Seaplane model water-entry experiment synchronous control system and method
By using a central control module for unified scheduling and real-time wave phase prediction, the synchronization problem of the seaplane model water entry experiment system was solved, enabling precise water entry control in a dynamic wave environment and significantly improving the repeatability and reliability of the experiment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI JIAOTONG UNIV
- Filing Date
- 2026-01-12
- Publication Date
- 2026-07-21
AI Technical Summary
The existing seaplane model water entry test system lacks a unified timing coordination mechanism, resulting in millisecond-level delays between devices, making it impossible to achieve microsecond-level synchronization and ensuring precise matching between the water entry timing and wave phase.
A central control module is used for unified scheduling. Wave phase prediction is performed by combining real-time wave height data and propagation models. Through adaptive control with position feedback and error compensation, microsecond-level timing synchronization and accurate experimental testing of multiple devices are achieved.
It achieves precise matching of the model's entry into the water with a specific wave phase in a dynamic wave environment, improving the repeatability and reliability of the experiment. It also has online self-learning and adaptive capabilities, breaking through the limitations of fixed delay.
Smart Images
Figure CN121857464B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a technology in the field of seaplane design and testing, specifically a synchronous control system and method for seaplane model water entry experiments. Background Technology
[0002] The water entry process of seaplanes is a key research topic at the intersection of aeronautical and marine engineering, involving multiple aspects such as fluid dynamics, structural impact, and noise control. To study phenomena such as structural response, hydrodynamic loads, and cavitation formation during water entry, model testing in a controlled water tank is necessary. Currently, experimental systems typically consist of a wave generator, motion control platform, data acquisition system, and high-speed camera system. Existing technologies involve independent control of each device, lacking unified scheduling, leading to time asynchrony and inconsistent data acquisition; they do not consider wave propagation delay and model freefall time, making precise water entry at specific phases impossible; and without closed-loop error compensation, experimental deviations accumulate over time. Summary of the Invention
[0003] This invention addresses the problem that existing technologies lack a unified timing coordination mechanism, resulting in millisecond-level delays between systems, making it impossible to achieve microsecond-level synchronization and ensuring precise matching between the water entry timing and wave phase. It proposes a synchronous control system and method for seaplane model water entry experiments. Through unified scheduling by a central control module, it achieves microsecond-level timing synchronization (<100μs) for multiple devices, while simultaneously enabling accurate experimental testing based on adaptive control and error compensation using position feedback.
[0004] This invention is achieved through the following technical solution:
[0005] This invention relates to a synchronous control system for a seaplane model water entry experiment, comprising: a central control module, a wave generator control module, a motor control module, and a triggering and data acquisition module. The central control module performs real-time calculation and prediction of wave propagation delay and model release delay based on received real-time wave height data, preset wave parameters, and system calibration parameters, obtaining precise synchronous triggering commands for each module, and performs closed-loop error compensation based on post-experiment image analysis results. The wave generator control module generates and transmits wave generator drive signals based on the start command and wave parameters issued by the central control module, obtaining specific waves that meet experimental requirements. The motor control module controls the position and speed of the linear motor based on the start command and preset motion trajectory issued by the central control module, achieving precise translation of the aircraft model to the target release point. The triggering and data acquisition module processes the electromagnet demagnetization release of the model, high-speed camera image acquisition, and sensor data synchronous recording based on the synchronous triggering command issued by the central control module at precise moments, obtaining time-aligned water entry process images and load data.
[0006] The central control module includes a wave phase prediction unit, a release timing calculation unit, a synchronization command scheduling unit, and an error compensation decision unit. Specifically, the wave phase prediction unit calculates the wave propagation delay to the experimental water area based on real-time wave surface data collected by the wave height meter, preset wave parameters, and pre-calibrated wave attenuation coefficient, wave speed, and propagation distance information, obtaining the wave propagation delay required for the wave to form a specific phase at the target location. The release timing calculation unit performs a comprehensive calculation of the free fall time and horizontal motion time based on the instantaneous speed feedback of the motor, the initial height of the model, gravitational acceleration, and target wave phase information, using a kinematic model. The system processes the data to obtain the model release delay required for the model to reach the target phase from release. The synchronization command scheduling unit performs high-precision timing scheduling and network command encapsulation processing based on the system startup time, the calculated wave propagation delay, and the model release delay. This results in a time-division multiplexing sequence of synchronization trigger commands sent to the wave generator control module, the motor control module, and the triggering and data acquisition module. The error compensation decision unit performs actual water entry phase calculation processing based on image analysis and deep learning based on the water entry process image sequence information acquired by the high-speed camera. This yields the phase error of the current experiment, and the Gaussian process regression compensation model is updated based on this error to obtain the timing compensation amount for subsequent experiments.
[0007] This invention relates to a synchronous control method for a seaplane model water entry experiment based on the above-mentioned system, comprising:
[0008] Step 1, Experimental Initialization and Wave Generation Start-up, specifically includes:
[0009] 1.1 Central Control Module Startup and Synchronization: The central control module deployed in the main control unit starts up and establishes a network connection with the wave generator control unit, linear motor control unit, and trigger and data acquisition control unit to complete system clock synchronization.
[0010] 1.2 Experimental Parameter Initialization: Set the target wave phase for this experiment in the central control module. Wave type parameters, initial model height H, preset motor speed, wave attenuation coefficient α, wave speed c, and distance L1 from wave generator to experimental water area.
[0011] 1.3 Wave generation start command issuance: At the preset experimental start time T0, the central control module sends a start command to the wave generator control unit to control the wave generator to start generating waves.
[0012] Step 2, Real-time wave phase prediction and motor start-up control, specifically includes:
[0013] 2.1 Real-time acquisition of wave height data: Wave height data h(t) is continuously acquired by a wave height meter located near the wave generator and sent to the central control module in real time.
[0014] 2.2 Wave Propagation Delay Calculation: The wave phase prediction unit of the central control module calculates the wave propagation to the target experimental water area and the formation of the target phase based on real-time wave height data h(t), wave type, and pre-calibrated wave attenuation coefficient α, wave speed c, and propagation distance L1 using a wave propagation prediction model. The required precise delay Δt1 = f(h(t), α, c, L1, ).
[0015] 2.3 Motor start command issuance: At time T0 + Δt1, the central control module sends a motor start command to the linear motor control unit, controlling the linear motor to drive the aircraft model to start horizontal movement.
[0016] Step 3, Model Release Timing Calculation and Synchronization Trigger, specifically includes:
[0017] 3.1 Motor motion status monitoring: The central control module reads the speed feedback v(t) of the linear motor in real time.
[0018] 3.2 Calculation of release delay Δt2: The release timing calculation unit of the central control module calculates the release time based on the instantaneous speed of the motor v(t), the initial height H of the model, the gravitational acceleration g, and the target phase. The theoretical free fall time t of the aircraft model was calculated using a kinematic model. fall ( ), and taking into account the horizontal motion, the precise release delay Δt2 required for the model to enter the water at the target phase was calculated.
[0019] 3.3 Synchronous Triggering of Model Release and Data Acquisition: At precise time T0+Δt1+Δt2, the central control module sends a sequence of synchronous triggering commands to the data acquisition control unit, including: a model release command to control the electromagnet to demagnetize and release the model, a camera start command to trigger the high-speed camera to start shooting, and a data acquisition start command to start sensor data acquisition, and the model falls freely into the water.
[0020] Step 4, Experiment Termination and Error Compensation Decision, specifically includes:
[0021] 4.1 Experiment Termination Control: After the model enters the water and slides away from the experimental section (after a preset delay Δt3), the central control module sends a stop command to all clients to stop all equipment and data acquisition.
[0022] 4.2 Actual Water Entry Phase Calculation: Based on a sequence of images captured by a high-speed camera with strictly aligned timestamps, the actual water entry phase of the aircraft model's first contact with the wave surface is calculated through image analysis and a deep learning regression network. .
[0023] 4.3 Error Calculation and Compensation Model Update: Calculate the phase error ε = in this experiment. - The error compensation decision unit adds the data pair consisting of the state vector of this experiment (including wave parameters, motor speed, etc.) and the error ε to the training set of the Gaussian process regression (GPR) compensation model, updates the model online, and provides a predicted compensation amount for the calculation of the release delay Δt2 in the next experiment, thus realizing closed-loop control.
[0024] Technical effect
[0025] This invention presents a wave phase feedforward prediction algorithm based on real-time wave height data and a propagation model: utilizing real-time feedback from a wave height meter, combined with calibrated wave attenuation characteristics and propagation distance, it dynamically predicts the wave phase of the target water area at future moments, rather than using a fixed delay. A dynamic release timing calculation algorithm integrating real-time motor speed and a free-fall model: based on the instantaneous speed feedback of the linear motor, combined with the model height and gravitational acceleration, it calculates and dynamically adjusts the release time in real time, ensuring the model accurately reaches the predicted wave phase after free fall. A closed-loop error compensation method based on image sequence deep learning regression and incremental Bayesian optimization: by analyzing continuous image sequences captured by a high-speed camera, it uses a trained neural network to regress the actual entry phase, and uses the error to update the Gaussian process regression model online, achieving adaptive compensation for the release timing. Compared with existing technologies, this invention achieves accurate and repeatable matching of the model's entry moment with specific wave phases (such as wave crests and troughs) in dynamic, unsteady wave environments, overcoming the limitation of fixed delays in adapting to wave changes. This system enables online self-learning and adaptive capabilities, allowing it to automatically optimize subsequent control parameters based on historical experimental errors. This significantly improves the repeatability and reliability of multiple experiments under complex conditions, achieving a qualitative leap from "open-loop execution" to "closed-loop optimization." It also creates a collaborative control paradigm across mechanical motion, wave generation, and visual sensing. Through unified scheduling and real-time computation, it elevates the synchronization accuracy of multiple devices to the microsecond level and ensures strict temporal alignment between kinematic data and visual information. Attached Figure Description
[0026] Figure 1 This is a schematic diagram of the structure of the present invention;
[0027] Figure 2 This is a flowchart of an implementation example;
[0028] Figure 3 Here is a flowchart of the error compensation mechanism;
[0029] Figure 4 This is a schematic diagram illustrating the effect of an example. Detailed Implementation
[0030] like Figure 1 As shown, this embodiment illustrates a synchronous control system for a seaplane model's water entry experiment. The system includes a central control module, a wave generator control module, a motor control module, and a triggering and data acquisition module. The central control module calculates and predicts the wave propagation delay Δt1 and model release delay Δt2 in real time based on received real-time wave height data, preset wave parameters, and system calibration parameters. This yields precise synchronous triggering commands for each module, and closed-loop error compensation is performed based on post-experiment image analysis. The wave generator control module generates and transmits wave generator drive signals based on the start command and wave parameters issued by the central control module, generating specific waves that meet experimental requirements. The motor control module controls the position and speed of the linear motor based on the start command and preset motion trajectory issued by the central control module, achieving precise translation of the seaplane model to the target release point. The triggering and data acquisition module processes the electromagnet demagnetization release of the model, high-speed camera image acquisition, and sensor data synchronous recording based on the synchronous triggering command issued by the central control module at precise moments, obtaining time-aligned images of the water entry process and load data.
[0031] like Figure 2 As shown, this embodiment relates to a synchronous control method for a seaplane model entering the water, comprising:
[0032] Step 1, T0: Wave generation starts, the central control module sends a wave generator start command to the wave generator control unit;
[0033] Step 2: Real-time wave data analysis, calculating Δt1, specifically: based on the wave height meter measurement value... Combining the wave attenuation coefficient α, wave velocity c, distance L1 from the wave generator to the experimental water area, and waveform type (sine wave, solitary wave, etc.) calibrated before the experiment, the central control module calculates the future moment that satisfies the "target phase generation": And thus obtain .
[0034] Step 3, T0+Δt1: Motor starts, the central control module sends a motor start command to the linear motor control unit;
[0035] Based on the instantaneous motor speed v(t) feedback, combined with the initial aircraft bottom height H, gravitational acceleration g, motor travel distance L2, and wave phase requirements, Predict the time from free fall to water contact time of the aircraft model after release. The target release time is obtained by combining the results. .
[0036] Step 4, T0+Δt1+Δt2: Release the model + start the camera / sensor
[0037] The central control module sends a model release command, a camera start command, and a data acquisition start command to the data acquisition control unit. The electromagnet instantly loses its magnetism, and the model slides into the water.
[0038] Step 5, Δt3: The model leaves the experimental section and stops uniformly. The central control module simultaneously sends all stop commands to the three control units.
[0039] Step 6, as follows Figure 3 The error compensation mechanism is shown below, and it specifically includes:
[0040] Step i, Spatiotemporal context feature extraction: The continuous image sequence {I(tk), ..., I(t), ..., I(t+m)} captured by the high-speed camera with strictly aligned timestamps, covering the entire process before, during, and after entering the water, is input into a multi-branch feature extraction network containing a lightweight CNN encoder to obtain fused features.
[0041] The aforementioned multi-branch refers to:
[0042] Wavefront dynamic evolution branch: Stack three consecutive frames of images (I(t-1), I(t), I(t+1)) and input them into a dedicated 3D convolutional subnetwork to extract the spatiotemporal motion features of the wavefront and capture the propagation speed and acceleration of the waves.
[0043] Static structure enhancement branch: Apply a U-Net encoder to the current frame I(t) to extract high-resolution wavefront texture and model contour structure features, focusing on enhancing the water-air interface and model edges.
[0044] Prior physical information injection: Known prior information (such as the main wave frequency and wave number initially estimated from wave height data through fast Fourier transform) is used as an additional channel and fused with image features at a specific layer to guide the network to focus on waveforms that conform to physical laws.
[0045] Step ii, Wavefront Phase Continuous Curve Regression: The fused features are input into the regression module of the Temporal Convolutional Network (TCN) combined with the Long Short-Term Memory Network (LSTM). This regression module outputs a continuous wavefront phase function. Discrete sampling. For each column of pixels x in the image (corresponding to the horizontal position in actual space), predict its absolute phase value (0°-360°) in each frame t of the time series.
[0046] The Temporal Convolutional Network (TCN) combined with the regression module of the Long Short-Term Memory Network (LSTM) uses training data obtained by simultaneously measuring the wave surface time history of a series of locations in real water using a high-precision laser wave height array or binocular vision technology in a "calibration experiment" without model interference. This time history is converted into absolute phase and used as the "true value". The loss function adopts a combination of smooth L1 loss and periodic cosine similarity loss to ensure that the phase prediction is both accurate and meets the periodic continuity of the waveform.
[0047] Step iii: Accurate calculation of model motion trajectory and water entry moment: In another parallel lightweight branch, an attention-based decoder is used to segment and track the bottom contour of the aircraft model from the image sequence in real time, obtaining its position-time trajectory L in the world coordinate system. model After (t), the instant of entry into the water t0 is no longer determined by the abrupt change of single-frame features, but by solving the trajectory-wave surface spatiotemporal intersection problem to determine the spatiotemporal coordinates of the contact point.
[0048] In the space (x, t), find the expression L. model (t) and =The point at which spatial contact first occurs at a specific threshold (such as the 90° phase contour line representing the wavefront) (x contact, t contact ).
[0049] This method makes comprehensive use of information from the entire motion process and has a strong ability to resist single-frame noise.
[0050] Step iv, actual phase acquisition: obtain the calculated spatiotemporal coordinates (x, y) of the contact point. contact, t contact Substitute the continuous phase field output by the regression network Direct interpolation yields the accurate actual water ingress phase. .
[0051] Step v, error modeling and compensation quantity generation, specifically includes:
[0052] 5.1 Calculate the phase error of this experiment The error ε is treated as a complex function of the system state (wave conditions, mechanical parameters, etc.). A Gaussian process regression (GPR) model is established to correlate the state vector S (including wave parameters, motor speed, ambient temperature, etc.) of each experiment with the observed error ε.
[0053] 5.2 Before conducting the next experiment, based on the preset state S for the next experiment... next The error ε is predicted using a trained GPR model. pred The compensation strategy is as follows: , where: Δt nominalThis is a theoretically calculated value, and K is the gain coefficient determined based on system dynamics.
[0054] Step vi, Online Incremental Learning: After each experiment, new (S,ε) data pairs are added to the training set of the GPR model. An online sparse Gaussian process technique is employed to update the model in real time with limited computational resources.
[0055] Compared with existing technologies, this invention significantly improves phase matching capability; avoids internal clock drift; can calculate wave phase in experimental water area in advance; phase prediction accuracy is much higher than the "fixed delay method"; the model can stably reach the specified wave phase; ensures controllable model entry time from a physical principle; improves consistency of multiple experiments of the system; and is a universal solution applicable to different waves / different models / different heights.
[0056] The above-described specific implementations can be partially adjusted by those skilled in the art in different ways without departing from the principles and purpose of the present invention. The scope of protection of the present invention is defined by the claims and is not limited to the above-described specific implementations. All implementation schemes within the scope of the claims are bound by the present invention.
Claims
1. A synchronous control method for a seaplane model water entry experiment based on a synchronous control system for a seaplane model water entry experiment, characterized in that, The system comprises a central control module, a wave generator control module, a motor control module, and a triggering and data acquisition module. Specifically: the central control module calculates and predicts wave propagation delay and model release delay in real time based on received real-time wave height data, preset wave parameters, and system calibration parameters, obtaining precise synchronous triggering commands for each module, and performs closed-loop error compensation based on post-experiment image analysis results; the wave generator control module generates and transmits wave generator drive signals based on the start command and wave parameters issued by the central control module, obtaining waves that meet experimental requirements; the motor control module controls the position and speed of the linear motor based on the start command and preset motion trajectory issued by the central control module, achieving precise translation of the aircraft model to the target release point; and the triggering and data acquisition module processes the electromagnet demagnetization release model, high-speed camera image acquisition, and sensor data synchronous recording based on the synchronous triggering command issued by the central control module at precise moments, obtaining time-aligned images of the water entry process and load data. The aforementioned synchronous control of the water entry experiment includes: Step 1: Wave generation starts at time T0. The central control module sends a wave generator start command to the wave generator control unit. Step 2: Real-time wave data analysis, calculating Δt1, specifically: based on the wave height meter measurement value... Based on the pre-test calibrated wave attenuation coefficient α, wave velocity c, distance L1 from the wave generator to the experimental water area, and waveform type, the central control module calculates the future moment that satisfies the "target phase generation": And thus obtain ; Step 3: At time T0+Δt1, the motor starts. The central control module sends a motor start command to the linear motor control unit and, based on the feedback of the instantaneous motor speed v(t), combined with the initial aircraft bottom height H, gravitational acceleration g, motor running distance L2, and wave phase requirement φ, target Predict the time from free fall to water contact time of the aircraft model after release. The target release time is obtained by combining the results. ; Step 4: Release the model and start the camera / sensor at time T0+Δt1+Δt2: The central control module sends a model release command, a camera start command, and a data acquisition start command to the data acquisition control unit. The electromagnet instantly loses its magnetism, and the model slides into the water. Step 5: At time Δt3, the model leaves the experimental section and stops uniformly. The central control module simultaneously sends all stop commands to the three control units. Step 6, error compensation, specifically includes: Step i, Spatiotemporal context feature extraction: The continuous image sequence {I(tk), ..., I(t), ..., I(t+m)} captured by the high-speed camera with strictly aligned timestamps, covering the entire process before, during and after water entry, is input into a multi-branch feature extraction network containing a lightweight CNN encoder to obtain fused features; Step ii, Wavefront Phase Continuous Curve Regression: The fused features are input into the Temporal Convolutional Network (TCN) and the regression module of the Long Short-Term Memory Network (LSTM). The regression module outputs discrete samples of the continuous wavefront phase function φ(x,t). For each column of pixels x in the image, the absolute phase value corresponding to each frame t in the time series is predicted. Step iii: Accurate calculation of model motion trajectory and water entry moment: In another parallel lightweight branch, an attention-based decoder is used to segment and track the bottom contour of the aircraft model from the image sequence in real time, obtaining its position-time trajectory L in the world coordinate system. model After (t), the instant of entry into the water t0 is no longer determined by the abrupt change of single-frame features, but by solving the trajectory-wave surface spatiotemporal intersection problem to determine the spatiotemporal coordinates of the contact point; Step iv, actual phase acquisition: obtain the calculated spatiotemporal coordinates (x, y) of the contact point. contact , t contact Substituting the wavefront phase function φ(x, t) output by the regression network, the actual water ingress phase φ is obtained by direct interpolation. actual ; Step v: Error modeling and compensation quantity generation; Step vi, Online Incremental Learning: After each experiment, new (S,ε) data pairs are added to the training set of the GPR model, and the model is updated in real time with limited computing resources using online sparse Gaussian process technology. The error modeling mentioned above refers to: calculating the phase error By treating the error ε as a complex function of the system state, a Gaussian process regression model is established, which associates the state vector S, including wave parameters, motor speed, and ambient temperature, with the observed error ε.
2. The synchronous control method for seaplane model water entry experiment according to claim 1, characterized in that, The aforementioned multi-branch structure includes: Wavefront dynamic evolution branch: Stack three consecutive frames of images (I(t-1), I(t), I(t+1)) and input them into a dedicated 3D convolutional subnetwork to extract the spatiotemporal motion features of the wavefront and capture the propagation speed and acceleration of the waves. Static structure enhancement branch: Apply a U-Net structured encoder to the current frame I(t) to extract high-resolution wavefront texture and model contour structured features, focusing on enhancing the water-air interface and model edges; Prior physical information injection: The main wave frequency and wave number initially estimated from the wave height meter data are used as additional channels and fused with image features at a specific layer to guide the network to focus on waveforms that conform to physical laws.
3. The synchronous control method for seaplane model water entry experiment according to claim 1, characterized in that, The Temporal Convolutional Network (TCN) combined with the regression module of the Long Short-Term Memory Network (LSTM) uses the following training data: in a calibration experiment without model interference, a high-precision laser altimeter array or binocular vision technology is used to simultaneously measure the wave surface time history of a series of locations in real water, which is then converted into absolute phase as the true value; its loss function is a combination of smooth L1 loss and periodic cosine similarity loss.
4. The synchronous control method for seaplane model water entry experiment according to claim 1, characterized in that, The aforementioned compensation quantity generation refers to: before conducting the next experiment, based on the preset state S for the next experiment... next The error ε is predicted using a trained Gaussian process regression model. pred The compensation strategy is as follows: , where: Δt nominal This is a theoretically calculated value, and K is the gain coefficient determined based on system dynamics.
5. A synchronous control system for a seaplane model water entry experiment, characterized in that, Implement the method described in any one of claims 1-4.
6. The synchronous control system for the seaplane model water entry experiment according to claim 5, characterized in that, The central control module includes a wave phase prediction unit, a release timing calculation unit, a synchronization command scheduling unit, and an error compensation decision unit. Specifically, the wave phase prediction unit calculates the wave propagation delay to the experimental water area based on real-time wave surface data collected by the wave height meter, preset wave parameters, and pre-calibrated wave attenuation coefficient, wave speed, and propagation distance information, obtaining the wave propagation delay required for the wave to form a phase at the target location. The release timing calculation unit performs a comprehensive calculation of the free fall time and horizontal motion time based on the instantaneous speed feedback of the motor, the initial height of the model, gravitational acceleration, and target wave phase information, using a kinematic model. The system calculates the model release delay required for the model to reach the target phase from release. The synchronization command scheduling unit performs high-precision timing scheduling and network command encapsulation processing based on the system startup time, the calculated wave propagation delay, and the model release delay. This results in a time-division multiplexing sequence of synchronization trigger commands sent to the wave generator control module, the motor control module, and the triggering and data acquisition module. The error compensation decision unit performs actual water entry phase calculation processing based on image analysis and deep learning based on the water entry process image sequence information acquired by the high-speed camera. This results in the phase error of the current experiment, and the Gaussian process regression compensation model is updated based on this error to obtain the timing compensation amount for subsequent experiments.