Wave-current coupling scour simulation test method in high gravity field

Abstract

The application discloses a kind of wave-current coupling scour simulation test methods in supergravity field, it is related to the cross field of geotechnical engineering and ocean engineering.This method synchronously collects multi-source observation signals under supergravity environment, constructs phase synchronous observation sequence with wave phase as index.Using low-dimensional filtering-inversion integrated method based on wave phase harmonic expansion, combined with adaptive basis function and scour starting physical constraint, high-precision inversion bed surface shear stress field.Through calculating the sensitivity of shear stress to wave-current phase difference and constructing nonlinear controllable domain, carry out shear stress controllability analysis.Based on this, adopt double-layer control strategy, outer layer planning target phase trajectory, inner layer combined with bed surface evolution state switch linear or nonlinear compensation control structure, generate control command of driving wave-making and flow-making mechanism.The application solves the problem that existing technology cannot simulate the dynamic change of wave-current phase and lacks real-time feedback of shear stress field, realizes the accurate closed-loop simulation of wave-current coupling scour.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of geotechnical engineering and ocean engineering, in particular to a wave-current coupling scour simulation test method in a supergravity field. BACKGROUND

[0002] Port engineering, cross-sea bridge and submarine pipeline and other marine structures face serious wave-current scour challenges in long-term service, and the stability of their foundations is directly related to national energy and transportation safety. In-depth revelation of the wave-current coupling scour mechanism under complex marine environment and calculation of the bed shear stress evolution law are of great strategic significance for engineering safety evaluation and life prediction.

[0003] Currently, the research on local scour mainly relies on conventional gravity field (1g) model tests or early centrifugal model tests. In the existing centrifugal simulation technology, wave making and current making are usually operated independently, or only simple linear superposition of waves and currents can be achieved. For tests involving wave-current coupling, existing equipment usually uses mechanical locking to maintain a fixed wave-current phase relationship, such as simulating only the same direction or the opposite direction, and the bed monitoring mainly relies on sparse point sensors to obtain single-point time history data of flow velocity or pressure.

[0004] However, the existing technology has deep technical bottlenecks in simulating real marine environment and precise control, such as distortion of wave-current dynamic coupling effect simulation and lack of bed response feedback control. Specifically, the fixed phase loading mode cuts off the spatiotemporal relationship of the wave-current phase difference with the dynamic evolution of tides, and cannot reproduce the nonlinear modulation process of the shear stress by wave-current interaction; the traditional discrete point observation and time domain filtering method cannot separate the non-steady-state interference of the bed form (such as sand wave) evolution on the flow field, resulting in insufficient accuracy of the shear stress field inversion and inability to construct an effective mapping relationship between the phase difference and the shear stress; due to the lack of real-time evaluation of the controllability of the shear stress, the control system has difficulty in self-adaptive adjustment of the phase adjustment strategy when facing bed response lag or nonlinear saturation, leading to deviation of the test process from the preset scour evolution trajectory. SUMMARY

[0005] The application aims to provide a wave-current coupling scour simulation test method in a supergravity field to solve one of the above problems in the prior art.

[0006] The technical scheme provides a wave-current coupling scour simulation test method in a supergravity field, which comprises the following steps:

[0007] Read the multi-source sensor monitoring signals and wave making and current making motion state signals in the supergravity test environment, analyze the real-time wave phase and current phase, map the monitoring signals to the wave phase domain, and generate a phase-synchronous observation sequence;

[0008] Based on the phase-synchronous observation sequence, the fluid dynamic characteristics of the bed surface are inverted and analyzed to obtain the estimated data of shear stress field reflecting the stress state of the bed surface;

[0009] Based on the deviation between the estimated shear stress field data and the preset target shear stress index, and combined with the sensitivity characteristics of the shear stress field to changes in wave-current phase difference, wave-current phase control commands are generated for real-time adjustment of the phase difference between the wave-generating mechanism and the current-generating mechanism.

[0010] Beneficial effects: This invention solves the problems of existing technologies being unable to simulate dynamic changes in wave-current phase and lacking real-time feedback of shear stress field, and realizes accurate closed-loop simulation of wave-current coupled scouring. Attached Figure Description

[0011] Figure 1 This is a flowchart illustrating the steps of a wave-current coupling scour simulation test method in a hypergravity field according to an embodiment of this application.

[0012] Figure 2 This is a flowchart illustrating the steps involved in generating a phase-synchronized observation sequence in an embodiment of this application.

[0013] Figure 3 This is a flowchart illustrating the steps involved in calculating and evaluating the scour response in an embodiment of this application.

[0014] Figure 4 This is a flowchart illustrating the steps for generating the shear stress-phase difference correlation sequence and calculating the shear stress sensitivity index in this embodiment of the application.

[0015] Figure 5 This is a three-dimensional structural diagram of the centrifugal simulation system for the scouring process in a hypergravity field in this embodiment of the application.

[0016] Figure 6 This is a top view of the centrifugal simulation system for the scouring process in a hypergravity field, as described in this application.

[0017] Figure 7 For along Figure 6 Schematic diagram of the cross-sectional structure along line AA.

[0018] Figure 8 For along Figure 6 Schematic diagram of the cross-sectional structure of the middle BB line.

[0019] In the figure: 1, wave height phase sensor; 2, displacement sensor; 3, flow meter; 4, wave absorbing net; 5, annular water tank; 6, water tank support structure; 7, servo oil cylinder; 8, transparent organic glass; 9, high-precision camera; 10, water flow phase sensor; 11, phase output end; 12, phase control transmission mechanism; 13, phase input end; 14, miniature pore water pressure sensor; 15, laser profilometer; 16, hydraulic model; 17, honeycomb flow stabilizing plate; 18, shear stress sensor; 19, pulsating pressure sensor; 20, sandy soil; 21, soil sample box; 22, propeller. DETAILED DESCRIPTION

[0020] In conjunction Figures 1 to 8 with the description of the implementation details of the present application.

[0021] Example 1, a centrifugal simulation system capable of accurately realizing wave-flow coupling loading in a supergravity field, real-time measuring bed surface shear stress and capturing local scouring process is constructed, solving the problem that the existing equipment cannot realize wave-flow dynamic phase control and shear stress full-field measurement.

[0022] The wave-flow coupling scouring simulation system mainly consists of three core subsystems, namely the dynamic phase adjustable wave-flow coupling system, the multi-source data acquisition system and the bed surface shear stress inversion and control system. Among them, the wave-flow coupling system is responsible for generating waves and water flow in the model box at the end of the centrifuge arm, and realizing the dynamic adjustment of the phase between the two; the multi-source data acquisition system is responsible for synchronously acquiring various physical quantities in the test process; the bed surface shear stress inversion and control system is responsible for processing data and generating feedback control instructions.

[0023] Specifically, the dynamic phase adjustable wave-flow coupling system includes a wave making device, a flow making device and a phase control transmission mechanism. The wave making device adopts a wedge-shaped block wave maker, which is driven by a servo oil cylinder to make vertical reciprocating motion, generates waves by pushing away water, and sets a wave absorbing net at the end. The movement period of the wedge-shaped block can be adjusted in the range of 0-10s, and regular waves and irregular waves with a wave height of 0-10m can be generated. The flow making device adopts a propeller, which can generate constant flow or pulsating flow of 0-3m / s. The phase control transmission mechanism includes a harmonic reducer and a gear linkage device inside, the phase input end is connected to the servo drive shaft of the wave making wedge-shaped block, and the phase output end is connected to the motor shaft of the flow making propeller. By combining mechanical linkage with electronic servo control, the phase difference between wave generation and water flow pulsation is adjusted, and dynamic phase change of 0-180 degrees is realized.

[0024] In terms of multi-source data acquisition system, the embodiment is configured with a water dynamic sensor group, an erosion sensor group and a bed surface observation sensor group. The water dynamic sensor group includes a wave height phase sensor arranged upstream of the test section for measuring wave elevation changes; a flow velocity meter arranged in the near-bottom region for measuring water flow velocity vectors; and a fluctuating pressure sensor buried in the model surface for obtaining dynamic pressure distribution under wave flow action. The erosion sensor group includes a non-contact laser displacement sensor for real-time monitoring of bed surface erosion depth; and a miniature pore water pressure sensor buried in the sand inside the soil sample box for monitoring seepage and excess pore pressure response in the sand. The bed surface observation sensor group includes a particle image velocimetry (PIV) system for obtaining near-bottom flow field structure and a digital image correlation system for monitoring bed surface micro-deformation. In the test environment, a hydraulic model for simulating structures such as piles and piers is arranged for studying local scouring effects. A honeycomb flow stabilizer plate is installed to stabilize the flow field. All sensors are triggered by a high-precision synchronous clock, and the collected data have a unified time reference.

[0025] The bed surface shear stress inversion and control system, as the core of data processing and decision-making, is deployed in the upper computer workstation. The system receives multi-source sensor data, inverts the bed surface shear stress field through the built-in algorithm, calculates control instructions according to the preset test target such as the target shear stress time history or the target erosion depth, adjusts the motion parameters of the wave-making wedge blocks and the rotation speed parameters of the current-making propellers through the servo controller, and realizes closed-loop control of the wave flow coupling conditions.

[0026] In the embodiment, in order to adapt to the supergravity environment, the model box and all water-related parts are made of high-strength corrosion-resistant materials such as stainless steel or aluminum alloy. The sensor signal transmission uses shielded cables resistant to electromagnetic interference and is fixed along the inner wall of the model box to reduce interference with the flow field. When the centrifuge is running, the gravitational acceleration is set to N times the gravitational acceleration, and the model geometric scale is scaled down by 1 / N to meet the similarity criteria of centrifugal model tests.

[0027] Embodiment 2 describes how to accurately extract the phase information of waves and water flow from the original sensor signals and construct a synchronous observation sequence indexed by wave phase in a supergravity wave flow coupling test.

[0028] Step 201, read the multi-source sensor monitoring signals and wave and current motion state signals in the supergravity test environment, and analyze the real-time wave phase and water flow phase.

[0029] Specifically, during the test, the data acquisition system receives signals from various monitoring devices such as wave height phase sensors, flow meters, pressure sensors, and state feedback signals from wave makers and current generators servo controllers in real time. In order to obtain accurate phase reference, this embodiment uses a specific signal processing method to extract wave phase and flow phase respectively.

[0030] For the analysis of real-time wave phase, the wave height phase sensor signal arranged upstream of the test section is extracted from the monitoring signal of multiple sensors. Since the wave surface presents periodic fluctuations over time, the peak detection algorithm can be used to identify the wave peak time. Let t wave_peak be the time of the last detected wave peak, Tw be the current wave period, which can be calculated in real time by the time difference between adjacent wave peaks or given by the wave making instruction, then the real-time wave phase θ w at the current time t can be calculated by linear interpolation, the formula is: θ w = 360°* (t - t wave_peak ) / Tw. The phase value varies between 0 and 360°, and the wave peak is defined as 0° or 360°. In order to improve the anti-noise performance, the wave height signal can be low-pass filtered before peak detection to filter out high-frequency burr interference.

[0031] For the analysis of real-time flow phase, mainly for the case of pulsating flow. The rotational speed pulse signal or encoder feedback signal of the current generating propeller is extracted from the wave making and current generating motion state signal. Since there is a positive correlation between water flow velocity and propeller rotational speed: v = k * n, the pulsating period Tf of the flow velocity is usually determined by the adjustment period of the propeller rotational speed. By identifying the peak time t flow_peak of the rotational speed signal or flow velocity signal and the pulsating period Tf, the real-time flow phase θ f at the current time t can be calculated, the formula is: θ f = 360°* (t - t flow_peak ) / Tf. If it is a constant flow, the flow phase can be considered as a constant or not involved in the phase difference calculation.

[0032] On this basis, the real-time wave phase and real-time flow phase are timestamped and aligned using high-precision synchronous clock signals. Since the sampling rates of different sensors and control systems may be different, the system uniformly uses a master clock, such as a PTP protocol or a GPS clock, as a reference, and all data points are stamped with timestamps accurate to microseconds. By aligning the timestamps, the wave-flow phase difference Δ θ at the same time can be accurately calculated, θ w - θ f .

[0033] Step 202, map the monitoring signal to the wave phase domain to generate a phase-synchronized observation sequence.

[0034] This step is used to convert the time series data originally with time t as the independent variable into phase domain data with wave phase θ w as the independent variable.

[0035] Divide the time-continuous multi-source sensor monitoring signals into several complete wave period units based on real-time wave phase.

[0036] Specifically, the system detects θ w The moment when 360° jumps back to 0° or the moment when it passes through 0° is marked as the beginning of a new wave period. The data segment between two adjacent zero-phase moments constitutes a wave period unit. Each unit contains time series fragments of all sensors in the period, such as wave height, flow rate, pressure, shear stress, etc.

[0037] For the wave period unit, extract the signal instantaneous value at the preset discrete phase node to form a period segmented observation set.

[0038] In this embodiment, the preset discrete phase nodes can be uniformly distributed, for example, taking a point every 10°, such as 0°, 10°,..., 350°, a total of 36 nodes. For each node, find the data point at the corresponding phase position in the wave period unit. If the sampling time of the original data does not exactly fall on this type of phase node, use linear interpolation or spline interpolation method to calculate the signal instantaneous value at the phase node. For the kth wave period, the observation vector D k (θ) is obtained, where θ takes all preset phase nodes.

[0039] Statistically superimpose the period segmented observation set at the same discrete phase node to generate a phase-synchronous observation sequence with wave phase as the independent variable index.

[0040] Specifically, since the signal of a single measurement may contain random turbulent fluctuations and measurement noise, in order to extract the deterministic wave period component, this embodiment uses the phase average method. Superimpose and average the data D k (θ) (k=1...N) of N consecutive wave periods to obtain the phase-synchronous observation sequence D avg (θ) = (1 / N) * Σ(D k (θ)). In addition, statistical quantities such as standard deviation can also be calculated to represent the turbulence intensity or the stability of periodic changes.

[0041] In some optional embodiments, in order to cope with slight fluctuations in wave period, a dynamic time warping algorithm can be used to align data of different periods before superimposed averaging. For non-stationary processes such as variable wave height or variable period working conditions, a sliding window average can be used, that is, only the last M periods are superimposed to retain the time-varying characteristics of the process.

[0042] It should also be noted that the generated phase-synchronized observation sequence not only contains the mean value of the original physical quantity, but also constructs a bidirectional index mapping structure between the time domain and the phase domain. The index table records which original time points the phase domain data points come from and the corresponding wave period ID. This bidirectional indexing mechanism enables rapid backtracking to the original time series for verification once an anomaly is found in the phase domain data, or accurately mapping the inversion results of the phase domain back to the time axis for real-time control.

[0043] Embodiment 3 describes the process of harmonic inversion based on physical constraints and multi-source fusion.

[0044] Based on the phase-synchronized observation sequence, the bed surface fluid dynamics characteristics are inverted to obtain the shear stress field estimation data reflecting the stress state of the bed surface.

[0045] This embodiment details a shear stress field inversion method based on wave phase harmonic expansion, which uses a phase-synchronized observation sequence to construct a harmonic model containing physical constraints, achieving high-precision, low-dimensional reconstruction of the bed surface shear stress field. Specifically:

[0046] Step 301, a preset shear stress harmonic model describing the change of bed surface spatial node shear stress with wave phase is adopted, and the model takes the harmonic coefficient vector and the phase parameter vector as the to-be-solved state quantities.

[0047] In this step, based on the periodic characteristics of wave action, it is assumed that the shear stress τ(x, t) at any point x on the bed surface can be represented as the superposition of the fundamental wave and its higher harmonics. Since the data has been mapped to the phase domain, the independent variable here is the wave phase θ. The shear stress harmonic model is specifically expressed as: τ(x, θ) = a0(x) + Σ m=1 M [a m (x) * cos(m * θ +φ m (x))] Where a0(x) represents the average shear stress, mainly contributed by the steady flow, M is the truncation order, for example, 3 or 5, a m (x) is the amplitude coefficient of the mth harmonic, and φ m (x) is the phase parameter of the mth harmonic. The to-be-solved state quantity is the parameter vector S(x) = [a0, a1,..., a M, φ1,..., φ M This modeling approach compresses the high-dimensional shear stress field, which changes rapidly with time, into a low-dimensional parameter field, which changes slowly with space, reducing the dimension and computational cost of the inversion problem.

[0048] At step 302, the characteristics of the bed elevation change in the multi-source sensor monitoring signal are analyzed to determine the bed evolution stage marker corresponding to the current monitoring area, including the flat bed stage or sand wave development stage, and to select the harmonic basis function family accordingly.

[0049] To improve the adaptability of the model, an adaptive basis function selection mechanism is introduced. The bed elevation data z bed (x, t) obtained by the laser profiler or multi-beam depth sounder is used to calculate its spatial spectrum or spatial gradient. If the root mean square height of the bed elevation is less than a preset threshold, and there is no obvious main peak in the spatial spectrum, it is determined to be a flat bed stage; if there is a spatial periodic fluctuation, and the wave form presents obvious asymmetry, the upflow surface is gentle and the backflow surface is steep, it is determined to be a sand wave development stage.

[0050] When the bed evolution stage marker is the flat bed stage, the symmetric harmonic basis function family composed of sine and cosine functions is selected. At this time, the standard Fourier series expansion described above is sufficient to describe the symmetric periodic change of the shear stress.

[0051] When the bed evolution stage marker is the sand wave development stage, the asymmetric harmonic basis function family containing asymmetric wave form characteristic functions is selected. For example, a second-order skew term can be introduced in the basis function, or a non-sine basis function such as the Jacobi elliptic function is used to better capture the shear stress mutation and asymmetry caused by flow separation on the backflow surface of the sand wave.

[0052] Using the selected harmonic basis function family, the function expression form of the shear stress harmonic model is constructed.

[0053] At step 303, an optimization solving system is established to minimize the difference between the output of the shear stress harmonic model and the phase-synchronized observation sequence, and a scour initiation physical constraint term is introduced.

[0054] To solve the above model parameters, the following optimization objective function J is constructed: J = J data + λ * J smooth +γ *J physics Where J data is the data fidelity term, representing the difference between the model calculated shear stress and the measured or calculated shear stress through flow rate in the phase-synchronized observation sequence, usually in the form of least squares: Σ ||G(S) - D obs || 2G is an observation operator that maps the shear stress parameters to the observation space, such as the velocity gradients measured by PIV. D obs is the actual observation sequence. J smooth is a spatial smoothing regularization term to suppress non-physical spatial oscillations, such as:∑||Laplacian(S)||2 2 λ is a regularization parameter.

[0055] In particular, the embodiment introduces J physics as a scour initiation physical constraint term. A critical shear stress threshold τ cr is introduced to describe the scour initiation condition of the sediment used in the experiment. The threshold can be determined by the Hillz curve according to the particle size of the sediment.

[0056] Based on the critical shear stress threshold and the shear stress amplitude calculated by the shear stress harmonic model, a scour initiation physical constraint term is constructed, which is configured to only numerically accumulate the spatial distribution area where the shear stress amplitude exceeds the critical shear stress threshold.

[0057] The specific form can be a penalty function: J physics =∫[max(0, |τ(x, θ)|-τ Max_physical ) 2 ]dx or a constraint for the initiation condition, the shear stress in the area where scour does not occur should not exceed τ cr In the embodiment, the constraint term is preferably used to punish the area where extremely high shear stress is physically impossible, such as the non-scour hot spot area, to improve the physical reasonableness of the inversion result. The scour initiation physical constraint term is added to the objective function of the optimization solution system as a weighted penalty component, and the weight coefficient γ can be adjusted according to the signal-to-noise ratio.

[0058] Step 304, using the optimization solution system to calculate the optimal harmonic coefficient vector and phase parameter vector, and reconstructing the shear stress field estimation data with unified spatial and temporal resolution.

[0059] A nonlinear optimization algorithm such as Levenberg-Marquardt algorithm or sequential quadratic programming method is used to solve the objective function J to obtain the optimal state parameter vector S opt (x). After obtaining the optimal parameters, they are substituted into the harmonic model formula, and the shear stress value τ(x, θ) at any phase θ and any position x can be reconstructed. Since θ and time t have a clear mapping relationship, the shear stress field estimation data τ(x, t) with unified spatial and temporal resolution can be further generated.

[0060] In some preferred embodiments, in order to further improve the accuracy, especially for local high gradient areas such as scour pit edges, a multi-source data fusion strategy is adopted. The traditional inversion, such as based on extended Kalman filter or simple interpolation, obtains the shear stress field as the basis, that is, the low frequency component, and the parameters obtained by the above harmonic inversion extract high frequency detailed features, such as high harmonic components, and the two are superimposed.

[0061] The specific operation can be: in the identified high shear stress gradient area, the weight of the high harmonic is increased; and in the flat area, the base wave or the traditional inversion result is mainly relied on. The final output shear stress field not only retains the macro trend of the whole field, but also clearly depicts the fine structure of the local scour hot spot.

[0062] Embodiment 4, constructing shear stress controllability analysis and nonlinear controllable domain.

[0063] According to the deviation of the shear stress field estimation data and the preset target shear stress index, and combining the response sensitivity characteristics of the shear stress field to the wave flow phase difference change, a wave flow phase control instruction for real-time adjusting the phase difference between the wave making mechanism and the flow making mechanism is generated.

[0064] Due to the evolution of the bed morphology, such as the generation and movement of sand waves and the nonlinear hydrodynamic effect in the supergravity environment, the influence of the wave flow phase difference on the bed shear stress is not constant, but shows time-varying and state-dependent. Therefore, it is necessary to evaluate whether the adjustment of the phase difference is effective and what range it should be adjusted to in the current working condition.

[0065] Step 401, generating a shear stress-phase difference correlation sequence and calculating a shear stress sensitivity index.

[0066] The key shear stress index sequence used as the control object is extracted from the shear stress field estimation data. The key shear stress index here is a scalar or vector used to represent the current scour state, which can specifically include: the spatially averaged effective shear stress in the monitoring area, the maximum shear stress peak value, or a specific scour hotspot position, such as the local shear stress around the structure. The wave flow phase difference sequence is obtained by using the real-time wave phase and the real-time flow phase, and calculating the difference between the real-time wave phase and the real-time flow phase. The shear stress index sequence and the wave flow phase difference sequence are aligned on the time axis to form a correlation sequence.

[0067] On this basis, the sliding time window technique is used to fit the local relationship between the key shear stress index sequence and the wave flow phase difference sequence, and the shear stress sensitivity index reflecting the response intensity of the shear stress to the unit phase difference change is calculated. A sliding window containing several wave periods is set, and the change rate of the shear stress index with respect to the wave flow phase difference is fitted in the window. This embodiment proposes a multi-time scale sensitivity analysis mechanism:

[0068] Based on the time scale of a single or a few wave periods, e.g. 1-3 periods, in the pre-set monitoring signals of multi-source sensors, the linear regression or difference method is used to fit the quick response relationship between the sequence of key shear stress indicators and the sequence of wave-current phase difference, to obtain the short-term instantaneous sensitivity κ short , for the immediate control effect of the phase difference adjustment on the shear stress.

[0069] Based on the time scale of the evolution of the bed surface morphology, e.g. dozens of periods or sand wave migration characteristic time, in the pre-set phase synchronization observation sequence, the long-term cumulative trend of the sequence of key shear stress indicators with the sequence of wave-current phase difference is fitted, to obtain the medium-term evolution trend parameter κ Medium , for evaluating the long-term impact of phase difference adjustment on the evolution direction of the bed surface.

[0070] Step 402, construct the nonlinear wave-current phase difference controllable domain.

[0071] In order to overcome the limitations of traditional linear control in the strong nonlinear region, the concept of nonlinear wave-current phase difference controllable domain is introduced in this embodiment. The controllable domain Ω(t) is defined as the set of all possible wave-current phase difference values that can make the predicted shear stress indicator maintain in the pre-set target range, i.e. the safety interval, at the current time t.

[0072] Based on the short-term instantaneous sensitivity and the medium-term evolution trend parameter, the possible wave-current phase difference value interval is scanned and predicted at the current time section, to determine the value set that can make the predicted shear stress value fall within the target range, forming the nonlinear wave-current phase difference controllable domain.

[0073] The specific construction process is as follows: the shear stress harmonic model is used as a fast predictor. At the current time section, the bed surface morphology and fluid parameters are kept unchanged, and the possible value interval of the wave-current phase difference Δθ is virtually scanned from 0 to 360° or 0 to 2π radians. For the virtual phase difference values in the scanning interval, the corresponding predicted shear stress indicators are calculated by substituting the harmonic model. The prediction results are compared with the upper limit τ Max and the lower limit τ Min of the target shear stress indicator. If the predicted value falls within the interval [τ Min , τ Max ], the virtual phase difference value is included in the controllable domain Ω(t).

[0074] The target shear stress index and the shear stress sensitivity index are combined to determine the effective adjustment range of the wave-current phase difference that can keep the shear stress within the preset safety interval, and the shear stress controllability analysis data is generated. The final generated shear stress controllability analysis data not only includes the above-mentioned sensitivity index κ and controllable domain Ω(t), but also includes controllability level determination. For example, when Ω(t) is empty, it is determined to be uncontrollable, at which time the system may need to alarm or adjust the wave height flow and other parameters; when the range of Ω(t) is wide and the sensitivity κ is moderate, it is determined to be highly controllable, which is suitable for fine adjustment.

[0075] In some optional embodiments, for the calculation of the sensitivity index, the recursive least squares method can be used to estimate the sensitivity coefficient online, and a forgetting factor can be introduced to adapt to the time-varying characteristics of the system. When constructing the controllable domain, the physical constraints of the actuator, such as the upper limit of the phase adjustment speed, can also be considered, and the phase difference values that cannot be reached in the next control period are removed from the controllable domain to obtain the dynamic reachable controllable domain.

[0076] Embodiment 5 describes a specific implementation of wave-current phase closed-loop control using controllability analysis results. This embodiment adopts a double-layer control structure and introduces a controller structure switching mechanism based on the physical state of the bed surface, solving the control problems caused by the response lag and nonlinear enhancement of the shear stress in the sand wave development process.

[0077] Step 501, outer control loop: calculate the target wave-current phase difference trajectory.

[0078] The outer control loop is mainly responsible for path planning. In the outer control loop, the deviation e τ between the current shear stress field estimation data and the target shear stress index is calculated, and the shear stress controllability analysis data and the nonlinear wave-current phase difference controllable domain are used as constraint conditions. According to the sign of the deviation e τ and the sign of the short-term instantaneous sensitivity κ short , the direction of phase difference adjustment is determined. For example, if the current shear stress is too large and the sensitivity is positive, the phase difference should be reduced.

[0079] Next, the target wave-current phase difference trajectory Δθ target in the next control period is calculated. The calculation logic is: Δθ target = Δθ current + α * (e τ / κ short ). Wherein, α is the outer adjustment gain. In order to ensure safety, the calculated Δθ targetThe actual phase difference must be limited within the controllable domain Ω(t) of nonlinear wave-current phase difference. If the calculated value exceeds the boundary of Ω(t), the boundary value is taken as the target, or the optimal working point within Ω(t) is searched according to the medium-term evolution trend parameter. In this process, the control target is always located within the physically safe region.

[0080] Step 502, inner layer control loop: instruction generation based on bed state switching.

[0081] In the inner layer control loop, the actual phase difference is calculated by the difference between the real-time wave phase and the real-time flow phase, and the wave-flow phase control instruction driving the wave-making mechanism and the flow-making mechanism to execute the phase adjustment action is generated based on the tracking error of the target wave-flow phase difference trajectory. The inner layer control loop is responsible for high-frequency tracking of the target trajectory given by the outer layer. The system analyzes the bed surface morphology estimation data, and determines the current bed evolution stage marker by calculating the spatial spectrum characteristics of the bed surface elevation, such as wavelength, wave height or topographic gradient. The inner layer control loop is configured with a controller structure switching mechanism based on the physical state of the bed, which analyzes the spatial spectrum characteristics and topographic gradient in the bed surface morphology estimation data to determine the current bed evolution stage marker, including the flat bed stage and the sand wave development stage.

[0082] When the bed evolution stage marker is the flat bed stage, the response of shear stress to phase difference is approximately linear and has no hysteresis. At this time, the linear phase tracking control structure is activated, and the linear feedback algorithm is used to calculate the wave-flow phase control instruction to achieve rapid tracking of the target wave-flow phase difference trajectory. This structure uses an incremental PID control algorithm. Specifically, the difference between the target wave-flow phase difference trajectory at the current sampling time k and the actual phase difference is calculated to obtain the phase tracking error e(k). The feedback adjustment amount Δu(k) is calculated using the proportional coefficient Kp, the integral coefficient Ki and the differential coefficient Kd. The specific formula is: Δu(k) = Kp * [e(k) - e(k-1)] + Ki * e(k) + Kd * [e(k) - 2e(k-1) + e(k-2)]. The first derivative of the target trajectory is calculated and multiplied by the feedforward coefficient kf to generate the feedforward compensation amount u ff The final wave-flow phase control instruction is the sum of the feedback adjustment amount and the feedforward compensation amount.

[0083] When the bed evolution stage marker is the sand wave development stage, the shear stress response caused by vortex shedding on the back flow surface of the sand wave has a nonlinear hysteresis effect, and a simple linear PID cannot maintain the control quality. At this time, the system activates the nonlinear compensation control structure, which superimposes a nonlinear compensation term on the basis of the linear feedback algorithm to offset the asymmetric shear stress response hysteresis caused by the sand wave terrain and generate the wave-flow phase control instruction. On the basis of the above linear PID algorithm, a nonlinear compensation term u nlThe compensation term is constructed based on the asymmetric term coefficient in the shear stress harmonic model, such as the second-order harmonic amplitude, to pre-compensate the shear stress waveform distortion caused by sand wave topography. In this stage, the system will dynamically adjust the PID parameters according to the sensitivity index, for example, appropriately increasing Kp when the sensitivity decreases to enhance the tracking ability.

[0084] In some preferred embodiments, the inner control loop further comprises a dead zone link. When the phase tracking error e(k) is less than a preset dead zone threshold, such as a phase difference of 1°, no adjustment action is taken to avoid mechanical wear caused by frequent fine tuning of the actuator. The control command will be subjected to amplitude limiting and rate limiting processing before being output to the wave and current generating mechanism, and the command will be within the physical safe operating range of the device.

[0085] Example 6 describes how to use multi-source data to calculate and evaluate the test results after completing a single round of wave-current coupling scouring test, and optimize the test parameters for the next round of test, forming a large closed loop of test-evaluation-optimization.

[0086] Step 601, scour response calculation and evaluation.

[0087] Based on the wave height data, flow velocity data and fluctuating pressure data in the multi-source sensor monitoring signal, the wave alone acting shear stress, the flow alone acting shear stress and the wave-current coupling interaction shear stress are calculated respectively; the wave alone acting shear stress, the flow alone acting shear stress and the wave-current coupling interaction shear stress are superimposed to obtain the effective shear stress.

[0088] After executing a round of wave-current phase control command, the system collects the shear stress field estimation data, bed scour depth monitoring data and fluctuating pressure data throughout the process. The ratio of the effective shear stress to the sum of the wave alone acting shear stress and the flow alone acting shear stress is calculated to generate the coupling effect coefficient for calculating the wave-current synergistic enhancement or weakening degree.

[0089] To calculate the wave-current coupling effect, the coupling effect coefficient η is used for evaluation in this embodiment. The calculation formula is: η = τ couple / (τ wave + τ current ). Wherein, τ couple is the measured effective shear stress, τ wave and τ current are the shear stress components calculated according to the wave and flow theory respectively. If η > 1, it indicates that the wave-current coupling promotes scouring; if η < 1, it indicates that it inhibits scouring.

[0090] A calculation correlation model of scouring rate is established. Based on the physical model in the disclosure, the scouring rate v erode can be expressed as: v erode = k * τcouple α * ΔP β Where k, a, b are undetermined coefficients, and ΔP is the fluctuating pressure. The measured scour depth-time curve, i.e. erode The measured values and the inverted shear stress and pressure data are used to fit the above coefficients k, a, b by multivariate nonlinear regression. Such coefficients reflect the unique scouring rules of the current sand material in the supergravity environment.

[0091] Step 602, parameter iteration and next round target generation.

[0092] Based on the above evaluation results, scour response evaluation data is generated. If the scour effect of the current test does not reach the expected failure mode, for example, the scour pit depth is insufficient, or the shear stress fluctuation in the control process is too large, the system will correct the test loading control target data of the next round.

[0093] The specific correction strategy includes: according to the fitted scour rate model, the ideal shear stress trajectory required to reach the target scour depth is back calculated, and the target shear stress index data is updated. Using controllable domain information, the change mode of the target wave phase difference is re-planned. For example, if it is found in the last round of test that the coupling coefficient η is maximum near 90 degrees in a certain phase difference interval, and the control sensitivity is higher, the holding time of the phase difference interval is increased in the next round of test, and the scouring process is accelerated.

[0094] The updated test loading control target data, including the new target wave sequence, flow sequence and phase difference planning, is included as the input of the next round of test, and the online control process is restarted. Through multiple iterations, the local scour mechanism of complex marine structures can be accurately simulated and the ultimate bearing capacity can be accurately determined.

[0095] In summary, the present application shows a complete supergravity wave-current coupling scour simulation technical solution from bottom layer hardware perception to high layer algorithm decision, which solves the problem of not being able to see by using phase synchronization and harmonic inversion, solves the problem of not being able to control by using controllability analysis and double-layer switching control, and realizes real wave-current coupling closed-loop simulation.

[0096] Embodiment 7, describes instantaneous shear stress rapid estimation and multi-modal fusion inversion.

[0097] Although the harmonic inversion method has advantages in dealing with periodic wave characteristics, in the short time scale of real-time control, or in the initial loading stage when the harmonic characteristics have not yet stabilized, instantaneous estimation based on physical approximation and state fusion based on traditional filtering are still key means to ensure system robustness.

[0098] Step 701, generating instantaneous shear stress preliminary calculation data.

[0099] To meet the low latency requirements of real-time control, typically millisecond-level latency is required. This step employs an explicit formula based on the boundary layer physical approximation to directly calculate the instantaneous shear stress. The system extracts the vertical component of flow velocity and the horizontal component of water pressure in the near-bed region from multi-source sensor monitoring signals. Specifically, flow velocity data at locations very close to the bed surface, such as 0.5 mm to 2 mm in height, is acquired using a PIV system or a high-frequency point-type velocimeter. Combined with this measurement point height, the near-bed flow velocity gradient du is calculated. dz The pressure gradient dp is calculated using an array of pulsating pressure sensors embedded in the bed surface along the flow direction. dx .

[0100] Based on the boundary layer momentum equation approximation in fluid dynamics, the initial calculation data τ0 of the instantaneous shear stress reflecting the transient forces acting on the bed surface is obtained. The calculation formula is: τ0 = μ * du dz + k p * dp dx Where μ is the dynamic viscosity coefficient of the fluid, and for sediment-laden water flows, this coefficient needs to be corrected according to the sediment concentration; k p This is the pressure gradient correction coefficient, used to characterize the nonlinear contribution of the pressure gradient to the bed surface shear stress. This coefficient was determined through previous clean water calibration tests. This calculation process does not involve complex iterative solutions, therefore the calculation speed is extremely fast, and it can serve as a rapid feedback signal in the real-time control loop.

[0101] Step 702: Construct a multimodal fusion filtering system.

[0102] To overcome the problems of high noise and insufficient spatial coverage of single sensors, this embodiment constructs a multi-modal fusion filtering module. This module aims to establish a unified state vector and observation equation. The state vector s(t) is defined to include three core physical quantities: the bed shear stress field τ(x, t), the near-bed velocity profile feature u(z, t), and the bed elevation z. bed (x, t). That is, s(t) = {τ, u, z} bed}

[0103] Constructing nonlinear state transition equations s k +1 = f(s k ) + w k The function f(s) k This describes the evolution of the system state over time, utilizing the sediment initiation and transport equations to describe the coupling relationship between shear stress and bed elevation changes. k Characterize the uncertainty and noise of the model. Construct the observation equation d. k = H k * s k + vk where d is the observation vector k contains real-time readings of all sensors, such as PIV velocity, pressure, laser displacement, the observation matrix H k describes how the state variable maps to the observation space, v k is the observation noise.

[0104] Step 703, performing extended Kalman filtering and regularization inversion.

[0105] The above state space model is solved recursively using the extended Kalman filtering algorithm. At each time step, the state estimation value at the last time is used to perform time update according to the state transition equation, that is, the prediction step, to obtain the prior state estimation; the current observation vector d k and the observation equation are used to perform measurement update, that is, the correction step, to obtain the posterior state estimation τ hat (x, t) and the corresponding error covariance matrix P(x, t).

[0106] In some preferred embodiments, in order to obtain a higher spatial resolution shear stress field, the system periodically, for example, every 10 time steps, runs a regularization high-resolution inversion module. This module takes the posterior estimation output by the extended Kalman filtering as the initial value, and constructs an optimization problem containing a data fidelity term and a spatial smoothing regularization term: min || G * τ - d|| 2 + λ * || L * τ || 2 . Where G is a forward operator, L is a Laplace operator or a gradient operator, and λ is a regularization parameter, for example, 0.01. The inversion process is solved using the alternating direction multiplier method or the conjugate gradient method, and the output is a full-field shear stress distribution with a spatial resolution of centimeters.

[0107] Example 8: Centrifugal simulation system of scouring process in supergravity field, which is divided into three parts: dynamic phase-adjustable wave-current coupling wave-flow coupling system, multi-source data acquisition system, bed shear stress inversion system.

[0108] The dynamic phase-adjustable wave-current coupling wave-flow coupling system includes a propeller, a wedge-shaped block, a phase sensor, a transmission device, and a servo control device.

[0109] The propeller can generate a water flow of 0-3 m / s, and the wedge-shaped block can generate regular waves and irregular waves with a period of 0-10 s and a wave height of 0-10 m. The phase sensor includes a wave height phase sensor and a water flow phase sensor.

[0110] Wave phase sensor, using a miniature liquid level gauge, non-contact, avoid interference flow field, installed on the model tank test section upstream 10 cm, probe to the water surface. Water flow phase sensor, installed on the flow generating propeller motor shaft end, by detecting the speed pulse to obtain the water flow phase, the flow rate and the propeller speed is positively correlated, the speed phase is the water flow phase.

[0111] Wave phase extraction method: through the wave height-time curve output by the liquid level gauge, the peak value detection algorithm is used to identify the wave peak time t 波峰 , the trough time t 波谷 , define the wave phase θ w =360°×(t-t 波峰 ) / T w , where T w is the wave period, θ w ranges from 0 to 360°.

[0112] Water flow phase extraction method: according to the vortex propeller speed n(r / min) to calculate the water flow velocity u=k u ×n, k u is the flow rate-speed calibration coefficient, define the water flow phase θ f =360°×(t-t 流速峰值 ) / T f , where T f is the water flow pulsation period, when there is no pulsation, T f takes 10 times of the wave period to ensure the phase synchronization.

[0113] Transmission device includes harmonic reducer and gear linkage mechanism, the input end is connected with the wave making wedge block servo device, the displacement signal of the up and down movement of the wedge block is obtained, the output end is connected with the flow generating propeller speed shaft through the gear linkage mechanism, realizing the linkage of wave and flow.

[0114] The wave making wedge block is driven by a servo oil cylinder, and the flow generating propeller is driven by a permanent magnet synchronous motor.

[0115] The specific steps of the closed-loop control algorithm of real-time phase adjustment are as follows:

[0116] Phase difference calculation: real-time acquisition of θ w and θ f , calculation of Δθ 实际 =θ w -θ f , if negative, add 360° to convert to 0-360° range;

[0117] Error processing: calculate the deviation e=Δθ 目标 -Δθ 实际 , use the incremental PID control:

[0118] Δu(k)=K p [e(k)-e(k-1)]+K i e(k)+K d [e(k)-2e(k-1)+e(k-2)], where K p =0.5, K i =0.01, K d =0.1;

[0119] Feedforward compensation: based on the first derivative dt / Δθ of the preset phase difference curve. 实际 / dt, output adjustment amount in advance, u 前馈 =k f ×dt / Δθ 实际 / dt, where k f =0.2;

[0120] Execution output: Total adjustment u(k) = Δu(k) + u 前馈 This is converted into control signals for wave generation and current generation;

[0121] Wave generation: the velocity of the wedge block, u w =k w ×u(k), where k w =0.1mm / s per;

[0122] Current generation: propeller speed n f =n f0 +k f ×u(k), where k f =5r / min per°, n f0 The reference speed;

[0123] The signal flow path is: phase sensor → data acquisition card → host computer → servo controller (executes PID algorithm) → wave / current generation mechanism → phase sensor (feedback closed loop). By using a synchronized clock, the timestamp error between sensor acquisition, algorithm calculation, and actuator action is ensured to be ≤1ms, achieving wave-current synchronization.

[0124] Dynamically phase-tunable wave-current coupling mechanism:

[0125] Wave-current motion linkage is achieved through a phase sensor and a transmission device: the wave phase sensor identifies the wave crest moment in real time, and the water flow phase sensor detects the peak flow velocity synchronously. The two are mechanically linked with the transmission device through a harmonic reducer to ensure the timing of the motion is matched.

[0126] The preset phase difference curve adjusts the time difference between the wedge block and the propeller through a PID algorithm, achieving stepless adjustment from 0 to 180° with a phase error of ≤ ±3°, accurately simulating the wave and current phase relationship of the real ocean.

[0127] The multi-source data acquisition system is composed of a water dynamic sensor group, a local scour sensor group, and a bed surface observation sensor group.

[0128] The water dynamic sensor group includes:

[0129] Wave height phase sensor: One is arranged upstream of the test section (10 cm away from the hydraulic model), and the other is arranged downstream (10 cm away from the hydraulic model), 5 cm away from the water surface height, and vertically downward measurement;

[0130] Current meter: Two are arranged at the cross section of the test section, one upstream and one downstream, 5 cm away from the bed surface (wave flow coupling core area), and the probe is directed to the flow direction;

[0131] Shear stress sensor: directly pasted on the bed surface of the soil sample box, a total of 3, 6 cm in front of the hydraulic model, evenly distributed;

[0132] Pulsating pressure sensor: buried in the water flow surface of the hydraulic model (1 at the bottom, middle and top), the sensor probe is flush with the model surface to avoid interfering with the flow field.

[0133] Effective shear stress τ couple =τ wave +τ current +τ interact , wherein τ wave is the shear stress (Pa) of the wave alone, calculated from the wave height phase sensor data, τ wave =0.5×ρ×g×H² / (2πL), wherein ρ is the density of water 1000 kg / m³, g is the acceleration of gravity, H is the wave height, and L is the wavelength;

[0134] τ current is the shear stress (Pa) of the water flow alone, calculated from the current meter data, τ current =f×ρ×v² / 8, wherein f is the bed surface friction coefficient, calibrated from the baseline data of the shear stress sensor; v is the average flow velocity of the water flow.

[0135] τ interact is the shear stress (Pa) of the wave flow coupling, derived from the pulsating pressure sensor data, τ interact =k1×ΔP, wherein k1 is the interaction coefficient, with a value range of 0.05-0.1; ΔP is the pulsating pressure.

[0136] The scour sensor group includes:

[0137] Displacement sensor: erected above the visible window on the top of the model box, aligned with the center of the soil sample box test section, laser vertically irradiates the soil surface, realizes non-contact measurement of scour depth;

[0138] Pore water pressure sensor: buried in the soil body (depth 5 cm, 10 cm each 1), 10 cm away from the water model foundation edge, avoid the sensor destroy the soil structure, output bed bottom pore pressure.

[0139] Coupling effect coefficient η calculation: η = τ couple / (τ wave +τ current );

[0140] η>1: wave flow synergistic enhancement (coupling promotes scour);

[0141] η=1: no synergistic effect (wave flow independent action);

[0142] η<1: wave flow synergistic weakening (coupling inhibits scour).

[0143] Bed surface observation sensor group includes:

[0144] Near-bottom velocity field acquisition module (PIV): through the sheet laser and high-speed camera to obtain the instantaneous velocity u(z, x, t) at a distance of 0.3-2mm from the bed surface, output the near-bed flow velocity profile u(z)=Ψu / Ψz, wherein Ψ is the partial derivative and z is the height from the bed surface;

[0145] Digital image DIC system: lay flexible film on the bed surface, and obtain the bed surface micro-deformation ε(x, t) by high-speed imaging system;

[0146] The bed surface observation sensor group is integrated in a high-precision camera, and all sensors in the system are collected synchronously through a high-precision synchronous clock.

[0147] The system is externally provided with a ring-shaped water tank, a water tank support structure and transparent organic glass. The ring-shaped water tank is installed in the system main frame through the water tank support structure, and the side wall is made of transparent organic glass material, which is convenient for high-precision camera to non-contact measurement and record the wave flow coupling scouring process in the super gravity field.

[0148] Bed surface morphology measurement module: laser profilometer is used to scan and obtain the bed surface elevation z bed (x, t);

[0149] Wire processing: use electromagnetic interference shielding wire, wire diameter ≤1mm, to reduce the interference to the flow field, wiring along the inner wall of the model box, and sealing glue packaging at the outlet;

[0150] Synchronous clock synchronous acquisition is adopted, the acquisition frequency is set to 100Hz, the synchronous clock sends a unified trigger signal, and all sensors start collecting at the same time to avoid time difference caused by asynchronous acquisition.

[0151] Before the test, each sensor is calibrated in 1g environment, the calibration coefficient is recorded, and written into the acquisition system.

[0152] Start the centrifuge to the required super gravity acceleration for the test, collect the baseline data of the sensor under the condition of no wave flow loading, and ensure the accuracy of the data.

[0153] Scouring rate v erode Model associated with coupling parameters: v erode =k×τ couple α ×ΔP β ;

[0154] Wherein, v erode is the instantaneous scouring rate (cm / s), which is calculated from the displacement sensor data: v erode =(h t -h {t-1} ) / Δt, h t is the scouring depth at t, h {t-1} is the scouring depth at t-1, and Δt=1ms. K, α, β: super gravity field scale correction coefficient, calibrated by test.

[0155] Specifically, the core of quantifying the wave flow coupling effect is to convert physical formula→programming logic→real-time calculation, and the algorithm is embedded in the servo control system, which is linked with the acquisition module to realize the real-time of acquiring 1 set of data→calculating 1 result→outputting 1 quantified index.

[0156] The coupling effect quantification algorithm of the application is realized based on multi-field synchronous data acquisition, and the specific steps are as follows:

[0157] Step 1, data preprocessing:

[0158] Obtain the synchronous collected wave height H, average flow velocity v, fluctuating pressure ΔP, scouring depth h t and other data, wherein the timestamp error of wave height H and flow velocity v is ≤1ms, and the scouring depth h t is recorded continuously at a sampling interval Δt=1ms;

[0159] Step 2, calculation of shear stress under single action:

[0160] a. Wave shear stress τ wave under single action: according to the wavelength L fitted from the wave height phase sensor data, the formula τ wave =0.5×ρ×g×H² / (2πL) is used for calculation, wherein ρ is the density of water 1000kg / m³, g is the acceleration of gravity, and n×g is used instead of g under super gravity field, and n is the centrifugal coefficient;

[0161] b. Water flow shear stress τ current under single action: according to the formula τ current= f x p x v2 / 8, where f is the bed friction coefficient, calibrated by the baseline data of shear stress sensor, with a calibration error of less than 5%;

[0162] Step 3, coupling interaction shear stress calculation:

[0163] According to the ΔP (unit: kPa) collected by the pulsating pressure sensor, the formula τ interact =k1x ΔPx 1000 is used for calculation (ΔP is converted to Pa), where k1 is the coupling interaction coefficient of hypergravity field, which is obtained by fitting test data, with a fitting range of 0.05-0.1 and a fitting error of less than 3%;

[0164] Step 4, effective shear stress and coupling effect coefficient calculation:

[0165] a. Effective shear stress τ couple =τ wave +τ current +τ interact ;

[0166] b. Coupling effect coefficient η = τ couple / (τ wave +τ current ), and the abnormal value is truncated (η ∈ [0.8, 1.5]);

[0167] Step 5, establish the quantitative correlation model of effective shear stress, pulsating pressure and scouring rate:

[0168] According to the formula v erode =(h t -h {t-1} ) / Δt, the measured scouring rate is calculated, and then the correlation model v erode =kxτ couple α x ΔP β is established through the hypergravity scale correction coefficients k, α and β;

[0169] Step 6, real-time correction:

[0170] Every 1 minute, compare the measured scouring rate v erode with the predicted value v erode_pred , if the error is more than 10%, re-fit the coefficients k, α and β to ensure the calculation accuracy.

[0171] The bed shear stress inversion system is the core module of data processing, which inverts the space-time field τ(x, t) from multiple source observations d(t) and gives the uncertainty estimation to meet the real-time control requirements. The core principles of the inversion algorithm include three levels of instantaneous shear stress fast estimation module, multi-modal fusion filtering module and regularization high-resolution inversion.

[0172] The bed shear stress inversion system is connected with a multi-source data acquisition system to obtain synchronous observation data.

[0173] a. Instantaneous shear stress rapid estimation module:

[0174] Based on the combined approximation formula of velocity gradient and pressure gradient, it has low delay characteristics (1-3 ms) and is used for real-time control: τ0(x,t)=μ*Ψu / Ψz| z=0 +k p *Ψp / Ψx, wherein Ψ is a partial derivative, and μ is dynamic viscosity.

[0175] b. Multi-modal fusion filtering module:

[0176] The state vector is established: s(t)=[τ(x,t),u(z,t),z bed (x,t)], wherein τ(x,t) is the shear stress field, u(z,t) is the near-bottom velocity profile, and z bed (x,t) is the bed elevation;

[0177] The state equation is constructed: s k+1 =f(s k )+w k , wherein f(s k ) is a state prediction function, and w k represents model uncertainty;

[0178] The observation equation is constructed: d k =H k *s k +v k , wherein d k represents an observation vector, H k represents a multi-modal observation matrix, and v k represents observation noise;

[0179] The extended Kalman filter (EKF) is used for real-time fusion, and the inversion shear stress τ̂(x,t) and the covariance matrix P(x,t) are output.

[0180] c. Regularization high-resolution inversion module:

[0181] The bed evolution is constrained to solve the regularization: min||G*τ-d|| 2 +λ||L*τ|| 2 , wherein G is a forward operator matrix, d is an observation data vector, λ=0.01 is a regularization parameter, L is a regularization operator, and the full-field shear stress distribution τ(x,t) with a spatial resolution of 1 cm×1 cm is output, and the calculation delay is ≤1 s.

[0182] The specific implementation steps are as follows:

[0183] a. System alignment and time synchronization: start PPS, confirm all acquisition terminal timestamps are consistent (error <1ms).

[0184] b. Sensor zero point and gain calibration: measure zero point, noise spectrum and record in static state.

[0185] c. Membrane calibration: apply known force / displacement, establish calibration curve G of membrane displacement→ apparent τ.

[0186] d. Pore pressure→τ calibration: apply known τ with reference wall shear stress sensor in circulating water tank, record pore pressure response.

[0187] e. Velocity gradient path calibration: use PIV to measure synchronously and compare with reference shear stress, get μ*Ψu / Ψz correction coefficient, where Ψ is partial derivative.

[0188] f. Matrix construction: all observed data are summarized into H matrix.

[0189] g. Start transient shear stress fast estimation module: output τ0 for each frame, send to controller.

[0190] h. Perform multi-modal fusion filtering module in parallel: extend Kalman filter (EKF) to fuse observations at 20-200Hz, update inverse shear stress τ̂ and covariance P.

[0191] i. Periodically perform regularization high-resolution inversion module: use alternating direction multiplier method to do high-precision inversion on recent time window data, save results for analysis.

[0192] j. Return τ̂ to host computer every control cycle, phase module makes corresponding closed-loop adjustment.

[0193] To solve the problem of simulation distortion of wave-current dynamic coupling effect, a hardware system with adjustable dynamic phase is adopted to cooperate with a double-layer wave-current control strategy. Through the outer control loop, a phase difference trajectory conforming to the real tidal law is planned, and the nonlinear compensation mechanism of the inner control loop is used to overcome the hysteresis effect of sand wave topography on the flow field, realizing continuous and accurate dynamic adjustment of the wave-current phase difference in the range of 0-180°, restoring the real wave-current coupling physical process.

[0194] In view of the problems of lack of bed surface response feedback control and insufficient shear stress field inversion accuracy, a shear stress field inversion method based on wave phase harmonic expansion is adopted. The method discards the traditional discrete point measurement, uses phase synchronous observation sequence and adaptive harmonic base function, reconstructs the sparse observation into continuous shear stress field, and effectively removes the interference of bed surface morphology evolution. At the same time, the controllability analysis and nonlinear controllable domain construction of shear stress are introduced, the mapping between phase adjustment amount and shear stress response is established, so that the control system has the ability of sensing controllability, and can adaptively adjust the control strategy according to the real-time bed surface state, and the scouring process strictly develops along the preset trajectory.

[0195] In summary, the transient estimation and extended Kalman filtering fusion method provided by the present application provides high-quality initial guess value for harmonic inversion, accelerates the convergence of harmonic parameters; on the other hand, in the transient process in which the harmonic model has not been established or the wave flow phase is not clear, it is directly used as the feedback source of the control system, ensuring the continuity and safety of the whole period control.

Claims

1. A method for simulating and testing wave-current coupling scour in a hypergravity field, characterized in that, include: Read the monitoring signals from multiple sensors and the wave-generating and current-generating motion state signals under the hypergravity test environment, analyze the real-time wave phase and water flow phase, map the monitoring signals to the wave phase domain, and generate a phase synchronization observation sequence; Based on the phase-synchronous observation sequence, the fluid dynamic characteristics of the bed surface are inverted and analyzed to obtain the estimated data of shear stress field reflecting the stress state of the bed surface; Based on the deviation between the estimated shear stress field data and the preset target shear stress index, and combined with the response sensitivity characteristics of the shear stress field to changes in wave-current phase difference, wave-current phase control commands are generated for real-time adjustment of the phase difference between the wave-generating mechanism and the current-generating mechanism. Based on the phase-synchronous observation sequence, the hydrodynamic characteristics of the bed surface were inverted and analyzed, and the estimated data of the shear stress field reflecting the stress state of the bed surface were obtained, including: A pre-defined shear stress harmonic model is used to describe the variation of shear stress at the bed surface nodes with wave phase. The model uses the harmonic coefficient vector and phase parameter vector as the state variables to be solved. An optimization solution system is established with the objective of minimizing the difference between the output of the shear stress harmonic model and the phase-synchronized observation sequence; The optimal harmonic coefficient vector and phase parameter vector are calculated using the optimization solution system, and the shear stress field estimation data with unified spatiotemporal resolution are reconstructed accordingly. The pre-defined shear stress harmonic model describing the variation of shear stress at the bed surface nodes with wave phase includes: The characteristics of bed elevation change in multi-source sensor monitoring signals were analyzed to determine the bed evolution stage markers corresponding to the current monitoring area, including the flat bed stage or the sand wave development stage. When the bed surface evolution stage is marked as the flat bed stage, a symmetric harmonic basis function family composed of sine and cosine functions is selected; When the bed surface evolution stage is marked as the sand wave development stage, an asymmetric harmonic basis function family containing asymmetric waveform characteristic functions is selected; The selected family of harmonic basis functions is used to construct the functional expression of the shear stress harmonic model.

2. The method according to claim 1, characterized in that, Mapping the monitoring signals to the wave phase domain to generate a phase-synchronized observation sequence includes: Based on the real-time wave phase, the time-continuous multi-source sensor monitoring signal is divided into several complete wave cycle units; For wave periodic units, the instantaneous signal values ​​at preset discrete phase nodes are extracted to form a periodic segmented observation set; By statistically superimposing the periodic segmented observation set at the same discrete phase node, a phase-synchronized observation sequence with wave phase as the independent variable index is generated.

3. The method according to claim 1, characterized in that, An optimization solution system with the objective of minimizing the difference between the output of the shear stress harmonic model and the phase-synchronized observation sequence includes: A critical shear stress threshold is introduced to describe the initiation conditions of the sediment used in the experiment; Based on the shear stress amplitude calculated from the critical shear stress threshold and the shear stress harmonic model, a scour-induced physical constraint term is constructed. The scour-induced physical constraint term is configured to accumulate values ​​only for spatial distribution regions where the shear stress amplitude exceeds the critical shear stress threshold. The physical constraints caused by scouring are added as weighted penalty components to the objective function of the optimization solution system.

4. The method according to claim 1, characterized in that, Inverting and analyzing the fluid dynamics characteristics of the bed surface also includes calculating the coupling effect coefficients, and the specific steps include: Based on wave height data, flow velocity data and pulsating pressure data from multi-source sensor monitoring signals, the shear stress under the action of waves alone, the shear stress under the action of water flow alone, and the interactive shear stress of wave-flow coupling are calculated respectively. The effective shear stress is obtained by superimposing the shear stress caused by wave action alone, the shear stress caused by water flow alone, and the wave-flow coupled interactive shear stress. Calculate the ratio of the effective shear stress to the sum of the shear stress caused by the wave alone and the shear stress caused by the water flow alone, and generate a coupling effect coefficient for calculating the degree of wave-flow synergistic enhancement or attenuation.

5. The method according to claim 1, characterized in that, By combining the sensitivity characteristics of the shear stress field to changes in wave-current phase difference, shear stress controllability analysis data are generated. The specific steps include: The key shear stress index sequence, which is the control object, is extracted from the shear stress field estimation data, and the wave-current phase difference sequence is obtained by calculating the difference between the real-time wave phase and the real-time water flow phase. The sliding time window technique was used to fit the local relationship between the key shear stress index sequence and the wave-current phase difference sequence, and the shear stress sensitivity index reflecting the shear stress response intensity caused by the unit phase difference change was calculated. By combining the numerical range of the target shear stress index with the shear stress sensitivity index, the effective adjustment range of the wave-current phase difference that can keep the shear stress within the preset safe range is determined, and shear stress controllability analysis data is generated.

6. The method according to claim 5, characterized in that, The calculated shear stress sensitivity indices, which reflect the shear stress response intensity caused by a unit phase difference change, include: Based on the time scale of a single wave cycle preset in the multi-source sensor monitoring signal, the rapid response relationship between the key shear stress index sequence and the wave-current phase difference sequence is fitted to obtain the short-term instantaneous sensitivity. Based on the time scale of the bed surface morphology evolution preset in the phase synchronous observation sequence, the long-term cumulative trend of the key shear stress index sequence with the change of wave-current phase difference sequence is fitted to obtain the mid-term evolution trend parameters. Based on short-term instantaneous sensitivity and medium-term evolution trend parameters, the possible wave-current phase difference range is scanned and predicted at the current time section to determine the set of values ​​that can make the predicted shear stress value fall within the target range, thus forming a nonlinear wave-current phase difference controllable domain.

7. The method according to claim 6, characterized in that, The wave-current phase control command generated for real-time adjustment of the phase difference between the wave-generating mechanism and the current-generating mechanism adopts a two-layer control structure, including: In the outer control loop, based on the deviation between the current shear stress field estimation data and the target shear stress index, and with the shear stress controllability analysis data and the nonlinear wave-current phase difference controllable domain as constraints, the target wave-current phase difference trajectory in the next control cycle is calculated. In the inner control loop, the difference between the real-time wave phase and the real-time water flow phase is calculated to obtain the actual phase difference. Based on the tracking error of the target wave-flow phase difference trajectory according to the actual phase difference, wave-flow phase control commands are generated to drive the wave-generating mechanism and the current-generating mechanism to perform phase adjustment actions.

8. The method according to claim 7, characterized in that, The inner control loop is equipped with a controller structure switching mechanism based on the physical state of the bed surface. The specific steps include: The spatial spectral characteristics and morphological gradients in the bed surface morphology estimation data are analyzed to determine the current bed surface evolution stage markers, including the flat bed stage and the sand wave development stage. When the bed evolution stage is marked as the flat bed stage, the linear phase tracking control structure is activated, and the wave-current phase control command is calculated using a linear feedback algorithm to achieve rapid tracking of the target wave-current phase difference trajectory. When the bed surface evolution stage is marked as the sand wave development stage, the nonlinear compensation control structure is activated. On the basis of the linear feedback algorithm, the nonlinear compensation term is superimposed to offset the asymmetric shear stress response lag caused by the sand wave terrain and generate wave flow phase control command.

Images