Multi-machine cooperative system for underwater dam detection

Abstract

The invention discloses a multi-machine cooperative system for underwater dam detection, and relates to the field of intelligent detection.The multi-machine cooperative system is provided with a shore-based control center, a water surface power buoy, an underwater detection robot and an auxiliary sub-machine, the water surface power buoy achieves navigation positioning through a GPS module, transmits acoustic signals to the underwater robot in combination with a USBL, and assists in determining the relative position; the underwater detection robot integrates a detection module, a power module, a navigation module and an auxiliary sub-machine, and detects the dam through cooperative work of the four modules; the shore-based control center is used as an information hub to realize information transmission among the buoy, the underwater robot and the auxiliary sub-machine through multiple links; the system ensures detection data integrity and operation stability through a multi-device cooperative sensing and high-precision data fusion technology, improves detection efficiency and safety evaluation timeliness by combining a multi-link transmission scheme of underwater acoustic communication and GPS positioning, and realizes real-time identification, positioning and quantification of dam defects.

Description

Technical Field

[0001] This invention relates to the field of intelligent detection, specifically a multi-machine collaborative system for underwater dam detection. Background Technology

[0002] In recent years, my country's hydropower construction has developed rapidly, with large-scale hydropower stations such as Xiangjiaba, Xiluodu, Longtan, Nuozhadu, Xiaowan, and Lianghekou being put into operation. The Yangtze River main stream boasts five hydropower stations, including the Three Gorges and Baihetan, ranking among the world's top ten in installed capacity, providing strong support for national strategies such as the "West-to-East Power Transmission" project and the development of the Yangtze River Economic Belt. However, hydraulic structures, subjected to long-term environmental erosion, material aging, and the continuous effects of various loads, commonly suffer from cavitation, erosion, and dissolution, leading to accumulated structural damage and weakened resistance, seriously threatening dam safety. Statistics show that up to 72.9% of dam overall damage and durability reduction problems are caused by cracks, and various defects often originate or manifest on the structural surface. Therefore, regular and comprehensive dam inspections are crucial for timely detection and identification of defects and hidden dangers, scientific evaluation of dam safety status, and providing key evidence for reinforcement and strengthening.

[0003] Traditional underwater inspections primarily rely on divers. However, the underwater environment of hydropower dams typically features significant challenges, including deep working depths, wide defect distribution, long maintenance times, complex turbulent currents, and extremely low visibility. These characteristics present manual inspection methods with severe challenges such as high risk, low efficiency, limited coverage, and insufficient technical capabilities, making it difficult to meet the high standards of safety and maintenance requirements of modern hydropower stations. Existing underwater inspection robots struggle to maintain stability and precise control in confined waterways with turbulent currents and have numerous blind spots. Furthermore, their deployment and retrieval require substantial financial and material resources. Therefore, there is an urgent need for a multi-robot collaborative underwater dam defect precision inspection system that can improve the accuracy of full-coverage inspection of the dam surface while simultaneously enhancing the efficiency and reliability of underwater inspection operations. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-machine collaborative system for underwater dam inspection, in order to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a multi-machine collaborative system for underwater dam inspection, the system comprising a shore-based control center, a surface dynamic buoy, and an underwater inspection robot; The shore-based control center serves as a primary decision-making hub. It uses a crane to lift the buoy and underwater robot to a safe area and release them into the water. It receives positioning, status, and sensor data from both, acquires high-definition video via a fiber optic cable, processes it in real time, and dynamically displays the detection progress. It monitors the equipment status, identifying communication interruptions, power anomalies, and attitude instability as equipment malfunctions. In such cases, it first attempts to repair the equipment remotely. If communication and power are not restored, or attitude is not adjusted, the remote repair is deemed ineffective, triggering an emergency jettison command. After the mission is completed, it coordinates the recovery, deeply processes the data, and archives reports. The surface dynamic buoy is a two-level collaborative layer. After power-on, it initializes dual GPS and merges positioning, navigates to the target position via PID, and releases the robot after receiving instructions from the shore-based control center. It also starts the USBL positioning robot and transmits coordinates back, while serving as a relay to extend communication. During recovery, it guides the robot to float up and locks it. The underwater robot is a three-level execution core. After release, it activates the navigation and calculation system (SINS), positioning sonar (USBL), and Doppler log (DVL). It achieves full-degree-of-freedom motion through positioning via SINS+USBL+DVL and cascaded PID control of the thrusters. It uses an acoustic array, high-definition camera, and 3D scanning module for detection. In narrow areas, it releases auxiliary sub-machines for coordination. After receiving instructions, it floats up to dock with a buoy for recovery. In case of emergency, it jettisons ballast and floats up.

[0006] The shore-based control center includes a mission and data management unit and a deployment, recovery, and emergency response unit. The task and data management unit formulates a global task plan and scheduling scheme based on the dam inspection requirements. It then establishes bidirectional data transmission with the surface dynamic buoy, underwater inspection robot, and auxiliary submachines through optical fiber, wireless communication, and an optoelectronic composite cable towed by an underwater robot. It receives attitude, depth, thruster status, and acoustic and optical sensor data transmitted back by the underwater robot in real time, and monitors the entire inspection operation process in real time. During this period, it also needs to issue instructions for gimbal pitch / roll control, camera parameter adjustment, and movement and inspection of auxiliary submachines, and monitor the status of each device simultaneously. The deployment, retrieval, and emergency response unit first deploys a crane and operators in an area near the pre-set deployment point. The crane lifts and transports the integrated surface-powered buoy and underwater detection robot smoothly to the pre-calculated safe deployment area. After the buoy triggers its self-detachment device and the robot completes deployment, the crane retrieves the connecting rope. During the retrieval phase, after the operation is completed, the surface-powered buoy and other equipment are coordinated with the underwater detection robot to complete the retrieval of itself and its auxiliary submachine. In emergency response, if the underwater equipment is detected to have communication interruption, power abnormality, or attitude instability that cannot be recovered via remote command, the operator or intelligent control system generates a decision and sends an encrypted emergency jettison command to the underwater robot through the remaining communication link. After the robot's main control cabin verifies the command, it releases the mechanical lock of the counterweight. After the counterweight separates, the robot rises using positive net buoyancy, and the mooring reel releases the communication cable in a synchronized and orderly manner.

[0007] The surface dynamic buoy includes a positioning and navigation unit and a communication relay unit; The positioning and navigation unit is based on two physically isolated high-precision GPS modules, working in conjunction with the navigation control algorithm and integrity monitoring program built into the buoy control board. After the system is powered on, the control board first initializes and configures the two GPS modules, setting the output data format and baud rate of the GGA and RMC statements of the NMEA-0183 protocol, and enabling the data reception modes of RTK and PPP. Subsequently, the two GPS modules work in parallel, independently receiving satellite navigation signals and differential correction signals from the shore-based control center reference station, and asynchronously outputting raw data packets containing latitude and longitude, UTC time, and positioning status at high frequencies of 5Hz and 10Hz. The control board synchronously receives data through two independent UART serial ports and adds internal timestamps to achieve synchronization alignment. The navigation processor performs multi-source fusion processing on the data, using RTK... The system fixes the position data and performs consistency checks. When the module data is out of tolerance or invalid, it automatically isolates and alarms. Finally, it compares the fused actual position P1 with the preset target position P0, subtracts the two to calculate the position deviation ΔP, and the navigation control algorithm calculates the control command based on ΔP to drive the buoy propeller so that the buoy dynamically maintains itself near the target position. The communication relay unit establishes data interaction links with other surface buoys in the operating area to share the location and environmental information collected by each buoy. On the other hand, it serves as a communication relay station between the shore-based control center and the underwater detection robot, receiving control commands issued by the shore-based control center and forwarding them to the underwater detection robot. At the same time, it collects the status data, detection data, and buoy operation data transmitted back by the underwater robot, summarizes them, and transmits them back to the shore-based control center.

[0008] The surface dynamic buoy includes a deployment and recovery auxiliary unit and an active auxiliary positioning unit; During the deployment phase, the deployment and recovery auxiliary unit first securely connects to the underwater detection robot via a control rod. After receiving the target deployment location command from the shore-based control center, it navigates to the target point using a positioning and navigation unit, triggering a self-detachment device to release the control rod and complete the precise deployment of the underwater detection robot. Simultaneously, in the initial deployment phase, the self-detachment device connects to the crane at the shore-based control center. Once the buoy is hoisted to the surface deployment area and its position is confirmed to be safe, the self-detachment device disconnects from the crane, and the crane retrieves the rope. During the recovery phase, after receiving a recovery command from the shore-based control center, the unit coordinates for the underwater detection robot to return to the buoy and reconnects to the robot via the control rod. If assistance from the shore-based control center is required, the robot is recovered using the crane; otherwise, it is recovered autonomously using the buoy's power to tow the robot to the designated recovery area. The active assisted positioning unit first obtains the buoy's high-precision absolute geographic coordinates from the positioning and navigation unit; then, the positioning sonar system transmits acoustic signals of a specific frequency to the underwater detection robot; after receiving the response signal returned by the underwater detection robot, the unit calculates the straight-line distance between the underwater detection robot and the buoy by measuring the round-trip time between the interrogation signal and the response signal and multiplying it by the known underwater speed of sound; simultaneously, by analyzing the phase difference between multiple hydrophone array elements of the received signal and using a direction-of-arrival estimation algorithm, the unit calculates the azimuth angle of the signal transmitted by the underwater detection robot reaching the buoy. After obtaining the distance and azimuth angle, and combining the depth information provided by the robot's depth sensor, the three parameters of distance, azimuth angle, and depth are transformed into coordinates, specifically: Using the buoy as the origin, the robot's two-dimensional relative coordinates on the horizontal plane are calculated using azimuth and distance. Combined with the depth value, a three-dimensional rectangular coordinate system is constructed for the underwater inspection robot relative to the buoy, thus obtaining its complete real-time relative position coordinates. The calculated three-dimensional spatial coordinates of the robot relative to the buoy are then superimposed on the absolute geographic coordinates of the buoy itself, which are obtained through BeiDou high-precision positioning. Through spatial geometric transformation, this local vector of relative coordinates is transformed and superimposed onto the global geographic coordinate system with the absolute position of the buoy as the origin, thus calculating the absolute geographic coordinates of the underwater inspection robot. The absolute geographic coordinates are the coordinates of a specific geographic location on the Earth's surface obtained through high-precision positioning technology based on the BeiDou satellite navigation system. This coordinate data is then synchronously transmitted to the shore-based control center and the underwater inspection robot.

[0009] The underwater robot includes a detection module, a power module, a navigation module, and an auxiliary sub-machine. The detection module includes an image acquisition unit and an acoustic detection unit; The image acquisition unit is based on high-resolution wide-angle optical components and large target surface sensing elements, combined with an adaptive optical compensation mechanism and a high-power controllable light source. It actively compensates for underwater light attenuation to maintain the effective observation distance in turbid waters and acquire basic images of the dam. Then, it eliminates fog perception through real-time defogging algorithms and corrects color cast through color correction technology. Finally, it outputs clear images of the dam with low distortion and high contrast. Low distortion refers to suppressing the interference of the optical system and underwater environment on the geometric shape of the image, and high contrast refers to enhancing the difference in brightness between the normal area and the defective area of ​​the dam in the image, so that the boundary between the two areas is clear. The acoustic detection unit generates electrical pulse signals of a specific waveform from the signal generation module within the control board. These signals are then amplified by a power amplifier to drive a specific array element of the acoustic array to emit acoustic pulses. After the sound waves are reflected by the target, the echo signals are received by multiple array elements of the acoustic array, amplified by a preamplifier, converted into digital signals by a high-speed analog-to-digital converter, and transmitted to the signal processing unit. Subsequently, all time-delayed and weighted signals are coherently superimposed, and the beam pointing angle is changed by electronic scanning to form a focused scan. The processed signal amplitude and transit time information are then mapped onto a two-dimensional image plane through polar coordinate-Cartesian coordinate transformation to generate an acoustic image with a resolution of 0.5-2cm.

[0010] The detection module includes a multi-degree-of-freedom gimbal unit and a three-dimensional scanning unit; The gimbal controller in the multi-degree-of-freedom gimbal unit acquires the three-axis angular velocity, three-axis acceleration, and three-axis attitude angle data of the robot's main inertial navigation system, as well as the angular velocity data of the gimbal's built-in IMU, in real time via a high-speed bus, and assigns a unified high-precision timestamp to all data. The carrier IMU data is used as the primary observation, and the difference in angular velocity between the gimbal IMU and the carrier IMU is used as the auxiliary observation. The gimbal controller uses the carrier IMU data with the unified timestamp as the main driver for Kalman filter state prediction, and calculates the theoretical expected attitude of the gimbal at high frequency. Simultaneously, the difference in angular velocity between the gimbal IMU and the carrier IMU is used as the key observation input filter. The predicted and observed values ​​are optimally fused using Kalman gain. Then, the feedforward compensation torque and the feedback torque calculated by the attitude deviation using a PID controller are combined to form the total control torque, which is then used to drive the waterproof servo motor to adjust the gimbal attitude through a field-oriented control algorithm. The 3D scanning unit emits a coded laser grid pattern of a specific wavelength onto the dam surface via a laser projector; a rigorously calibrated high-resolution binocular camera synchronously acquires the modulated and deformed laser pattern on the dam surface, obtaining two digital images, left and right; based on the predetermined scanning accuracy and point cloud acquisition rate, a semi-global matching algorithm is used to perform dense stereo matching calculations on the image pairs to obtain a disparity map, which is then converted into an initial 3D point cloud based on the left camera coordinate system using triangulation principles; combined with synchronous positioning and map building algorithms, an iterative nearest-point algorithm is used to register the current frame point cloud with the global model, optimizing the relative pose transformation matrix of the scanning unit, and then registering the point cloud... The point cloud is incrementally fused into the global model. Finally, based on the high-precision global model, a region growing algorithm is used to automatically segment the defect region. The defect depth is quantified by calculating the Euclidean distance between the defect region point cloud and the reference model. Based on the segmented defect region point cloud, a convex hull construction algorithm is used to calculate the smallest convex polyhedron that can completely enclose all defect points. Then, by calculating the closed space enclosed between the convex hull model and the reference model representing the original intact state, the space is decomposed into multiple tetrahedrons and the volume is accumulated to calculate the missing and redundant volumes of the defect. Combining the obtained defect depth and area, the defect volume, maximum depth, surface area, and location are output.

[0011] The power module includes a vertical propulsion unit and a horizontal propulsion unit; The outer ring position controller of the vertical propulsion unit is based on the depth deviation. Calculate the target's vertical velocity Z t Z represents the target depth, and Z represents the actual depth. For depth deviation, the inner loop speed controller adjusts the speed according to the deviation. Calculate the total vertical thrust required V t Let F be the target vertical velocity and V be the actual vertical velocity. Since the two thrusters have identical performance and are symmetrically arranged, a thrust distribution algorithm is used to distribute F... t The signal is evenly distributed to the left and right vertical thrusters to generate the corresponding PWM control signal; The horizontal propulsion unit includes several propulsion components symmetrically distributed around the center of the robot. Each component has rated power and a predetermined maximum thrust. It adopts a ±180° deflection vector nozzle design, and the rotation speed and nozzle direction of different propellers can be adjusted.

[0012] The navigation module includes a data acquisition and processing unit and a cooperative positioning unit; The data acquisition and processing unit integrates a gyroscope, a quartz accelerometer, a depth pressure sensor, a Doppler log, and a magnetometer to acquire three-axis angular velocity, three-axis specific force, depth pressure value, ground velocity, and geomagnetic field vector in real time. It uses a strapdown inertial navigation algorithm to generate position, velocity, attitude, and depth information, specifically: Using the ω output from the gyroscope, the attitude matrix is ​​updated in real time using the quaternion method or the direction cosine method to calculate the roll angle φ, pitch angle θ, and yaw angle ψ. Then, the specific force f measured by the accelerometer is transformed from the vehicle coordinate system to the navigation coordinate system through the attitude matrix. Gravitational acceleration and Coriolis acceleration terms are subtracted, and the three-dimensional velocity V of the vehicle in the n-frame is obtained by a first integration. The displacement relative to the initial position is obtained by a second integration. Depth information is directly obtained by performing temperature compensation and static conversion on the depth pressure value. At the same time, a Kalman filter is used to fuse the position, velocity, and attitude calculated by SINS with the ground velocity and depth values ​​of the DVL sensor, suppressing SINS integral divergence and outputting optimized relative motion data.

[0013] When the cooperative positioning unit receives the acoustic signal emitted by the surface dynamic buoy to the underwater detection robot, it sends a response signal. The cooperative positioning unit obtains the real-time position data of the underwater detection robot relative to the surface dynamic buoy and the absolute geographical location of the surface dynamic buoy. After correcting the error by extended Kalman filtering, it outputs its own absolute geographical coordinates.

[0014] The auxiliary submachine includes a motion control unit, a detection and operation unit, and an adsorption control unit; After receiving instructions from the shore-based control center forwarded by the underwater inspection robot, the motion control unit first calculates the target motion direction and observation pose required for the robot to achieve path tracking based on the geometric features and operational requirements of the preset inspection path through the trajectory tracking controller. The preset inspection path is generated by the shore-based control center and issued in the form of a series of waypoints and parametric curves. The path geometric features include curvature, length, and priority. After receiving the instructions, the motion control unit parses the path data. The navigation module of the auxiliary submachine is the same as that of the underwater inspection robot, and the pose of the auxiliary submachine is obtained through the same method. The multi-degree-of-freedom motion mode of the auxiliary submachine is consistent with that of the underwater inspection robot. The detection unit determines the activation plan of the detection equipment according to the detection scenario. Then, after the motion control unit completes the pose adjustment, it starts the deep-water camera to collect image information of the dam surface and the narrow gap area, and various sensors simultaneously collect defect-related data. The adsorption control unit consists of a negative pressure pump, a suction cup, and a high-precision pressure sensor. When the main control system issues an adsorption command, the negative pressure pump immediately starts and runs at high speed, continuously discharging water from the adsorption chamber between the suction cup and the dam surface through a sealed pipeline, creating a negative pressure environment within the adsorption chamber. Simultaneously, the high-precision pressure sensor monitors the relative pressure value within the adsorption chamber in real time and feeds the pressure data back to the main control system via a communication link. The main control system employs a closed-loop PID control algorithm, comparing the real-time monitored pressure with a preset target negative pressure threshold. Based on the pressure deviation and the rate of change of the deviation, it dynamically adjusts the duty cycle of the PWM control signal output to the negative pressure pump motor, thereby precisely controlling the pumping speed and power of the negative pressure pump.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention overcomes the limitations of a single device by using a master-slave collaborative mode between an underwater inspection robot and an auxiliary sub-machine, and achieves blind-spot-free coverage and high-precision detection of the underwater area of ​​the dam based on multi-source data.

[0016] 2. This invention reduces reliance on manpower through automated and collaborative operation methods, and enables communication between modules through multiple links, ensuring real-time monitoring of the operation process, guaranteeing the full controllability of the operation, and reducing the risks of underwater operations.

[0017] 3. This invention resists turbulent interference through the coordinated layout of the thrusters and the stable control of the gimbal, adapts to turbid water bodies through sonar, and meets the needs of various scenarios, thus reducing the cost of its later upgrades and maintenance. Attached Figure Description

[0018] Figure 1 This is a system architecture diagram of a multi-machine collaborative system for underwater dam inspection according to the present invention; Figure 2 This is a schematic diagram of the operation scenario of a multi-machine collaborative system for underwater dam inspection according to the present invention.

[0019] Figure 3 This is a flowchart of the operation of a multi-machine collaborative system for underwater dam inspection according to the present invention. Detailed Implementation

[0020] 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, and 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.

[0021] Example 1: As Figures 1-3 As shown, the present invention provides a technical solution: a multi-machine collaborative system for underwater dam inspection. The system includes a shore-based control center, surface dynamic buoys, and underwater inspection robots; The shore-based control center serves as a primary decision-making hub. It uses a crane to lift the buoy and underwater robot to a safe area and release them into the water. It receives positioning, status, and sensor data from both, acquires high-definition video via a fiber optic cable, processes it in real time, and dynamically displays the detection progress. It monitors the equipment status, identifying communication interruptions, power anomalies, and attitude instability as equipment malfunctions. In such cases, it first attempts to repair the equipment remotely. If communication and power are not restored, or attitude is not adjusted, the remote repair is deemed ineffective, triggering an emergency jettison command. After the mission is completed, it coordinates the recovery, deeply processes the data, and archives reports. The surface dynamic buoy is a two-level collaborative layer. After power-on, it initializes dual GPS and merges positioning, navigates to the target position via PID, and releases the robot after receiving instructions from the shore-based control center. It also starts the USBL positioning robot and transmits coordinates back, while serving as a relay to extend communication. During recovery, it guides the robot to float up and locks it. The underwater robot is a three-level execution core. After release, it activates the navigation and calculation system (SINS), positioning sonar (USBL), and Doppler log (DVL). It achieves full-degree-of-freedom motion through positioning via SINS+USBL+DVL and cascaded PID control of the thrusters. It uses an acoustic array, high-definition camera, and 3D scanning module for detection. In narrow areas, it releases auxiliary sub-machines for coordination. After receiving instructions, it floats up to dock with a buoy for recovery. In case of emergency, it jettisons ballast and floats up.

[0022] The shore-based control center includes a mission and data management unit and a deployment, recovery, and emergency response unit. The task and data management unit formulates a global task plan and scheduling scheme based on the dam inspection requirements. It then establishes bidirectional data transmission with the surface dynamic buoy, underwater inspection robot, and auxiliary submachines through optical fiber, wireless communication, and an optoelectronic composite cable towed by an underwater robot. It receives attitude, depth, thruster status, and acoustic and optical sensor data transmitted back by the underwater robot in real time, and monitors the entire inspection operation process in real time. During this period, it also needs to issue instructions for gimbal pitch / roll control, camera parameter adjustment, and movement and inspection of auxiliary submachines, and monitor the status of each device simultaneously. The deployment, retrieval, and emergency response unit first deploys a crane and operators in an area near the pre-set deployment point. The crane lifts and transports the integrated surface-powered buoy and underwater detection robot smoothly to the pre-calculated safe deployment area. After the buoy triggers its self-detachment device and the robot completes deployment, the crane retrieves the connecting rope. During the retrieval phase, after the operation is completed, the surface-powered buoy and other equipment are coordinated with the underwater detection robot to complete the retrieval of itself and its auxiliary submachine. In emergency response, if the underwater equipment is detected to have communication interruption, power abnormality, or attitude instability that cannot be recovered via remote command, the operator or intelligent control system generates a decision and sends an encrypted emergency jettison command to the underwater robot through the remaining communication link. After the robot's main control cabin verifies the command, it releases the mechanical lock of the counterweight. After the counterweight separates, the robot rises using positive net buoyancy, and the mooring reel releases the communication cable in a synchronized and orderly manner.

[0023] The surface dynamic buoy includes a positioning and navigation unit and a communication relay unit; The positioning and navigation unit is based on two physically isolated high-precision GPS modules, working in conjunction with the navigation control algorithm and integrity monitoring program built into the buoy control board. After the system is powered on, the control board first initializes and configures the two GPS modules, setting the output data format and baud rate of the GGA and RMC statements of the NMEA-0183 protocol, and enabling the data reception modes of RTK and PPP. Subsequently, the two GPS modules work in parallel, independently receiving satellite navigation signals and differential correction signals from the shore-based control center reference station, and asynchronously outputting raw data packets containing latitude and longitude, UTC time, and positioning status at high frequencies of 5Hz and 10Hz. The control board synchronously receives data through two independent UART serial ports and adds internal timestamps to achieve synchronization alignment. The navigation processor performs multi-source fusion processing on the data, using RTK... The system fixes the position data and performs consistency checks. When the module data is out of tolerance or invalid, it automatically isolates and alarms. Finally, it compares the fused actual position P1 with the preset target position P0, subtracts the two to calculate the position deviation ΔP, and the navigation control algorithm calculates the control command based on ΔP to drive the buoy propeller so that the buoy dynamically maintains itself near the target position. The communication relay unit establishes data interaction links with other surface buoys in the operating area to share the location and environmental information collected by each buoy. On the other hand, it serves as a communication relay station between the shore-based control center and the underwater detection robot, receiving control commands issued by the shore-based control center and forwarding them to the underwater detection robot. At the same time, it collects the status data, detection data, and buoy operation data transmitted back by the underwater robot, summarizes them, and transmits them back to the shore-based control center.

[0024] The surface dynamic buoy includes a deployment and recovery auxiliary unit and an active auxiliary positioning unit; During the deployment phase, the deployment and recovery auxiliary unit first securely connects to the underwater detection robot via a control rod. After receiving the target deployment location command from the shore-based control center, it navigates to the target point using a positioning and navigation unit, triggering a self-detachment device to release the control rod and complete the precise deployment of the underwater detection robot. Simultaneously, in the initial deployment phase, the self-detachment device connects to the crane at the shore-based control center. Once the buoy is hoisted to the surface deployment area and its position is confirmed to be safe, the self-detachment device disconnects from the crane, and the crane retrieves the rope. During the recovery phase, after receiving a recovery command from the shore-based control center, the unit coordinates for the underwater detection robot to return to the buoy and reconnects to the robot via the control rod. If assistance from the shore-based control center is required, the robot is recovered using the crane; otherwise, it is recovered autonomously using the buoy's power to tow the robot to the designated recovery area. The active assisted positioning unit first obtains the buoy's high-precision absolute geographic coordinates from the positioning and navigation unit; then, the positioning sonar system transmits acoustic signals of a specific frequency to the underwater detection robot; after receiving the response signal returned by the underwater detection robot, the unit calculates the straight-line distance between the underwater detection robot and the buoy by measuring the round-trip time between the interrogation signal and the response signal and multiplying it by the known underwater speed of sound; simultaneously, by analyzing the phase difference between multiple hydrophone array elements of the received signal and using a direction-of-arrival estimation algorithm, the unit calculates the azimuth angle of the signal transmitted by the underwater detection robot reaching the buoy. After obtaining the distance and azimuth angle, and combining the depth information provided by the robot's depth sensor, the three parameters of distance, azimuth angle, and depth are transformed into coordinates, specifically: Using the buoy as the origin, the robot's two-dimensional relative coordinates on the horizontal plane are calculated using azimuth and distance. Combined with the depth value, a three-dimensional rectangular coordinate system is constructed for the underwater inspection robot relative to the buoy, thus obtaining its complete real-time relative position coordinates. The calculated three-dimensional spatial coordinates of the robot relative to the buoy are then superimposed on the absolute geographic coordinates of the buoy itself, which are obtained through BeiDou high-precision positioning. Through spatial geometric transformation, this local vector of relative coordinates is transformed and superimposed onto the global geographic coordinate system with the absolute position of the buoy as the origin, thus calculating the absolute geographic coordinates of the underwater inspection robot. The absolute geographic coordinates are the coordinates of a specific geographic location on the Earth's surface obtained through high-precision positioning technology based on the BeiDou satellite navigation system. This coordinate data is then synchronously transmitted to the shore-based control center and the underwater inspection robot.

[0025] The underwater robot includes a detection module, a power module, a navigation module, and an auxiliary sub-machine. The detection module includes an image acquisition unit and an acoustic detection unit; The image acquisition unit is based on high-resolution wide-angle optical components and large target surface sensing elements, combined with an adaptive optical compensation mechanism and a high-power controllable light source. It actively compensates for underwater light attenuation to maintain the effective observation distance in turbid waters and acquire basic images of the dam. Then, it eliminates fog perception through real-time defogging algorithms and corrects color cast through color correction technology. Finally, it outputs clear images of the dam with low distortion and high contrast. Low distortion refers to suppressing the interference of the optical system and underwater environment on the geometric shape of the image, and high contrast refers to enhancing the difference in brightness between the normal area and the defective area of ​​the dam in the image, so that the boundary between the two areas is clear. The acoustic detection unit generates electrical pulse signals of a specific waveform from the signal generation module within the control board. These signals are then amplified by a power amplifier to drive a specific array element of the acoustic array to emit acoustic pulses. After the sound waves are reflected by the target, the echo signals are received by multiple array elements of the acoustic array, amplified by a preamplifier, converted into digital signals by a high-speed analog-to-digital converter, and transmitted to the signal processing unit. Subsequently, all time-delayed and weighted signals are coherently superimposed, and the beam pointing angle is changed by electronic scanning to form a focused scan. The processed signal amplitude and transit time information are then mapped onto a two-dimensional image plane through polar coordinate-Cartesian coordinate transformation to generate an acoustic image with a resolution of 0.5-2cm.

[0026] The detection module includes a multi-degree-of-freedom gimbal unit and a three-dimensional scanning unit; The gimbal controller in the multi-degree-of-freedom gimbal unit acquires the three-axis angular velocity, three-axis acceleration, and three-axis attitude angle data of the robot's main inertial navigation system, as well as the angular velocity data of the gimbal's built-in IMU, in real time via a high-speed bus, and assigns a unified high-precision timestamp to all data. The carrier IMU data is used as the primary observation, and the difference in angular velocity between the gimbal IMU and the carrier IMU is used as the auxiliary observation. The gimbal controller uses the carrier IMU data with the unified timestamp as the main driver for Kalman filter state prediction, and calculates the theoretical expected attitude of the gimbal at high frequency. Simultaneously, the difference in angular velocity between the gimbal IMU and the carrier IMU is used as the key observation input filter. The predicted and observed values ​​are optimally fused using Kalman gain. Then, the feedforward compensation torque and the feedback torque calculated by the attitude deviation using a PID controller are combined to form the total control torque, which is then used to drive the waterproof servo motor to adjust the gimbal attitude through a field-oriented control algorithm. The 3D scanning unit emits a coded laser grid pattern of a specific wavelength onto the dam surface via a laser projector; a rigorously calibrated high-resolution binocular camera synchronously acquires the modulated and deformed laser pattern on the dam surface, obtaining two digital images, left and right; based on the predetermined scanning accuracy and point cloud acquisition rate, a semi-global matching algorithm is used to perform dense stereo matching calculations on the image pairs to obtain a disparity map, which is then converted into an initial 3D point cloud based on the left camera coordinate system using triangulation principles; combined with synchronous positioning and map building algorithms, an iterative nearest-point algorithm is used to register the current frame point cloud with the global model, optimizing the relative pose transformation matrix of the scanning unit, and then registering the point cloud... The point cloud is incrementally fused into the global model. Finally, based on the high-precision global model, a region growing algorithm is used to automatically segment the defect region. The defect depth is quantified by calculating the Euclidean distance between the defect region point cloud and the reference model. Based on the segmented defect region point cloud, a convex hull construction algorithm is used to calculate the smallest convex polyhedron that can completely enclose all defect points. Then, by calculating the closed space enclosed between the convex hull model and the reference model representing the original intact state, the space is decomposed into multiple tetrahedrons and the volume is accumulated to calculate the missing and redundant volumes of the defect. Combining the obtained defect depth and area, the defect volume, maximum depth, surface area, and location are output.

[0027] The power module includes a vertical propulsion unit and a horizontal propulsion unit; The outer ring position controller of the vertical propulsion unit is based on the depth deviation. Calculate the target's vertical velocity Z t Z represents the target depth, and Z represents the actual depth. For depth deviation, the inner loop speed controller adjusts the speed according to the deviation. Calculate the total vertical thrust required V t Let F be the target vertical velocity and V be the actual vertical velocity. Since the two thrusters have identical performance and are symmetrically arranged, a thrust distribution algorithm is used to distribute F... t The signal is evenly distributed to the left and right vertical thrusters to generate the corresponding PWM control signal; The horizontal propulsion unit includes several propulsion components symmetrically distributed around the center of the robot. Each component has rated power and a predetermined maximum thrust. It adopts a ±180° deflection vector nozzle design, and the rotation speed and nozzle direction of different propellers can be adjusted.

[0028] The navigation module includes a data acquisition and processing unit and a cooperative positioning unit; The data acquisition and processing unit integrates a gyroscope, a quartz accelerometer, a depth pressure sensor, a Doppler log, and a magnetometer to acquire three-axis angular velocity, three-axis specific force, depth pressure value, ground velocity, and geomagnetic field vector in real time. It uses a strapdown inertial navigation algorithm to generate position, velocity, attitude, and depth information, specifically: Using the ω output from the gyroscope, the attitude matrix is ​​updated in real time using the quaternion method or the direction cosine method to calculate the roll angle φ, pitch angle θ, and yaw angle ψ. Then, the specific force f measured by the accelerometer is transformed from the vehicle coordinate system to the navigation coordinate system through the attitude matrix. Gravitational acceleration and Coriolis acceleration terms are subtracted, and the three-dimensional velocity V of the vehicle in the n-frame is obtained by a first integration. The displacement relative to the initial position is obtained by a second integration. Depth information is directly obtained by performing temperature compensation and static conversion on the depth pressure value. At the same time, a Kalman filter is used to fuse the position, velocity, and attitude calculated by SINS with the ground velocity and depth values ​​of the DVL sensor, suppressing SINS integral divergence and outputting optimized relative motion data.

[0029] When the cooperative positioning unit receives the acoustic signal emitted by the surface dynamic buoy to the underwater detection robot, it sends a response signal. The cooperative positioning unit obtains the real-time position data of the underwater detection robot relative to the surface dynamic buoy and the absolute geographical location of the surface dynamic buoy. After correcting the error by extended Kalman filtering, it outputs its own absolute geographical coordinates.

[0030] The auxiliary submachine includes a motion control unit, a detection and operation unit, and an adsorption control unit; After receiving instructions from the shore-based control center forwarded by the underwater inspection robot, the motion control unit first calculates the target motion direction and observation pose required for the robot to achieve path tracking based on the geometric features and operational requirements of the preset inspection path through the trajectory tracking controller. The preset inspection path is generated by the shore-based control center and issued in the form of a series of waypoints and parametric curves. The path geometric features include curvature, length, and priority. After receiving the instructions, the motion control unit parses the path data. The navigation module of the auxiliary submachine is the same as that of the underwater inspection robot, and the pose of the auxiliary submachine is obtained through the same method. The multi-degree-of-freedom motion mode of the auxiliary submachine is consistent with that of the underwater inspection robot. The detection unit determines the activation plan of the detection equipment according to the detection scenario. Then, after the motion control unit completes the pose adjustment, it starts the deep-water camera to collect image information of the dam surface and the narrow gap area, and various sensors simultaneously collect defect-related data. The adsorption control unit consists of a negative pressure pump, a suction cup, and a high-precision pressure sensor. When the main control system issues an adsorption command, the negative pressure pump immediately starts and runs at high speed, continuously discharging water from the adsorption chamber between the suction cup and the dam surface through a sealed pipeline, creating a negative pressure environment within the adsorption chamber. Simultaneously, the high-precision pressure sensor monitors the relative pressure value within the adsorption chamber in real time and feeds the pressure data back to the main control system via a communication link. The main control system employs a closed-loop PID control algorithm, comparing the real-time monitored pressure with a preset target negative pressure threshold. Based on the pressure deviation and the rate of change of the deviation, it dynamically adjusts the duty cycle of the PWM control signal output to the negative pressure pump motor, thereby precisely controlling the pumping speed and power of the negative pressure pump.

[0031] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A multi-machine collaborative system for underwater dam inspection, characterized in that: The multi-machine collaborative system includes a shore-based control center, a surface dynamic buoy, and an underwater detection robot; The shore-based control center serves as a primary decision-making hub. It uses a crane to lift the "buoy + underwater robot" to a safe water area and release it into the water. It receives positioning, status, and sensor data from both, acquires high-definition video via a fiber optic composite cable, processes it in real time, and dynamically displays the detection progress. It monitors the equipment status, identifying communication interruptions, power anomalies, and attitude instability as equipment malfunctions. In such cases, it first attempts to repair the equipment remotely. If communication and power are not restored, or attitude is not adjusted, the remote repair is deemed ineffective, triggering an emergency jettison command. After the mission is completed, it coordinates the recovery, deeply processes the data, and archives reports. The surface dynamic buoy is a two-level collaborative layer. After power-on, it initializes dual GPS and merges positioning, navigates to the target position via PID, and releases the robot after receiving instructions from the shore-based control center. It also starts the USBL positioning robot and transmits coordinates back, while serving as a relay to extend communication. During recovery, it guides the robot to float up and locks it. The underwater robot is a three-level execution core. After release, it activates the navigation and calculation system (SINS), positioning sonar (USBL), and Doppler log (DVL). It achieves full-degree-of-freedom motion through positioning via SINS+USBL+DVL and cascaded PID control of the thrusters. It uses an acoustic array, high-definition camera, and 3D scanning module for detection. In narrow areas, it releases auxiliary sub-machines for coordination. After receiving instructions, it floats up to dock with a buoy for recovery. In case of emergency, it jettisons ballast and floats up.

2. The multi-machine collaborative system for underwater dam inspection according to claim 1, characterized in that: The shore-based control center includes a mission and data management unit and a deployment, recovery, and emergency response unit. The task and data management unit formulates a global task plan and scheduling scheme based on the dam inspection requirements. It then establishes bidirectional data transmission with the surface dynamic buoy, underwater inspection robot, and auxiliary submachines through optical fiber, wireless communication, and an optoelectronic composite cable towed by an underwater robot. It receives attitude, depth, thruster status, and acoustic and optical sensor data transmitted back by the underwater robot in real time, and monitors the entire inspection operation process in real time. During this period, it also needs to issue instructions for gimbal pitch / roll control, camera parameter adjustment, and movement and inspection of auxiliary submachines, and monitor the status of each device simultaneously. The deployment, retrieval, and emergency response unit first deploys a crane and operators in an area near the pre-set deployment point. The crane lifts and transports the integrated surface-powered buoy and underwater detection robot smoothly to the pre-calculated safe deployment area. After the buoy triggers its self-detachment device and the robot completes deployment, the crane retrieves the connecting rope. During the retrieval phase, after the operation is completed, the surface-powered buoy and other equipment are coordinated with the underwater detection robot to complete the retrieval of itself and its auxiliary submachine. In emergency response, if the underwater equipment is detected to have communication interruption, power abnormality, or attitude instability that cannot be recovered via remote command, the operator or intelligent control system generates a decision and sends an encrypted emergency jettison command to the underwater robot through the remaining communication link. After the robot's main control cabin verifies the command, it releases the mechanical lock of the counterweight. After the counterweight separates, the robot rises using positive net buoyancy, and the mooring reel releases the communication cable in a synchronized and orderly manner.

3. The multi-machine collaborative system for underwater dam inspection according to claim 1, characterized in that: The surface dynamic buoy includes a positioning and navigation unit and a communication relay unit; The positioning and navigation unit is based on two physically isolated high-precision GPS modules, working in conjunction with the navigation control algorithm and integrity monitoring program built into the buoy control board. After the system is powered on, the control board first initializes and configures the two GPS modules, setting the output data format and baud rate of the GGA and RMC statements of the NMEA-0183 protocol, and enabling the data reception modes of RTK and PPP. Subsequently, the two GPS modules work in parallel, independently receiving satellite navigation signals and differential correction signals from the shore-based control center reference station, and asynchronously outputting raw data packets containing latitude and longitude, UTC time, and positioning status at high frequencies of 5Hz and 10Hz. The control board synchronously receives data through two independent UART serial ports and adds internal timestamps to achieve synchronization alignment. The navigation processor performs multi-source fusion processing on the data, using RTK... The system fixes the position data and performs consistency checks. When the module data is out of tolerance or invalid, it automatically isolates and alarms. Finally, it compares the fused actual position P1 with the preset target position P0, subtracts the two to calculate the position deviation ΔP, and the navigation control algorithm calculates the control command based on ΔP to drive the buoy propeller to keep the buoy dynamically near the target position. The communication relay unit establishes data interaction links with other surface buoys in the operating area to share the location and environmental information collected by each buoy. On the other hand, it serves as a communication relay station between the shore-based control center and the underwater detection robot, receiving control commands issued by the shore-based control center and forwarding them to the underwater detection robot. At the same time, it collects the status data, detection data, and buoy operation data transmitted back by the underwater robot, summarizes them, and transmits them back to the shore-based control center.

4. A multi-machine collaborative system for underwater dam inspection according to claim 1, characterized in that: The surface dynamic buoy includes a deployment and recovery auxiliary unit and an active auxiliary positioning unit; During the deployment phase, the deployment and recovery auxiliary unit first securely connects to the underwater detection robot via a control rod. After receiving the target deployment location command from the shore-based control center, it navigates to the target point using a positioning and navigation unit, triggering a self-detachment device to release the control rod and complete the precise deployment of the underwater detection robot. Simultaneously, in the initial deployment phase, the self-detachment device connects to the crane at the shore-based control center. Once the buoy is hoisted to the surface deployment area and its position is confirmed to be safe, the self-detachment device disconnects from the crane, and the crane retrieves the rope. During the recovery phase, after receiving a recovery command from the shore-based control center, the unit coordinates for the underwater detection robot to return to the buoy and reconnects to the robot via the control rod. If assistance from the shore-based control center is required, the robot is recovered using the crane; otherwise, it is recovered autonomously using the buoy's power to tow the robot to the designated recovery area. The active assisted positioning unit first obtains the buoy's high-precision absolute geographic coordinates from the positioning and navigation unit; then, the positioning sonar system transmits acoustic signals of a specific frequency to the underwater detection robot; after receiving the response signal returned by the underwater detection robot, the unit calculates the straight-line distance between the underwater detection robot and the buoy by measuring the round-trip time between the interrogation signal and the response signal and multiplying it by the known underwater speed of sound; simultaneously, by analyzing the phase difference between multiple hydrophone array elements of the received signal and using a direction-of-arrival estimation algorithm, the unit calculates the azimuth angle of the signal transmitted by the underwater detection robot reaching the buoy. After obtaining the distance and azimuth angle, and combining the depth information provided by the robot's depth sensor, the three parameters of distance, azimuth angle, and depth are transformed into coordinates, specifically: Using the buoy as the origin, the robot's two-dimensional relative coordinates on the horizontal plane are calculated using azimuth and distance. Combined with the depth value, a three-dimensional rectangular coordinate system is constructed for the underwater inspection robot relative to the buoy, thus obtaining its complete real-time relative position coordinates. The calculated three-dimensional spatial coordinates of the robot relative to the buoy are then superimposed on the absolute geographic coordinates of the buoy itself, which are obtained through BeiDou high-precision positioning. Through spatial geometric transformation, this local vector of relative coordinates is transformed and superimposed onto the global geographic coordinate system with the absolute position of the buoy as the origin, thus calculating the absolute geographic coordinates of the underwater inspection robot. The absolute geographic coordinates are the coordinates of a specific geographic location on the Earth's surface obtained through high-precision positioning technology based on the BeiDou satellite navigation system. This coordinate data is then synchronously transmitted to the shore-based control center and the underwater inspection robot.

5. A multi-machine collaborative system for underwater dam inspection according to claim 1, characterized in that: The underwater robot includes a detection module, a power module, a navigation module, and an auxiliary sub-machine. The detection module includes an image acquisition unit and an acoustic detection unit; The image acquisition unit is based on high-resolution wide-angle optical components and large target surface sensing elements, combined with an adaptive optical compensation mechanism and a high-power controllable light source. It actively compensates for underwater light attenuation to maintain the effective observation distance in turbid waters and acquire basic images of the dam. Then, it eliminates fog perception through real-time defogging algorithms and corrects color cast through color correction technology. Finally, it outputs clear images of the dam with low distortion and high contrast. Low distortion refers to suppressing the interference of the optical system and underwater environment on the geometric shape of the image, and high contrast refers to enhancing the difference in brightness between the normal area and the defective area of ​​the dam in the image, so that the boundary between the two areas is clear. The acoustic detection unit generates electrical pulse signals of a specific waveform from the signal generation module in the control board, which are then driven by a power amplifier to drive a specific array element group of the designated acoustic array to emit acoustic pulses. After the sound waves are reflected by the target, the echo signals are received by multiple array elements of the acoustic array, amplified by a preamplifier, converted into digital signals by a high-speed analog-to-digital converter, and transmitted to the signal processing unit. Then, all time-delayed and weighted signals are coherently superimposed, and the beam pointing angle is changed by electronic scanning to form a focused scan. The processed signal amplitude and transit time information are then mapped onto a two-dimensional image plane through polar coordinate-cartes transformation to generate an acoustic image with a resolution of 0.5-2cm.

6. A multi-machine collaborative system for underwater dam inspection according to claim 5, characterized in that: The detection module includes a multi-degree-of-freedom gimbal unit and a three-dimensional scanning unit; The gimbal controller in the multi-degree-of-freedom gimbal unit acquires the three-axis angular velocity, three-axis acceleration, and three-axis attitude angle data of the robot's main inertial navigation system, as well as the angular velocity data of the gimbal's built-in IMU, in real time via a high-speed bus, and assigns a unified high-precision timestamp to all data. The carrier IMU data is used as the primary observation, and the difference in angular velocity between the gimbal IMU and the carrier IMU is used as the auxiliary observation. The gimbal controller uses the carrier IMU data with the unified timestamp as the main driver for Kalman filter state prediction, and calculates the theoretical expected attitude of the gimbal at high frequency. Simultaneously, the difference in angular velocity between the gimbal IMU and the carrier IMU is used as the key observation input filter. The predicted and observed values ​​are optimally fused using Kalman gain. Then, the feedforward compensation torque and the feedback torque calculated by the attitude deviation using a PID controller are combined to form the total control torque, which is then used to drive the waterproof servo motor to adjust the gimbal attitude through a field-oriented control algorithm. The 3D scanning unit emits a coded laser grid pattern of a specific wavelength onto the dam surface via a laser projector; a rigorously calibrated high-resolution binocular camera synchronously acquires the modulated and deformed laser pattern on the dam surface, obtaining two digital images, left and right; based on the predetermined scanning accuracy and point cloud acquisition rate, a semi-global matching algorithm is used to perform dense stereo matching calculations on the image pairs to obtain a disparity map, which is then converted into an initial 3D point cloud based on the left camera coordinate system using triangulation principles; combined with synchronous positioning and map building algorithms, an iterative nearest-point algorithm is used to register the current frame point cloud with the global model, optimizing the relative pose transformation matrix of the scanning unit, and then registering the point cloud... The point cloud is incrementally fused into the global model. Finally, based on the high-precision global model, a region growing algorithm is used to automatically segment the defect region. The defect depth is quantified by calculating the Euclidean distance between the defect region point cloud and the reference model. Based on the segmented defect region point cloud, a convex hull construction algorithm is used to calculate the smallest convex polyhedron that can completely enclose all defect points. Then, by calculating the closed space enclosed between the convex hull model and the reference model representing the original intact state, the space is decomposed into multiple tetrahedrons and the volume is accumulated to calculate the missing and redundant volumes of the defect. Combining the obtained defect depth and area, the defect volume, maximum depth, surface area, and location are output.

7. A multi-machine collaborative system for underwater dam inspection according to claim 5, characterized in that: The power module includes a vertical propulsion unit and a horizontal propulsion unit; The outer ring position controller of the vertical propulsion unit is based on the depth deviation. Calculate the target's vertical velocity, where Z t Z represents the target depth, and Z represents the actual depth. For depth deviation, the inner loop speed controller adjusts the speed according to the deviation. Calculate the total vertical thrust required V t Let F be the target vertical velocity and V be the actual vertical velocity. Since the two thrusters have identical performance and are symmetrically arranged, a thrust distribution algorithm is used to allocate F... t The signal is evenly distributed to the left and right vertical thrusters to generate the corresponding PWM control signal; The horizontal propulsion unit includes several propulsion components symmetrically distributed around the center of the robot. Each component has rated power and a predetermined maximum thrust. It adopts a ±180° deflection vector nozzle design, and the rotation speed and nozzle direction of different propellers can be adjusted.

8. A multi-machine collaborative system for underwater dam inspection according to claim 5, characterized in that: The navigation module includes a data acquisition and processing unit and a cooperative positioning unit; The data acquisition and processing unit integrates a gyroscope, a quartz accelerometer, a depth pressure sensor, a Doppler log, and a magnetometer to acquire three-axis angular velocity, three-axis specific force, depth pressure value, ground velocity, and geomagnetic field vector in real time. It uses a strapdown inertial navigation algorithm to generate position, velocity, attitude, and depth information, specifically: Using the ω output from the gyroscope, the attitude matrix is ​​updated in real time using the quaternion method or the direction cosine method to calculate the roll angle φ, pitch angle θ, and yaw angle ψ. Then, the specific force f measured by the accelerometer is transformed from the vehicle coordinate system to the navigation coordinate system through the attitude matrix. Gravitational acceleration and Coriolis acceleration terms are subtracted, and the three-dimensional velocity V of the vehicle in the n-frame is obtained by a first integration. The displacement relative to the initial position is obtained by a second integration. Depth information is directly obtained by performing temperature compensation and static conversion on the depth pressure value. At the same time, a Kalman filter is used to fuse the position, velocity, and attitude calculated by SINS with the ground velocity and depth values ​​of the DVL sensor, suppressing SINS integral divergence and outputting optimized relative motion data.

9. A multi-machine collaborative system for underwater dam inspection according to claim 8, characterized in that: When the cooperative positioning unit receives the acoustic signal emitted by the surface dynamic buoy to the underwater detection robot, it sends a response signal. The cooperative positioning unit obtains the real-time position data of the underwater detection robot relative to the surface dynamic buoy and the absolute geographical location of the surface dynamic buoy. After correcting the error by extended Kalman filtering, it outputs its own absolute geographical coordinates.

10. A multi-machine collaborative system for underwater dam inspection according to claim 5, characterized in that: The auxiliary submachine includes a motion control unit, a detection and operation unit, and an adsorption control unit; After receiving the instructions from the shore-based control center forwarded by the underwater inspection robot, the motion control unit first calculates the target motion direction and observation pose required for the robot to achieve path tracking based on the geometric features and operational requirements of the preset inspection path through the trajectory tracking controller. The preset detection path is generated by the shore-based control center and issued in the form of a series of waypoints and parametric curves; the path geometry features include curvature, length, and priority; after receiving the command, the motion control unit parses the path data; The navigation module of the auxiliary submachine is the same as that of the underwater inspection robot, and the pose of the auxiliary submachine is obtained in the same way. The auxiliary submachine's multi-degree-of-freedom motion mode is consistent with that of the underwater inspection robot; The detection unit determines the activation plan of the detection equipment according to the detection scenario. Then, after the motion control unit completes the pose adjustment, it starts the deep-water camera to collect image information of the dam surface and the narrow gap area, and various sensors simultaneously collect defect-related data. The adsorption control unit consists of a negative pressure pump, a suction cup, and a high-precision pressure sensor. When the main control system issues an adsorption command, the negative pressure pump immediately starts and runs at high speed, continuously discharging the water in the adsorption chamber between the suction cup and the dam surface through a sealed pipeline, thus creating a negative pressure environment in the adsorption chamber. At the same time, the high-precision pressure sensor monitors the relative pressure value in the adsorption chamber in real time and feeds back the pressure data to the main control system through a communication link. The main control system adopts a closed-loop PID control algorithm, which compares the real-time monitored pressure with the preset target negative pressure threshold. Based on the pressure deviation and the rate of change of the deviation, it dynamically adjusts the duty cycle of the PWM control signal output to the negative pressure pump motor, thereby accurately controlling the pumping speed and power of the negative pressure pump.

Images