Gate control method and system based on heterogeneous data calibration and confidence fusion

Abstract

The invention discloses a gate control method and system based on heterogeneous data calibration and confidence fusion, and belongs to the technical field of gate control, and the method comprises the following steps: S1, collecting a visual image, water level data and gate opening through a perception and execution layer; s2, performing consistency judgment according to the visual image, the water level data and the gate opening through edge calculation and a fusion layer, and outputting a reliable gate opening; and S3, gate opening and closing control is executed according to the reliable gate opening degree. Through a multi-source fusion mechanism of an execution mechanism encoder, vision and hydraulics, in combination with consistency judgment and abnormal switching logic, a water conservancy virtual sensor is constructed through an edge calculation unit, reliable sensing and safety control of the gate opening degree under the complex working condition are achieved, and the safety of the gate opening degree under the complex working condition is improved. The problem of precision drift of a heterogeneous auxiliary sensor in a long-term idle state and the problem of distortion of fixed weight fusion in a complex working condition are solved.

Description

Technical Field

[0001] This invention belongs to the field of gate control technology, specifically relating to a gate control method and system based on heterogeneous data calibration and confidence fusion. Background Technology

[0002] Traditional gate opening measurement mainly relies on a single sensor (such as an encoder or a wire displacement sensor), which suffers from susceptibility to environmental influences (such as water flow impact and siltation), poor long-term operational stability, and a lack of self-diagnostic capabilities. Especially under complex operating conditions (such as turbid water, floating objects, and nighttime operations), the reliability of a single sensor is significantly reduced, leading to an increased risk of gate control failure.

[0003] In existing technologies, some systems attempt to introduce vision or water level sensors as auxiliary measurement methods, but they lack multi-source data fusion and anomaly detection mechanisms, and still have the following shortcomings: (1) Accuracy drift of heterogeneous auxiliary sensors under long-term idle conditions: In traditional redundant designs, vision and hydraulic backpropagation models are typically used as backups. However, due to the lack of real-time calibration, when the primary sensor (encoder) fails and the system switches to the backup, the accuracy of the backup data often drops sharply due to environmental changes (such as gate wear or changes in flow regime), failing to meet control requirements. (2) Distortion of fixed-weight fusion under complex working conditions: Existing technologies mostly use fixed thresholds to judge the quality of sensors, lacking dynamic quantitative assessment of the current "health" of sensors, and are unable to automatically reduce the weight of interfered data in complex scenarios such as muddy water, nighttime, and turbulent flow. Summary of the Invention

[0004] In view of the above-mentioned shortcomings in the prior art, the gate control method and system based on heterogeneous data calibration and confidence fusion provided by the present invention solves the following technical problems existing in the prior art.

[0005] (1) The accuracy drift problem of heterogeneous auxiliary sensors under long-term idleness.

[0006] (2) Distortion problem of fixed weight fusion under complex working conditions.

[0007] To achieve the aforementioned objectives, the present invention employs the following technical solution: a gate control method based on heterogeneous data calibration and confidence fusion, comprising the following steps: S1. Visual images, water level data, and gate opening are collected through the perception and execution layer; S2. Through edge computing and fusion layer, consistency judgment is performed based on visual images, water level data and gate opening, and a reliable gate opening is output; S3. Perform gate opening and closing control based on reliable gate opening degree.

[0008] Furthermore: S1 specifically refers to: The gate opening information is obtained in real time by an encoder to obtain the gate opening; images of the gate and water surface are collected by a camera to generate a visual image; and water level information of upstream and downstream is collected by an ultrasonic or radar level gauge to obtain water level data.

[0009] Furthermore: S2 specifically refers to: Determine whether the encoder is in good working order based on the gate opening degree; If so, the gate opening is taken as the reliable gate opening. Based on the water level data and the gate opening, the comprehensive flow coefficient and the "pixel-actual distance" conversion ratio factor are calculated through the online parameter self-learning mechanism and stored in the dynamic parameter library of the edge computing unit. If not, the latest comprehensive flow coefficient and "pixel-to-actual distance" conversion ratio factor are read from the dynamic parameter library. The confidence weight is calculated by the edge computing unit based on the comprehensive flow coefficient and "pixel-to-actual distance" conversion ratio factor. The virtual gate opening is calculated based on the dynamic confidence fusion mechanism of working condition perception as a reliable gate opening.

[0010] Furthermore, the specific method of the online parameter self-learning mechanism is as follows: (1) Input the water level data into the preset gate hydraulic flow model, reverse calculate and update the comprehensive flow coefficient in the gate hydraulic model, and store the comprehensive flow coefficient in the dynamic parameter library of the edge computing unit; (2) Based on the gate opening, the scaling factor of the "pixel-actual distance" conversion in the visual recognition algorithm is adjusted in real time. This is to eliminate the cumulative errors caused by thermal expansion and contraction or minute displacement of the camera.

[0011] Furthermore, the specific method for determining whether the encoder's operating status is healthy is as follows: The rate of change of the gate opening is obtained based on the gate opening degree. It is then determined whether the rate of change exceeds the threshold. If it does, the encoder is considered unhealthy; otherwise, the encoder is considered healthy.

[0012] Furthermore: In S22, the comprehensive flow coefficient in the gate hydraulic model is calculated. The specific expression is: In the formula, For traffic, For the gate opening, The upstream water level, The upstream water level, It is the acceleration due to gravity. The width of the gate; In the formula, This refers to the overall flow coefficient in the gate hydraulic model before the update.

[0013] Furthermore: The virtual gate opening is calculated based on the dynamic confidence fusion mechanism of operational condition perception. The specific expression is: In the formula, To visually identify the gate opening degree, To adjust the gate opening using hydraulic reverse thrust, The first confidence level weight is used. This is the weight for the second confidence level.

[0014] A gate control system based on heterogeneous data calibration and confidence fusion includes: The perception and execution layer is used to acquire basic physical information about the gate's operation and execute control commands. The perception and execution layer includes: The edge computing and fusion layer calculates reliable gate opening through edge computing units; The control and decision-making layer, including the PL controller and safety interlock module, is used to receive reliable gate opening degree output from the edge computing unit and execute gate opening and closing control and safety interlock logic; The monitoring and scheduling layer, including a host computer or remote monitoring system, is used to display the gate's operating status, sensor health status, and alarm information, and supports remote scheduling and operation and maintenance management.

[0015] Furthermore: the perception and execution layer includes: The actuator and main sensing unit include a gate and an encoder. The gate is driven by a servo motor, a reducer and a ball screw, and the encoder acquires the gate's opening information in real time. The visual perception unit includes a camera and supplementary lighting device arranged above or to the side of the gate, used to collect images of the gate and the water surface, and to visually identify the opening information of the gate. The water level sensing unit includes ultrasonic or radar water level gauges deployed in the stable water areas upstream and downstream of the gate, used to collect upstream and downstream water levels in real time.

[0016] The beneficial effects of this invention are as follows: This invention provides a gate control method and system based on heterogeneous data calibration and confidence fusion. Through a multi-source fusion mechanism of "actuator encoder, vision and hydraulics", combined with consistency discrimination and anomaly switching logic, and constructed through edge computing units, a virtual sensor is built to achieve reliable perception and safe control of gate opening under complex working conditions. Compared with the prior art, it has the following advantages: (1) Upgrade from "static standby" to "dynamic calibration": Unlike the long-term "blind waiting" of standby sensors in the prior art, this invention designs an online parameter self-learning mechanism, which allows the standby sensor to be "run-in" and calibrated by the main sensor during daily operation. This enables the virtual gate opening to be calculated by visual perception and hydraulic back-inference information during the period when the main sensor fails, thus eliminating the system error caused by long-term operation and solving the accuracy drift problem of heterogeneous auxiliary sensors under long-term idleness.

[0017] (2) Dynamic weight fusion: It can automatically adjust the confidence weight according to the latest comprehensive flow coefficient and "pixel-actual distance" conversion ratio factor read from the dynamic parameter library, realize high-precision and disturbance-free gate opening control takeover, avoid the data fusion failure caused by a single fixed logic under harsh working conditions, significantly improve the system's all-weather working capability, and solve the distortion problem of fixed weight fusion under complex working conditions.

[0018] (3) Physical consistency constraints were constructed: By real-time collision verification of hydraulic formulas (fluid) and kinematic formulas (mechanical), control accidents caused by sensor logic errors were eliminated from the perspective of physical principles. Attached Figure Description

[0019] Figure 1 This is a flowchart of the gate control method based on heterogeneous data calibration and confidence fusion of the present invention.

[0020] Figure 2 This is a schematic diagram of the gate control system based on heterogeneous data calibration and confidence fusion according to the present invention. Detailed Implementation

[0021] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0022] like Figure 1 As shown, in one embodiment of the present invention, the gate control method based on heterogeneous data calibration and confidence fusion includes the following steps: S1. Visual images, water level data, and gate opening are collected through the perception and execution layer; S2. Through edge computing and fusion layer, consistency judgment is performed based on visual images, water level data and gate opening, and a reliable gate opening is output; S3. Perform gate opening and closing control based on reliable gate opening degree.

[0023] S1 specifically refers to: The gate opening information is obtained in real time by an encoder to obtain the gate opening; images of the gate and water surface are collected by a camera to generate a visual image; and water level information of upstream and downstream is collected by an ultrasonic or radar level gauge to obtain water level data.

[0024] In this embodiment, the encoder is used as the main sensor. When the system determines that the main sensor is in a healthy state, the gate opening degree output by the encoder is used as a reliable reference value to start the online parameter self-learning mechanism, so as to realize the real-time calibration of the auxiliary sensor and physical model parameters.

[0025] S2 specifically refers to: Determine whether the encoder is in good working order based on the gate opening degree; If so, the gate opening is taken as the reliable gate opening. Based on the water level data and the gate opening, the comprehensive flow coefficient and the "pixel-actual distance" conversion ratio factor are calculated through the online parameter self-learning mechanism and stored in the dynamic parameter library of the edge computing unit. If not, the latest comprehensive flow coefficient and "pixel-to-actual distance" conversion ratio factor are read from the dynamic parameter library. The confidence weight is calculated by the edge computing unit based on the comprehensive flow coefficient and "pixel-to-actual distance" conversion ratio factor. The virtual gate opening is calculated based on the dynamic confidence fusion mechanism of working condition perception as a reliable gate opening.

[0026] In this embodiment, when an anomaly is detected in the main sensor data or a physical consistency conflict exists with other sensor data, the self-learning process based on the online parameter self-learning mechanism is immediately frozen to prevent abnormal data from polluting the dynamic parameter library. The comprehensive flow coefficient and "pixel-actual distance" conversion ratio factor of the most recently valid calibration in the parameter library are called, and the virtual gate opening output mode based on auxiliary sensor fusion is switched to achieve smooth takeover of gate control.

[0027] The specific method of the online parameter self-learning mechanism is as follows: (1) Input the water level data into the preset gate hydraulic flow model, reverse calculate and update the comprehensive flow coefficient in the gate hydraulic model, and store the comprehensive flow coefficient in the dynamic parameter library of the edge computing unit; (2) Based on the gate opening, the scaling factor of the "pixel-actual distance" conversion in the visual recognition algorithm is adjusted in real time. This is to eliminate the cumulative errors caused by thermal expansion and contraction or minute displacement of the camera.

[0028] The above methods ensure that the visual perception unit and the gate hydraulic flow model are always in a calibrated state during normal system operation.

[0029] The specific method for determining whether the encoder is in a healthy working state is as follows: The rate of change of the gate opening is obtained based on the gate opening degree. It is then determined whether the rate of change exceeds the threshold. If it does, the encoder is considered unhealthy; otherwise, the encoder is considered healthy.

[0030] In S22, the comprehensive flow coefficient in the gate hydraulic model is calculated. The specific expression is: In the formula, For traffic, For the gate opening, The upstream water level, The upstream water level, It is the acceleration due to gravity. The width of the gate; In this embodiment, the invention innovatively proposes using the "true value" of the main sensor to train the parameters of the gate hydraulic model in real time during operation. When the encoder is working normally, the system collects the opening degree measured by the encoder, as well as real-time water level and flow data, and calculates the current comprehensive flow coefficient in reverse. The system stores the calculated comprehensive flow coefficient in the dynamic parameter library of the edge computing unit. When the encoder malfunctions, the system immediately calls the most recently calibrated... The opening value is used to inversely calculate the value. This ensures that the backup system remains accurate even after sedimentation or wear alters the gate's flow characteristics.

[0031] In the formula, This refers to the overall flow coefficient in the gate hydraulic model before the update.

[0032] The virtual gate opening is calculated based on the dynamic confidence fusion mechanism of working condition perception. The specific expression is: In the formula, To visually identify the gate opening degree, The first confidence level weight is used. As the second confidence level weight, , The gate opening is determined by hydraulic back-calculation, which is obtained by back-calculation using the comprehensive flow coefficient formula in the gate hydraulic model.

[0033] In this embodiment, the first confidence weight is based on the scaling factor of "pixel-actual distance" conversion. Calculation, based on Histogram analysis is performed on the image to extract contrast and signal-to-noise ratio. The geometric consistency and fluctuation characteristics of the visual recognition gate opening under the current operating conditions are evaluated to obtain the visual recognition residual. The edge computing unit adjusts the first confidence weight based on the visual recognition residual. For example: If low illumination (nighttime) or high turbidity (blurred feature points) is detected, the system automatically reduces the weight of the first confidence level. (For example, reducing it from 0.3 to 0.05) to prevent visual noise interference control.

[0034] The second confidence weight is calculated based on the comprehensive flow coefficient. The model residual between the gate hydraulic flow model back-calculation results and real-time water level and flow data is calculated using the comprehensive flow coefficient. The edge computing unit adjusts the second confidence weight based on the model residual. For example: By monitoring the time-domain variance (fluctuation frequency) of upstream and downstream water levels, if turbulent water flow is detected causing high-frequency oscillations in the water level reading, the weight of the second confidence level is automatically reduced. To prevent control commands from oscillating.

[0035] like Figure 2 As shown, the gate control system based on heterogeneous data calibration and confidence fusion includes: The perception and execution layer is used to acquire basic physical information about the gate's operation and execute control commands. The perception and execution layer includes: The actuator and main sensing unit include a gate and an encoder. The gate is driven by a servo motor, a reducer and a ball screw. The encoder acquires the opening information of the gate in real time. In this embodiment, the encoder is used as the main sensor to provide a high-precision opening reference value under normal system operation. The visual perception unit includes a camera and supplementary lighting device arranged above or to the side of the gate, used to collect images of the gate and the water surface, and to visually identify the opening information of the gate. The water level sensing unit includes ultrasonic or radar water level gauges deployed in the stable water areas upstream and downstream of the gate for real-time collection of upstream and downstream water levels. The edge computing and fusion layer calculates reliable gate opening through edge computing units; The control and decision-making layer, including the PL controller and safety interlock module, is used to receive reliable gate opening degree output from the edge computing unit and execute gate opening and closing control and safety interlock logic; The monitoring and scheduling layer, including a host computer or remote monitoring system, is used to display the gate's operating status, sensor health status, and alarm information, and supports remote scheduling and operation and maintenance management.

[0036] To verify the effectiveness of the method of the present invention, the following experimental cases are provided in this embodiment: I. Normal Operation and Self-Learning Phase: The system operates under clean water and daytime conditions. Encoder readings are accurate. The edge computing unit continuously calculates the overall flow coefficient in the background.

[0037] Data performance: After 100 hours of operation and learning, the system detected slight siltation at the bottom of the gate, resulting in actual drift. The system automatically updated the parameter library, completing a "soft calibration" without manual intervention.

[0038] II. High-precision takeover after encoder failure: If traditional techniques (using fixed coefficients) are used, the reverse calculation of the opening error may reach 15mm; This system uses coefficients updated through self-learning just before the fault to back-calculate, and the actual opening error is only 2mm.

[0039] Result: Seamless take-off and maintenance in compliance with engineering standards was achieved without downtime.

[0040] III. Interference resistance under high turbidity conditions: Action: The image processing module detects that the signal-to-noise ratio is lower than the threshold and automatically adjusts the first confidence weight from... Down to At the same time, increase the hydraulic thrust weight. Result: The output fusion opening curve is smooth and does not produce erroneous actions due to visual noise.

[0041] In the description of this invention, the above are merely preferred embodiments and are not intended to limit the scope of protection of this invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A gate control method based on heterogeneous data calibration and confidence fusion, characterized in that, Includes the following steps: S1. Visual images, water level data, and gate opening are collected through the perception and execution layer; S2. Through edge computing and fusion layer, consistency judgment is performed based on visual images, water level data and gate opening, and a reliable gate opening is output; S3. Perform gate opening and closing control based on reliable gate opening degree.

2. The gate control method based on heterogeneous data calibration and confidence fusion according to claim 1, characterized in that, S1 specifically refers to: The gate opening information is obtained in real time by an encoder to obtain the gate opening; images of the gate and water surface are collected by a camera to generate a visual image; and water level information of upstream and downstream is collected by an ultrasonic or radar level gauge to obtain water level data.

3. The gate control method based on heterogeneous data calibration and confidence fusion according to claim 2, characterized in that, S2 specifically refers to: Determine whether the encoder is in good working order based on the gate opening degree; If so, the gate opening is taken as the reliable gate opening. Based on the water level data and the gate opening, the comprehensive flow coefficient and the "pixel-actual distance" conversion ratio factor are calculated through the online parameter self-learning mechanism and stored in the dynamic parameter library of the edge computing unit. If not, the latest comprehensive flow coefficient and "pixel-to-actual distance" conversion ratio factor are read from the dynamic parameter library. The confidence weight is calculated by the edge computing unit based on the comprehensive flow coefficient and "pixel-to-actual distance" conversion ratio factor. The virtual gate opening is calculated based on the dynamic confidence fusion mechanism of working condition perception as a reliable gate opening.

4. The gate control method based on heterogeneous data calibration and confidence fusion according to claim 3, characterized in that, The specific method of the online parameter self-learning mechanism is as follows: (1) Input the water level data into the preset gate hydraulic flow model, reverse calculate and update the comprehensive flow coefficient in the gate hydraulic model, and store the comprehensive flow coefficient in the dynamic parameter library of the edge computing unit; (2) Based on the gate opening, the scaling factor of the "pixel-actual distance" conversion in the visual recognition algorithm is adjusted in real time. This is to eliminate the cumulative errors caused by thermal expansion and contraction or minute displacement of the camera.

5. The gate control method based on heterogeneous data calibration and confidence fusion according to claim 3, characterized in that, The specific method for determining whether the encoder is in a healthy working state is as follows: The rate of change of the gate opening is obtained based on the gate opening degree. It is then determined whether the rate of change exceeds the threshold. If it does, the encoder is considered unhealthy; otherwise, the encoder is considered healthy.

6. The gate control method based on heterogeneous data calibration and confidence fusion according to claim 4, characterized in that, In S22, the comprehensive flow coefficient in the gate hydraulic model is calculated. The specific expression is: In the formula, For traffic, For the gate opening, The upstream water level, The upstream water level, It is the acceleration due to gravity. The width of the gate; In the formula, This refers to the overall flow coefficient in the gate hydraulic model before the update.

7. The gate control method based on heterogeneous data calibration and confidence fusion according to claim 6, characterized in that, The virtual gate opening is calculated based on the dynamic confidence fusion mechanism of working condition perception. The specific expression is: In the formula, To visually identify the gate opening degree, To adjust the gate opening using hydraulic reverse thrust, The first confidence level weight is used. This is the weight for the second confidence level.

8. A gate control system based on heterogeneous data calibration and confidence fusion, applied to the gate control method based on heterogeneous data calibration and confidence fusion as described in any one of claims 1 to 7, characterized in that, The system includes: The perception and execution layer is used to acquire basic physical information about the gate's operation and execute control commands. The perception and execution layer includes: The edge computing and fusion layer calculates reliable gate opening through edge computing units; The control and decision-making layer, including the PL controller and safety interlock module, is used to receive reliable gate opening degree output from the edge computing unit and execute gate opening and closing control and safety interlock logic; The monitoring and scheduling layer, including a host computer or remote monitoring system, is used to display the gate's operating status, sensor health status, and alarm information, and supports remote scheduling and operation and maintenance management.

9. The gate control system based on heterogeneous data calibration and confidence fusion according to claim 8, characterized in that, The perception and execution layer includes: The actuator and main sensing unit include a gate and an encoder. The gate is driven by a servo motor, a reducer and a ball screw, and the encoder acquires the gate's opening information in real time. The visual perception unit includes a camera and supplementary lighting device arranged above or to the side of the gate, used to collect images of the gate and the water surface, and to visually identify the opening information of the gate. The water level sensing unit includes ultrasonic or radar water level gauges deployed in the stable water areas upstream and downstream of the gate, used to collect upstream and downstream water levels in real time.

Images