Twin and AI Prediction Container: Loads an LSTM model trained offline and fine-tuned online, takes as input the water level, rainfall, and flow velocity sequences of the past 2 hours, and outputs water level prediction information for the next 60 minutes (i.e., the preset future time period). Runs a simplified gate-hydraulic coupling model (such as a reduced-order version of the Saint-Venant equations), combining predicted water level and current flow rate to solve for the optimal gate opening sequence in real time; employs a triggered model reduction mechanism, only initiating high-precision model calculations when the water level change rate exceeds a threshold or a control command is received, and using a low-order approximation model at other times, keeping the average CPU utilization below 10%, and supporting independent operation 7×24 hours a day even without network access.

[0065] Collaborative control strategy container: Subscribes to all data and prediction results on the DDS bus and executes the corresponding preset control strategy. For example: When the events of "predicted water level > warning water level 0.3 meters" and "sediment concentration > 5%" are triggered simultaneously, an automatic "pre-discharge + flushing" control sequence is generated: the gate (31) is controlled to open according to a specific curve through the EtherCAT bus, and the hydraulic grab bucket cleaning machine (32) is started in conjunction. When the "external power interruption" signal (acquired through DI) is triggered, the highest priority "emergency gate closure" command is immediately sent to the MCU, and the hybrid energy switching process is started.

[0066] Log and communication management container: Responsible for compressing and storing local logs and key image / video clips identified by AI; automatically uploading to the cloud platform when the public network (4G / 5G) is available; when the public network is interrupted, automatically switching to BeiDou short message service to transmit the most important alarm information and key data (such as "gates are fully closed") back to the remote control center. Figure 4 The SCADA (Supervisory Control and Data Acquisition) system in China.

[0067] For example, taking the scenario of "emergency flood control gate closure" as an example: 1. Sensing and prediction: Heavy rain causes water levels to rise rapidly, and the masts of the sensor group report data in real time; the LSTM model in the digital twin container predicts that the water level will exceed the warning line by 0.5 meters in the next 30 minutes. 2. Decision and local control: The collaborative control strategy container receives the warning prediction event and, combined with the current power grid status (assuming that the external power is interrupted at this time), immediately generates an emergency control command for "emergency gate closure". 3. Heterogeneous collaborative execution: This command is issued with the highest priority through the DDS bus. The application processor unit learns of the command within 1 ms through shared memory and interrupts and executes it immediately: ① Disconnect the main circuit contactor and switch to lithium battery power supply; ② Send a closed-loop position control command to the permanent magnet synchronous gate motor through the EtherCAT bus, driving the dam gate to close at the fastest safe speed. 4. Energy guarantee: During the entire 80-second gate closure process, the hybrid energy management module 13 ensures that the lithium battery provides stable power, and the supercapacitor smooths the instantaneous power surge when the motor starts.

[0068] In this embodiment, the present invention utilizes the heterogeneous computing module 12 in the edge computing terminal to achieve high-precision local water level prediction using the application processor unit and the real-time microcontroller unit to achieve rapid response of gate control. The hybrid energy management module 13 ensures power supply during external power outages, enabling the real-time microcontroller unit to take over control in a timely manner during external power outages and drive the dam gate to close within a preset time period, thereby improving the automation and safety of hydropower station operation.

[0069] Corresponding to the terminal embodiment above, this embodiment of the invention also provides an edge computing system. The edge computing system described below and the edge computing terminal described above can be referred to in correspondence.

[0070] An edge computing system includes: an edge computing terminal as provided in the above embodiments and a sensor array connected to the edge computing terminal.

[0071] In some embodiments, the system may further include: a self-powered device and / or an external power supply device connected to the edge computing terminal.

[0072] In some embodiments, the system may further include an AI vision PTZ camera connected to an edge computing terminal.

[0073] Corresponding to the terminal embodiment above, this embodiment of the invention also provides a gate control method. The gate control method described below and the edge computing terminal described above can be referred to each other.

[0074] Please refer to Figure 6 , Figure 6A flowchart illustrating a gate control method provided in an embodiment of the present invention. This method is applied to an edge computing terminal as provided in the above embodiment, and includes:

[0075] Step 101: Receive sensor data collected by the sensor group using pluggable I / O components; wherein the sensor group includes a water level gauge and a flow rate gauge.

[0076] Step 102: Using the application processor unit, generate a gate opening sequence corresponding to the water level prediction information within a preset future time period based on sensor data, and generate the opening control command corresponding to the gate opening sequence.

[0077] Step 103: Using the real-time microcontroller unit, the dam gate control components are controlled to adjust the opening of the dam gate according to the opening control command.

[0078] The method also includes:

[0079] Using a real-time microcontroller unit, the dam gate is controlled to close within a preset time period by the dam gate control component based on the external power failure command triggered by the hybrid energy management module. The hybrid energy management module is used to provide power from its own power supply equipment when an external power failure is detected, and to trigger the external power failure command.

[0080] In some embodiments, the method may further include:

[0081] Using the application processor unit, an emergency control instruction is generated based on an external power failure instruction; the emergency control instruction is written to shared memory and a high-priority interrupt is triggered.

[0082] Correspondingly, using a real-time microcontroller unit, based on an external power outage command triggered by the hybrid energy management module, the dam gate control components are used to control the dam gates to close within a preset time period, including:

[0083] Using a real-time microcontroller unit, emergency control instructions are read from shared memory based on high-priority interrupts; based on the emergency control instructions, the dam gate control components control the dam gates to close within a preset time period.

[0084] In some embodiments, the application processor unit generates a gate opening sequence corresponding to water level prediction information within a preset future time period based on sensor data, and generates an opening control command corresponding to the gate opening sequence, including:

[0085] Using the application processor unit, based on sensor data, a long short-term memory network model is used to generate water level prediction information for a preset future time period;

[0086] Based on water level prediction information, a gate opening sequence is generated using a gate-hydraulic coupling model.

[0087] Generate opening control commands based on the gate opening sequence.

[0088] In some embodiments, the long short-term memory network model includes a high-precision prediction model and a lightweight prediction model; based on sensor data, the long short-term memory network model is used to generate water level prediction information for a preset future time period, including:

[0089] When the preset triggering conditions are not met, a lightweight prediction model is used to generate water level prediction information for a preset future time period based on the target sensor data; where the target sensor data is data collected by some sensors in the sensor group.

[0090] When preset triggering conditions are met, water level prediction information for a preset future time period is generated based on sensor data using a high-precision prediction model; wherein the preset triggering conditions include at least one of the following: water level change rate exceeds change rate threshold, water level prediction error reaches error threshold, receiving preset external control command, and abnormal data collected by key sensors in the sensor group.

[0091] In some embodiments, the external actuator further includes a cleaning device; the method may also include:

[0092] Using the application processor unit, based on water level prediction information and sensor data, it is determined whether the target control strategy is met;

[0093] If so, a preset opening control sequence and / or preset cleaning instructions corresponding to the target control strategy are generated to control the real-time microcontroller unit to drive the dam gate control components and / or cleaning equipment; wherein, the target control strategy is any preset control strategy.

[0094] In some embodiments, the self-powered device includes a supercapacitor, a lithium battery, and a diesel generator; the method may further include:

[0095] The hybrid energy management module can switch to supercapacitors, lithium batteries and diesel generators for power supply when an external power outage is detected.

[0096] In some embodiments, the edge computing terminal further includes: a communication unit; the communication unit includes a mobile communication module and a satellite communication module; the method may further include:

[0097] By utilizing heterogeneous computing modules, alarm information can be sent through satellite communication modules when the signal quality of mobile communication modules does not meet communication requirements.

[0098] In some embodiments, the sensor group further includes at least one of a rain gauge, a weather instrument, and a sediment concentration meter; the sensor group is mounted on the same mast.

[0099] Pluggable I / O components are also used to connect to an AI vision PTZ camera; the external actuator also includes a cleaning device; the method may also include:

[0100] Using the application processor unit, cleaning control commands are generated based on the cleaning event information output by the AI ​​vision PTZ camera, in order to control the real-time microcontroller unit to drive the cleaning equipment to perform cleaning.

[0101] In this embodiment, the present invention utilizes an application processor unit to achieve high-precision local water level prediction and a real-time microcontroller unit to achieve rapid response in gate control. A hybrid energy management module ensures power supply during external power outages, enabling the real-time microcontroller unit to take over control in a timely manner and drive the dam gates to close within a preset time period, thereby improving the automation and safety of hydropower station operation.

[0102] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. Regarding the systems and methods disclosed in the embodiments, since they correspond to the terminals disclosed in the embodiments, the descriptions are relatively simple; relevant details can be found in the terminal section.

[0103] The foregoing has provided a detailed description of an edge computing terminal, system, and gate control method provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and core ideas of the present invention. It should be noted that those skilled in the art can make various improvements and modifications to the present invention without departing from its principles, and these improvements and modifications also fall within the protection scope of the present invention.

Claims

1. An edge computing terminal, characterized in that, include: Pluggable I / O components for connecting sensor arrays and external actuators; wherein the sensor arrays include level gauges and flow meters; and the external actuators include dam gate control components. A heterogeneous computing module; the heterogeneous computing module includes an application processor unit and a real-time microcontroller unit. The application processor unit is used to generate a gate opening sequence corresponding to water level prediction information within a preset future time period based on sensor data collected by the sensor group, and to generate an opening control command corresponding to the gate opening sequence. The real-time microcontroller unit is used to control the dam gate control components to adjust the opening of the dam gate according to the opening control command. The hybrid energy management module is used to connect external power supply equipment and self-powered equipment; when an external power outage is detected, power is supplied using the self-powered equipment, and an external power outage command is triggered to control the real-time microcontroller unit to control the dam gate to close within a preset time period through the dam gate control component.

2. The edge computing terminal according to claim 1, characterized in that, The heterogeneous computing module further includes: shared memory to which both the application processor unit and the real-time microcontroller unit are connected; The application processor unit is configured to generate an emergency control instruction based on the external power failure instruction; write the emergency control instruction into the shared memory and trigger a high-priority interrupt; The real-time microcontroller unit is used to read the emergency control instruction from the shared memory according to the high-priority interrupt; and according to the emergency control instruction, to control the dam gate to close within the preset time period through the dam gate control component.

3. The edge computing terminal according to claim 1, characterized in that, The application processor unit is specifically used to generate water level prediction information for the preset future time period based on the sensor data using a long short-term memory network model; generate the gate opening sequence based on the water level prediction information using a gate-hydraulic coupling model; and generate the opening control command based on the gate opening sequence.

4. The edge computing terminal according to claim 3, characterized in that, The long short-term memory network model includes a high-precision prediction model and a lightweight prediction model; the step of generating water level prediction information for the preset future time period using the long short-term memory network model based on the sensor data includes: When the preset triggering conditions are not met, the lightweight prediction model generates water level prediction information for the preset future time period based on the target sensor data; wherein, the target sensor data is data collected by some sensors in the sensor group; When a preset trigger condition is met, water level prediction information for the preset future time period is generated using the high-precision prediction model based on the sensor data; wherein, the preset trigger condition includes at least one of the following: water level change rate exceeds a change rate threshold, water level prediction error reaches an error threshold, a preset external control command is received, and data collected by key sensors in the sensor group is abnormal.

5. The edge computing terminal according to claim 1, characterized in that, The external actuator also includes cleaning equipment; The application processor unit is further configured to detect whether the target control strategy is met based on the water level prediction information and the sensor data; if so, generate a preset opening control sequence and / or a preset cleaning instruction corresponding to the target control strategy to control the real-time microcontroller unit to drive the dam gate control component and / or the cleaning equipment; wherein, the target control strategy is any preset control strategy.

6. The edge computing terminal according to claim 1, characterized in that, The self-powered equipment includes a supercapacitor, a lithium battery, and a diesel generator; The hybrid energy management module is specifically used to switch to the supercapacitor, the lithium battery, and the diesel generator for power supply when an external power outage is detected.

7. The edge computing terminal according to claim 1, characterized in that, Also includes: Communication unit; The communication unit includes a mobile communication module and a satellite communication module; The heterogeneous computing module is used to send alarm information through the satellite communication module when the signal quality of the mobile communication module does not meet the communication requirements.

8. The edge computing terminal according to any one of claims 1 to 7, characterized in that, The sensor group also includes at least one of a rain gauge, a weather instrument, and a sediment concentration meter; the sensor group is mounted on the same mast. The pluggable I / O component is also used to connect to the AI ​​vision PTZ camera; the external actuator also includes a cleaning device; the application processor unit is also used to generate a cleaning control command based on the cleaning event information output by the AI ​​vision PTZ camera, so as to control the real-time microcontroller unit to drive the cleaning device to perform cleaning.

9. An edge computing system, characterized in that, include: The edge computing terminal and the sensor array connected to the edge computing terminal as described in any one of claims 1 to 8.

10. A gate control method, characterized in that, Applied to the edge computing terminal as described in any one of claims 1 to 8, comprising: The system utilizes pluggable I / O components to receive sensor data collected by a sensor array, which includes a water level gauge and a flow rate gauge. Using the application processor unit, based on the sensor data, a gate opening sequence corresponding to the water level prediction information within a preset future time period is generated, and an opening control command corresponding to the gate opening sequence is generated. Using a real-time microcontroller unit, the dam gate control components are controlled to adjust the opening of the dam gate according to the opening control command; Using the real-time microcontroller unit, based on the external power outage command triggered by the hybrid energy management module, the dam gate control component controls the dam gate to close within a preset time period; wherein, the hybrid energy management module is used to provide power using its self-powered equipment when an external power outage is detected, and to trigger the external power outage command.

Images