Method and Detection Device for Collaborative Control of Ship Passage Safety in Large Water-Saving Ship Locks

Abstract

This invention discloses a collaborative control method and detection device for safe ship passage through large water-saving locks. The method includes: deriving hydrodynamic spectrum risks based on valve opening plans to obtain local flow velocity prediction, frequency band energy ratio, and a basic constraint table; constructing a joint criterion based on mooring tension signal analysis (such as fractional derivative and wavelet sparsity), and combining it with local flow velocity prediction to identify transient events and quantify and generate feedback constraint factors; coupling the basic constraint table and the feedback constraint factors to synthesize a final constraint table; and fusing the final constraint table, frequency band energy ratio, and valve opening plans (e.g., through S-shaped Jerk and notch filters) to generate an execution trajectory. This invention achieves closed-loop collaborative suppression of dual risks, improving the safety performance of ship passage through locks under complex hydraulic conditions.

Description

Technical Field

[0001] This invention relates to the field of ship lock control methods, and in particular to a ship lock safety collaborative control method and detection device for large water-saving locks. Background Technology

[0002] Shipping is a vital pillar of the national economy. With the construction of large-scale canal projects, new and complex navigation structures, such as high-head, multi-stage, strong-tidal, and water-saving locks, have begun operation. The operation and control of large water-saving locks, especially the complex hydraulic characteristics during their filling and emptying processes, pose unprecedented challenges to the safety of ships passing through the locks (particularly mooring safety). Therefore, researching collaborative control methods for ship passage safety adapted to the operational characteristics of these new locks is of great significance for ensuring the safety of transportation hubs, improving navigation efficiency, and serving major national strategies.

[0003] Currently, the safety control of conventional ship locks mainly relies on offline-defined valve operation curves and on-site monitoring and alarm systems. Valve operation curves are typically based on simplified hydraulic or physical model tests, pre-set to a fixed slow-fast-slow or phased opening pattern to mitigate water flow fluctuations in the lock chamber under normal operating conditions. On-site safety monitoring often employs video surveillance, laser or millimeter-wave radar to monitor the vessel's profile and position, or force sensors installed on mooring hooks to trigger audible and visual alarms when the mooring force exceeds a fixed, preset safety threshold. At the control execution level, the system primarily executes preset opening plans, lacking the ability to online perceive and adjust for changes in operating conditions in real time.

[0004] Existing technologies for dealing with complex hydraulic disturbances in large, water-saving ship locks suffer from feedforward blindness in control strategies and feedback lag in monitoring systems, with a lack of effective coordination mechanisms between the two. Fixed valve opening plans are open-loop controls that do not consider whether the plan itself will excite predetermined resonant frequencies in the lock chamber-ship system. When a frequency component in the opening plan (e.g., 0.5-2Hz) couples with the inherent frequency of the mooring system, it can trigger ship resonance even with a small total flow rate, and existing systems lack the ability to predict and mitigate such spectral risks online. Mooring force alarms based on fixed thresholds are reactive, only responding when the mooring force has reached a dangerous level. Existing technologies fail to detect the precursors of micro-transient events, from cable slack to instantaneous tension, missing the optimal time for intervention before the peak force arrives. Summary of the Invention

[0005] The purpose of this invention is to provide a method and detection device for safe and coordinated control of ship passage through large water-saving locks, which solves the problem of difficulty in coordinating the control of feedforward spectrum risk of valve planning and feedback transient risk of mooring emergencies.

[0006] According to one aspect of this application, a collaborative control method for safe passage of ships through a large water-saving ship lock includes:

[0007] Based on the valve opening plan, hydrodynamic spectrum risk is extrapolated to obtain local flow velocity prediction, frequency band energy ratio and basic constraint table;

[0008] Based on mooring tension and local flow velocity prediction, transient events are identified and feedback constraint factors are quantified and generated.

[0009] The coupling base constraint table and feedback constraint factor are used to synthesize the final constraint table, and the final constraint table, frequency band energy ratio and valve opening plan are integrated to generate the execution trajectory.

[0010] Optionally, based on mooring tension and local flow velocity prediction, transient events are identified and feedback constraint factors are quantified, including:

[0011] Analyze mooring tension signals to construct a joint criterion;

[0012] Identifying transient events based on triggering conditions using joint criteria;

[0013] In response to transient events, the upper limit of short-term forces is deduced by combining local flow velocity prediction;

[0014] By comparing the upper limit of short-term stress with the upper limit of allowable tension, the feedback limiting factor is obtained through quantification.

[0015] Optionally, the mooring tension signal is analyzed to construct a joint criterion, including:

[0016] Calculate the fractional derivative of the mooring tension signal;

[0017] Evaluate the wavelet sparsity of mooring tension signals;

[0018] We construct a joint criterion by weighting the fractional derivative and wavelet sparsity.

[0019] Optionally, transient events are identified based on triggering conditions according to joint criteria, including:

[0020] Set an entry threshold and an exit threshold, where the entry threshold is higher than the exit threshold to form a hysteresis interval;

[0021] The joint criteria and entry threshold will be continuously monitored and compared.

[0022] Transient events are confirmed when the joint criterion is not less than the entry threshold and the duration of the state meets the minimum dwell time.

[0023] Optionally, in response to transient events, the upper bound of short-term forces is deduced by combining local flow velocity predictions, and the feedback constraint factor is quantified, including:

[0024] In response to identified transient events, update the equivalent mechanical parameters of the mooring system;

[0025] Based on the updated equivalent mechanical parameters and local velocity prediction, a hydrodynamic equivalent model is constructed and solved to deduce the upper limit of short-term forces.

[0026] The feedback limiting factor is determined based on the comparison between the upper limit of short-term force and the preset upper limit of allowable tension.

[0027] Optionally, based on the valve opening plan, hydrodynamic spectrum risk is extrapolated to obtain local velocity predictions, frequency band energy ratios, and basic constraint tables, including:

[0028] A feedforward kernel is applied to valve opening planning to obtain local flow velocity predictions;

[0029] Short-time spectral estimation is performed on local velocity predictions to determine the band energy proportions;

[0030] The frequency band energy ratio is converted into a basic bounding table using a closed mapping function.

[0031] Optionally, the feedforward kernel is obtained through an online identification method;

[0032] Online identification methods include: acquiring historical execution trajectories and corresponding local flow velocities;

[0033] A recursive least squares algorithm with a forgetting factor is applied to obtain a feedforward kernel based on historical execution trajectories and local flow velocity identification.

[0034] Optionally, the basic conformal table includes at least: the maximum Jerk limit; the maximum acceleration limit; the maximum velocity limit; and notch filter parameters.

[0035] Optionally, determining the notch filter parameters includes:

[0036] Short-time spectral estimation is performed on local velocity predictions to extract spectral peaks in hazardous frequency bands;

[0037] Determine the center frequency of the notch filter parameters based on the spectral peaks of the dangerous frequency band;

[0038] Based on the frequency band energy ratio, the bandwidth and depth of the notch filter parameters are dynamically set.

[0039] Optionally, the basic constraint table is coupled with the feedback constraint factor to synthesize the final constraint table, including:

[0040] The feedback limiting factor is applied as a multiplicative strengthening factor;

[0041] Apply a multiplicative strengthening factor to at least one constraint in the basic constraint table to dynamically tighten the constraint of that constraint; synthesize the final constraint table.

[0042] Optionally, the final form constraint table, frequency band energy ratio, and valve opening plan are integrated to generate an execution trajectory, including:

[0043] The notch filter parameters included in the final form control table are used to perform spectral shaping on the valve opening plan to obtain a spectral shaping reference.

[0044] Furthermore, a three-segment S-shaped Jerk trajectory generator is applied to generate an execution trajectory based on a spectral shaping reference, under the constraints of Jerk, acceleration, and velocity limits defined in the final shape table.

[0045] Optionally, the process of generating an execution trajectory by integrating the final form constraint table, frequency band energy ratio, and valve opening plan also includes:

[0046] Enable a four-state machine with normal, slow start, pause and emergency stop;

[0047] The four-state machine performs unidirectional main loop transitions between the four states based on the threshold conditions of the frequency band energy ratio and the threshold conditions of the feedback form factor, in order to output the current state.

[0048] The final generation of the execution trajectory is controlled by the current state.

[0049] Optionally, the final generation of the execution trajectory is controlled by the current state, including:

[0050] Configure parameter remapping ratio tables for each of the four states;

[0051] Select the corresponding parameter remapping ratio table based on the current state;

[0052] Apply the corresponding parameter remapping ratio table to remap the Jerk, acceleration, and velocity limits in the final form constraint table;

[0053] The execution trajectory is then reshaped and trimmed based on the remapped limits.

[0054] Optionally, the coupling of the basic constraint table and the feedback constraint factor to synthesize the final constraint table further includes:

[0055] Obtain external robust time window constraints and actuator physical upper limits; when synthesizing the final constraint table, perform point-by-point value fusion on the constraints defined by the base constraint table, feedback constraint factor, robust time window constraints and actuator physical upper limits.

[0056] Optionally, the method further includes: acquiring tidal bore monitoring data through a tidal bore monitoring unit; predicting the arrival time and risk level of the tidal bore based on the tidal bore monitoring data; and generating robust time window constraints based on the prediction results.

[0057] Optionally, the method further includes:

[0058] Perform a closed-loop update step, based on the execution trajectory and historical measurement data, to generate updated feedforward kernel parameters and updated trigger threshold parameters;

[0059] Among them, the updated feedforward kernel parameters are used in the step of extrapolating hydrodynamic spectrum risks based on valve opening plans;

[0060] The updated trigger threshold parameter is used to identify transient events and quantify the generation of feedback constraint factors based on mooring tension and local flow velocity prediction.

[0061] According to another aspect of this application, a collaborative detection device for ship passage safety in a large water-saving ship lock includes:

[0062] The global state awareness module is configured to provide the real-time data required for decision-making in the operational security hub.

[0063] The operation safety hub is configured to receive and process data from the global status perception module, and it has a built-in expert knowledge base for the safe operation of the water-saving ship lock and a multi-objective optimization algorithm.

[0064] The collaborative safety control module is configured to output control signals to adjust the water-saving process or perform coordination between hubs based on the instructions of the operating safety hub.

[0065] Both the global state perception module and the collaborative safety control module are connected to the operation safety center for data and control signals.

[0066] Beneficial effects: Through the above technical solutions, this invention solves the problem of the difficulty in coordinating the control of feedforward spectrum risk of valve planning and feedback transient risk of mooring emergencies, realizes closed-loop coordinated suppression of dual risks, and improves the safety of ships passing through locks under complex hydraulic conditions. Attached Figure Description

[0067] Figure 1 This is a schematic diagram of a collaborative control method for safe passage of ships through large water-saving locks.

[0068] Figure 2 This is a flowchart illustrating the process of identifying transient events and quantifying feedback constraint factors based on mooring tension and local flow velocity prediction.

[0069] Figure 3 This is a flowchart illustrating the process of constructing a joint criterion by analyzing mooring tension signals.

[0070] Figure 4 This is a flowchart illustrating the process of identifying transient events based on joint criteria trigger conditions. Detailed Implementation

[0071] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0072] It should be noted that the terms "first," "second," etc., used in the specification and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein.

[0073] It should be noted that, for the purpose of clearly demonstrating the steps of this application, each step has been numbered in the specification. These numbers are for ease of explanation only and do not limit the execution order of the steps. In actual operation, depending on the technical requirements of the specific implementation scenario, the steps may be executed in a different order than that shown in the specification, and in some cases, parallel processing between steps can also be achieved.

[0074] Example 1: A collaborative detection device for ship passage safety in large water-saving locks is provided. This device can be deployed in large water-saving lock environments (such as multi-level canal hubs). Preferably, the device includes:

[0075] The system comprises a global state awareness module, an operational safety center, and a collaborative safety control module. Both the global state awareness module and the collaborative safety control module connect to the operational safety center for data and control signals.

[0076] Specifically, the global state awareness module provides the operational security hub with the real-time data needed for decision-making. This module includes:

[0077] Ship position, speed, and berthing attitude sensing units, such as those deployed in pilotways, lockheads, and lock chambers, acquire the ship's real-time position, speed, heading, and berthing attitude through the fusion of AIS, lidar, or millimeter-wave radar.

[0078] The water-saving process sensing unit is a key component adapted to large water-saving ship locks (distinct from conventional ship locks), and it includes:

[0079] The water-saving tank valve status sensor is used to detect the opening and closing status and real-time opening degree of the water-saving tank's inlet and outlet valves. plan (t).

[0080] The water flow disturbance monitoring unit for the provincial water reservoir, for example, uses flow velocity and direction sensors (ADCP or electromagnetic velocity meters) deployed at the entrance of the provincial water reservoir connecting corridor and the gate chamber wall to monitor the disturbance intensity of the water in the gate chamber during the filling and emptying of the provincial water reservoir in real time, and obtain historical flow velocity data. raw (t).

[0081] The tidal bore monitoring unit, targeting downstream hubs of canals, uses tidal level stations and radar current meters at the downstream estuary to monitor the generation, intensity, and estimated arrival time of tidal bores in real time.

[0082] The mooring force sensing unit is a multi-dimensional force sensor integrated on the mooring hook. It is used to monitor the tension and rate of change of the mooring line in three or more dimensions (X, Y, Z) in real time and obtain the mooring tension F(t).

[0083] The gate body status sensing unit is used to monitor the closing alignment error and vibration status of the miter gate and the tide barrier gate.

[0084] The operational safety central module, serving as the decision-making center (brain) of the device, receives and processes data from the overall state perception module. It incorporates a water-saving lock safety operation expert knowledge base and multi-objective optimization algorithms (i.e., the SRM-CF (Spectrum Risk Closed-Loop Deduction and Basic Boundary Mapping) and ET-OPU (Transient Event Triggering and Parameter Upper Bound Deduction) algorithms detailed in subsequent embodiments). This is used for data based on the water-saving process perception unit (e.g., valve opening plan u). plan (t) and historical flow rate raw (t)), and the data from the mooring force sensing unit (mooring tension F(t)), to execute a cooperative control method.

[0085] Specifically, the operational safety center is used to simulate the mooring safety status of ships during the water-saving process (e.g., to simulate the frequency band energy ratio η). _band (t) and feedback limiting factor κ _fb (t)).

[0086] Preferably, the operational safety central module is also used to predict risks based on data from the tidal surge monitoring unit and generate collaborative constraint instructions (e.g., robust time window constraint schedule). _θ (t)).

[0087] The collaborative safety control module, acting as the device's execution end (hands and feet), outputs control signals to adjust the water-saving process or perform coordination between key components based on instructions from the operational safety center. Specifically, it includes:

[0088] The water-saving tank operation process safety interlock unit receives decision results from the operation safety center (e.g., the final limit table). _table_finaland state(t)). When the deduction result (such as κ) _fb When (t) indicates a risk of mooring instability, actively output a control signal (e.g., control command ctrl) to the water-saving tank control system (e.g., PLC). _cmd (t) to automatically pause, slow down or adjust the operation of the water-saving pool valves until the ship's condition is restored to safety.

[0089] The cross-hub safety interlocking interface is used to achieve safety coordination between multiple levels of hubs. For example, when the operation safety center predicts a risk in a downstream hub (such as the Youth Hub) based on tidal monitoring data, it sends a coordinated stop-ship scheduling signal to the mid- and upstream hubs (such as the Madao and Qishi Hubs) through this interface to prevent vessel backlog.

[0090] The intelligent warning and guidance unit is used to send specific hazard avoidance instructions to the vessel via VHF broadcasts, shipborne intelligent terminal APPs, and other means.

[0091] The device in this embodiment combines the sensing of the water-saving process, the dynamic extrapolation of mooring force, and the real-time interlocking of valve control to achieve proactive, closed-loop collaborative control of the unique risks of large water-saving locks.

[0092] Example 2: A method for coordinated control of ship passage safety in large water-saving ship locks is provided. This method can be applied to a coordinated detection device for ship passage safety in large water-saving ship locks. Figure 1 As shown, it includes:

[0093] Step one involves extrapolating the hydrodynamic spectrum risk based on the valve opening plan, resulting in local velocity predictions, frequency band energy ratios, and a basic constraint table. The basic constraint table is a set of constraints used to quantify feedforward risk; its detailed implementation will be discussed in Example three.

[0094] Step two involves identifying transient events and quantifying them to generate feedback constraint factors based on mooring tension and local velocity prediction. Alternatively, the process can also involve detecting transient events and generating feedback constraint factors based on mooring tension, local velocity prediction, and trigger threshold parameters. The feedback constraint factor is a scalar used to quantify feedback risk. Its detailed implementation will be discussed in Example four.

[0095] Step 3: Combine the coupling base constraint table and the feedback constraint factor to synthesize the final constraint table, and integrate the final constraint table, frequency band energy ratio and valve opening plan to generate the execution trajectory.

[0096] The execution trajectory is the control instruction that is finally issued to the actuator, and its detailed implementation will be discussed in Examples 5 to 7.

[0097] Specifically, before the method in this embodiment begins, a data preparation step is included, which involves obtaining input data from the global state perception module, such as obtaining the valve opening plan u. plan (t), Multi-source flow velocity / water level measurement (historical flow velocity) raw (t) and mooring tension F(t). After acquiring the data, it is preferably time-aligned and resampled according to the system clock (e.g., 10ms or 100ms period) to obtain aligned data packets for use in subsequent steps.

[0098] In a preferred embodiment, this method further includes a closed-loop update step, which generates and distributes the execution trajectory u. _traj Executed after (t). Utilizing u _traj (t) and the flow rate / water level at the field extraction site (historical flow rate) raw Historical data such as F(t) and mooring tension F(t) are used to adaptively update the model parameters (such as feedforward kernel) and threshold parameters (such as trigger threshold) used in steps one and two online.

[0099] In this way, the method of the present invention forms a strong coupling between feedforward risk (step one) and feedback risk (step two) (step three), and achieves long-term adaptive capability to changes in the environment and model through closed-loop updates.

[0100] Example 3 details the process of extrapolating hydrodynamic spectrum risks based on valve opening plans to obtain local velocity predictions, frequency band energy ratios, and basic constraint tables. It includes the following steps:

[0101] Step 3.1: Apply the feedforward kernel to the valve opening plan to obtain the local flow velocity prediction.

[0102] Feedforward kernel, also known as impulse response kernel h r This metric physically characterizes the hydrodynamic transfer function between changes in valve opening and the local velocity response within the gate chamber. It is a dynamic model used to predict the planned valve opening, u. plan (t) will cause a local flow velocity v in the future hat (t).

[0103] Specifically, to obtain v hat The process of (t) is a convolution calculation: v hat (t)=h r *u plan (t), i.e., v hat The value at time t is equal to h. r The sequence and u plan The convolution sum of the historical sequence. For example, if h r For a finite impulse response (FIR) kernel of length L, h r= [h(0),h(1),...,h(L-1)], then v hat (t)=Σ i＝0 L-1 h(i)*u plan (ti).

[0104] In this application, the feedforward core h r These are not fixed parameters preset offline, but rather obtained through online identification methods.

[0105] Specifically, the online identification method includes: acquiring historical execution trajectories and corresponding local flow velocities; and applying a recursive least squares algorithm with a forgetting factor to identify a feedforward kernel based on the historical execution trajectories and local flow velocities.

[0106] Historical execution trajectory refers to the sequence of valve openings actually issued to the actuator within the previous control cycle or an earlier cycle. _traj_hist (t). Local flow velocity refers to the flow velocity v actually measured by the water-saving process sensing unit within the corresponding time period. _meas_hist (t).

[0107] Recursive Least Squares (RLS) with a forgetting factor is a classic adaptive filtering algorithm that minimizes the weighted sum of squared historical prediction errors: minΣ k λ t-k *(v _meas_hist (k)-h r T *U(k)) 2 Where λ is the forgetting factor, typically ranging from 0.95 to 1.0; the smaller λ is, the greater the weight given to new data, and the stronger the adaptability; U(k) is determined by the historical opening u. _traj_hist The vector formed. The RLS algorithm updates the covariance matrix and kernel vector h recursively. r , achieved h r The data is updated online on a sample-by-sample basis.

[0108] In a preferred embodiment, in order to improve the feedforward core h r To improve the accuracy and time-varying adaptability of identification, the online identification process may also include:

[0109] To achieve adaptive forgetting factor, specifically, the forgetting factor λ is not fixed but is adjusted based on the historical predicted residuals (v). _meas_hist With v hat The variance of the residuals is adaptively adjusted. For example, when the residual variance increases (indicating model mismatch), λ is decreased to strengthen the weights of new samples; when the residuals decrease, λ is increased to maintain model stability.

[0110] Regularization and selection are performed; specifically, an L2 regularization term β*||h is introduced into the RLS algorithm. r || 2 To prevent h r Excessively large parameters can lead to overfitting. The regularization coefficient β can be adaptively adjusted based on the signal-to-noise ratio (SNR) of the measured signal (e.g., a larger β for higher noise). Meanwhile, the kernel length L (e.g., between 3 and 5 taps) can be determined through cross-validation (selecting the L that minimizes the prediction error).

[0111] For nuclear bank management, specifically, due to the significant differences in the hydrodynamic characteristics of valves across different opening ranges (e.g., small opening, medium opening), this invention preferably maintains an independent kernel parameter set for each opening range r. _bank [r]. During identification, the corresponding h is loaded or updated based on the current opening interval r. r This enables segmented identification.

[0112] Model drift detection is performed, specifically by continuously monitoring h. r The vector's angle change or the proportion of the predicted residual variance. If the change exceeds a preset threshold (indicating a sudden change in hydrodynamic characteristics or model drift), a soft reset is performed (e.g., immediately reducing λ and increasing β), and a rollback to the kernel is possible. _bank The system stores the previous health snapshot to ensure the reliability of the prediction.

[0113] Step 3.2: Perform short-time spectral estimation on the local velocity prediction to determine the frequency band energy ratio.

[0114] In step 3.1, the local flow velocity v is obtained. hat After (t), a short-time spectral estimation, such as a short-time Fourier transform (STFT), is performed on it to analyze the local flow velocity v. hat (t) is the joint distribution in the time and frequency domains.

[0115] Bandwidth energy ratio η _band η(t) refers to the proportion of spectral energy within a predetermined hazardous frequency band (e.g., 0.2Hz to 2Hz, which typically coincides with the natural frequency of the ship's mooring system or the dominant frequency of swells) to the total frequency band energy (e.g., 0 to the Nyquist frequency). An example of the calculation formula is: η _band (t)=[∫ f1 f2 |V hat (f,t)∣ 2 df] / [∫0 f_Ny |V hat (f,t)∣ 2 df]. Wherein, V hat (f,t) is v hatThe STFT results for (t) show that [f1,f2] is the danger band, and f _Ny It is the Nyquist frequency.

[0116] η _band (t) is a scalar that varies with time, quantifying the valve opening plan u. plan (t) represents the relative intensity of the velocity disturbance component that will be most detrimental to ship mooring safety in the future.

[0117] Step 3.3: Convert the frequency band energy ratio into a basic form table using a closed mapping function.

[0118] Basic limit table limit _table_base This is the first constraint for achieving proactive safety control, determined by the frequency band energy ratio (feedforward risk) η. _band (t) is generated directly. A closed-form mapping function is a pre-defined, monotonic, nonlinear function used to map η... _band The risk (between 0 and 1) of (t) is mapped to specific physical constraint values. The basic constraint table includes at least: the maximum Jerk limit, the maximum acceleration limit, the maximum velocity limit, and the notch filter parameters.

[0119] Specifically, the mapping (transformation) process includes, for example:

[0120] Jerk / Acceleration / Velocity Limit Mapping: j _max_base (t)=j _nom / (1+ρ*η _band (t));

[0121] u _dot_max_base (t)=u _dot_nom / (1+γ*η _band (t));

[0122] Among them, the maximum acceleration limit a _ddot_max_base can be generated by j _max_base and u _dot_max_base Cooperative settings (e.g., a) _ddot_max_base =min(a _ddot_nom ,sqrt(2*j _max_base *u _dot_max_base This ensures the solvability of the three S-shaped trajectories.

[0123] Where, j _nom ,u _dot_nom ,a _ddot_nom These are the valve's nominal (maximum) Jerk, velocity, and acceleration; ρ and γ are calibrable mapping coefficients. This mapping relationship guarantees the current feed risk η. _bandWhen (t) increases, the valve’s maximum permissible speed, acceleration and Jerk are tightened non-linearly and smoothly (i.e. the denominator becomes larger and the limit becomes smaller), actively suppressing risk sources.

[0124] Determining the parameters of the notch filter:

[0125] Short-time spectral estimation is performed on the local velocity prediction to extract the spectral peaks of the danger band. That is, in the short-time Fourier transform (STFT) analysis in step 3.2, the strongest spectral peak f0(t) in the danger band [f1,f2] is found.

[0126] The center frequency of the notch filter parameters is determined based on the spectral peaks in the dangerous frequency band. That is, the center frequency f0 of the notch filter is set to be equal to the extracted spectral peak f0(t).

[0127] Based on the frequency band energy ratio, the bandwidth and depth of the notch filter parameters are dynamically set. That is, bandwidth df = df0 + df1 * η _band (t); Depth = d0 + d1 * η _band (t). Where df0, df1, d0, d1 are calibrable constants. When η _band As the risk (t) increases, the bandwidth and depth of the notch filter also increase to more accurately filter out the frequency components of the opening plan that will excite the water flow disturbance at f0(t).

[0128] In a preferred embodiment, to enhance the robustness of the mapping, the following steps are also performed:

[0129] Perform a conservative mapping, specifically, if the predicted residuals identified or output in step 3.1 are... _id (t) (i.e., v) _meas_hist With v hat If the difference between the two values ​​increases significantly (indicating model inaccuracy), then the mapping coefficients (γ, ρ) are temporarily increased to improve the limit of the basis form table. _table_base The output is more conservative (i.e., the limit is lower).

[0130] Perform parameter smoothing, specifically, smooth the generated notch filter parameters. _param Perform exponential smoothing (e.g., notch) _param (t)=α*notch _param (t-1)+(1-α)*notch _paramraw This avoids drastic changes in the center frequency or bandwidth of the notch filter causing impact on the system.

[0131] Perform snapshots and backtracking; specifically, use the limit table generated in this cycle. _table_baseThe relevant mapping coefficients are stored as a health snapshot. If the subsequent trajectory generation steps fail, or if step 3.1 determines that the model has drifted, the system can backtrack to the basic form table generated by the most recent health snapshot to ensure safety and redundancy.

[0132] Example 4 describes in detail the implementation process of identifying transient events and quantifying feedback constraint factors based on mooring tension and local flow velocity prediction, which is used to capture sudden changes in mooring tension caused by factors not covered by the model (such as sudden surges and the ship's own motion).

[0133] In a preferred embodiment, to reduce high-frequency noise interference and computational load, this step is preferably activated and executed only within the micro-charge / bleed window, where the micro-charge / bleed window flag is used. _micro The valve opening (e.g., <5%) and the gate chamber flow rate (e.g., <10m³) can be compared. 3 The threshold is determined by a combination of thresholds ( / s). Outside this window, transient detection is suppressed.

[0134] like Figure 2 As shown, the specific implementation of this step includes:

[0135] Step 4.1: Analyze the mooring tension signal to construct a joint criterion. Extract the precursor features of the slack-tight micro-transient event from the mooring tension F(t) against a strong noise background. Specifically, analyze the mooring tension signal to construct a joint criterion, such as... Figure 3 As shown, the process includes: calculating the fractional derivative of the mooring tension signal; evaluating the wavelet sparsity of the mooring tension signal; and weighting the fractional derivative and wavelet sparsity to construct a joint criterion.

[0136] The raw F(t) signal obtained from the mooring force sensing unit is detrended (e.g., by subtracting the moving average) and bandpass filtered (e.g., 1-25 Hz, to focus on millisecond-level transients).

[0137] Calculate the fractional derivative D α F(t), specifically, the fractional derivative order α (between 0 and 1) is more sensitive to abrupt changes (peaks) in the signal than the integer derivative. It is calculated using the Grinwald-Letnikov (GL) formula: D α F(t _k )=Σ i＝0 M [(-1) i *C(α,i)*F _pre (t k-i )]. Among them, F _preThe signal is the preprocessed tension signal; M is the cut-off length (number of taps); C(α,i) is the fractional combination coefficient (Γ(α+1) / (Γ(i+1)Γ(α-i+1))); α is the fractional order. Preferably, the order α (e.g., 0.3-0.7) can be adaptively adjusted according to the signal-to-noise ratio (SNR) of the tension signal: a larger α is used when the SNR is high to enhance sensitivity; a smaller α is used when the SNR is low to improve robustness.

[0138] Evaluation of wavelet sparsity S _w (t), specifically, the loose-tight event in the wavelet domain manifests as a sparse abrupt change in coefficients. For F _pre (t) Perform discrete wavelet decomposition (e.g., using db4 or coif3 wavelets), calculate the sparsity of the absolute values ​​of the wavelet coefficients at the millisecond-level scale of interest, and perform time-series sliding window aggregation to obtain S. _w (t).

[0139] We perform a weighted combination, specifically, we combine the two above with weights to obtain the joint criterion J(t):

[0140] J(t) = k1*∣D α F(t)∣+k2*S _w J(t) is a transient risk indicator that combines the amplitude and sparsity of signal abrupt changes.

[0141] Step 4.2: Identify transient events based on triggering conditions using joint criteria. For example... Figure 4 As shown, specifically, it includes: setting an entry threshold and an exit threshold, wherein the entry threshold is higher than the exit threshold to form a hysteresis interval; continuously monitoring and comparing the joint criterion and the entry threshold; and confirming the identification of a transient event when the joint criterion is not less than the entry threshold and the duration of the state meets the minimum dwell time.

[0142] Set parameters, specifically, set the entry threshold J. _hi Exit threshold J _lo (J) _hi >J _lo ), and minimum stay time T _min (e.g., 50ms). Hysteresis interval (J) _lo To J _hi This is used to prevent false triggering caused by J(t) jittering near the threshold edge.

[0143] Triggering logic:

[0144] Enter: When J(t)≧J _hi State duration T _duration ≧T _min When a transient event is confirmed to have been triggered, the event flag is set. _flag=1.

[0145] Exit: When the event _flag After =1, if J(t)≦J _lo If the event is resolved, the event is set. _flag =0.

[0146] As a preferred implementation method, to improve the reliability of identification, the following can also be added:

[0147] Multi-point consensus voting is performed. If the mooring force sensing unit contains sensors with multiple mooring hooks, then J is calculated for each sensor. _i (t), generates global event flags using a majority vote (e.g., 2 / 3 or 3 / 4). _flag Eliminate FSR (Fail-Safe Recovery) or noise from individual sensors.

[0148] Implement anti-chain jump logic; specifically, set the generated event flag. _flag Applying debouncing delay and minimum event interval T _gap This prevents short-term noise from causing continuous jumps in event flags.

[0149] Step 4.3: In response to transient events, combine local flow velocity prediction to deduce the short-term upper limit of force, compare the short-term upper limit of force with the preset upper limit of allowable tension, and quantify to obtain the feedback limiting factor.

[0150] This step is the coupling point between transient events (feedback) and velocity prediction (feedforward). Specifically, in response to transient events, the short-term upper bound of forces is derived by combining local velocity prediction, and the feedback constraint factor is quantified, including:

[0151] In response to the identified transient events, the equivalent mechanical parameters of the mooring system are updated.

[0152] Based on the updated equivalent mechanical parameters and local velocity predictions, a hydrodynamic equivalent model is constructed and solved to deduce the upper limit of short-term forces.

[0153] The feedback limiting factor is determined based on the comparison between the upper limit of short-term force and the preset upper limit of allowable tension.

[0154] Update the equivalent mechanical parameters; specifically, when step 4.2 identifies the event... _flag When the value is 1, it means that the mooring system may have changed from relaxed to taut, and its mechanical properties (such as stiffness and damping) have changed. The initial equivalent mechanical parameters of the mooring system are then expressed as mech. _param ={k _line ,c _line Updated to mech _param_plusWhere k is stiffness and c is damping. line represents the mooring line.

[0155] For example, the update law can be related to the peak value J of J(t). _peak hook up:

[0156] Updated mooring equivalent stiffness k _line_plus =k _line *(1+μ _k *J _norm );

[0157] Updated mooring equivalent damping c _line_plus =c _line *(1+μ _c *J _norm );

[0158] Among them, J _norm It is J _peak The normalized value within the hysteresis interval, μ _k and μ _c It is an update gain.

[0159] To extrapolate the upper bound of short-term forces, specifically, a simplified hydrodynamic equivalent model is constructed, for example:

[0160] m*x _ddot +c _line_plus *x _dot +k _line_plus *x=F _hydro_hat (t); where m is the mass of the ship (including additional water); x is the ship's displacement, dot represents the first derivative, and ddot represents the second derivative; F _hydro_hat (t) represents hydrodynamics, which is based on the prediction of local velocity v. hat (t) is used for modeling (e.g., F) _hydro_hat (t)=0.5*ρ0*C _d *A*∣v hat |*v hat +m _a *dv hat Solving this differential equation (e.g., using a linearized envelope, the Chebyshev upper bound method, etc.) allows us to deduce the future short-term prediction window (e.g., ΔT = 5 seconds) from the feedforward v. hat (t) and feedback mech _param_plus Under the combined action, the mooring tension k _line_plus The upper bound F that *x will reach hat_plus (t), and extract its peak value F. _peak_plus ρ0 is the fluid density, and A is the flow area. dv hat / dt represents the fluid acceleration.

[0161] Preferably, to address hydrodynamic model parameters (such as C) _d ,m _a Due to the uncertainty of the resistance coefficient C, the above solution process employs interval arithmetic. _d Additional mass m _a Using the interval [min, max] as input, the upper bound of tension (peak value of upper bound of force) F with confidence envelope is directly derived through the interval propagation algorithm. _peak_plus This makes the results more conservative and secure.

[0162] Quantitative feedback limiting factor κ _fb Specifically, the upper limit peak value of the force obtained from the deduction, F _peak_plus With the preset allowable tension limit F _allow (For example, 70% of the cable's rated tensile strength) is compared to quantify the feedback constraint factor (feedback risk) κ. _fb (t).

[0163] An example of the calculation formula is: κ _fb =clip((F _peak_plus / F _allow )-1,0,κ _max Where clip is the clipping function, κ _max It is the maximum factor value (e.g., 1.0). If F _peak_plus ≦F _allow , then κ _fb =0 (no feedback risk); if F _peak_plus >F _allow , then κ _fb κ is a positive value; the larger the value, the higher the risk of exceeding limits caused by the superposition of transient events and feedforward flow. _fb This will serve as the basis for strengthening trajectory constraints.

[0164] Example 5: Detailed description of how to generate an executable trajectory based on a given final constraint table. Final constraint table (limit) _table_final The synthesis process will be described in detail in Example 6.

[0165] The specific process of generating the execution trajectory by integrating the final limit table, frequency band energy ratio, and valve opening plan is achieved through two steps: spectrum shaping and S-shaped limit Jerk trajectory generation.

[0166] Step 5.1: Apply the notch filter parameters contained in the final constraint table to perform spectral shaping on the valve opening plan to obtain a spectral shaping reference.

[0167] Valve opening plan u plan (t) is the original, unoptimized expected trajectory. The final shape table is `limit`. _table_finalIt includes the frequency band energy ratio η _band (t) and feedback limiting factor κ _fb (t) The dynamically adjusted notch filter parameters jointly determined by the notch filter. _param_final (Including center frequency f0, bandwidth df, and depth).

[0168] This step filters out the valve opening plan before trajectory generation. plan The harmonic components of the dangerous frequency f0 of the water flow may be excited in (t). In practice, it is implemented as a digital filter.

[0169] According to the notch parameter _param_final The values ​​of f0, df, depth, and the sampling frequency Fs of the current control system are given. _act Calculate the filter coefficients (e.g., a0, a1, a2, b0, b1, b2) of the digital dual second-order notch filter.

[0170] will u plan (t) is used as the input signal, which is filtered by this digital notch filter to obtain the output signal, i.e., the spectrum shaping reference u. ref (t).

[0171] In a preferred embodiment, to prevent filter parameters from notch _param_final System shocks caused by abrupt changes between different control cycles (e.g., f0 jumping from 0.5Hz to 0.8Hz) also include coefficient smoothing transitions. Specifically, newly calculated filter coefficients are not applied immediately, but are linearly interpolated or exponentially smoothed with the old coefficients from the previous cycle, smoothly transitioning to the new coefficient values ​​over one or more control cycles.

[0172] Furthermore, this step may also include an assessment of the group delay effect, as the notch filter introduces a phase delay (i.e., group delay τ) while filtering out a predetermined frequency. _g ), τ can be estimated online _g If τ _g If a certain threshold is exceeded (which may lead to control instability), the notch filter parameters are actively adjusted, such as appropriately increasing the bandwidth df or decreasing the depth, sacrificing some notch filter performance in exchange for a smaller phase delay.

[0173] Step 5.2: Apply the three-segment S-shaped Jerk trajectory generator to generate the execution trajectory based on the spectral shaping reference under the constraints of Jerk, acceleration, and velocity limits defined in the final shape table.

[0174] The three-segment S-shaped limit Jerk trajectory generator is an algorithm used to generate smooth trajectories. It ensures that the rate of change of acceleration (Jerk) is constant or zero, thus avoiding rigid impacts on actuators (valves) and hydrodynamic systems.

[0175] This step uses the u generated in step 5.1 ref (t) is the target, with limit _table_final The maximum velocity u that varies with time as defined in [the original text] _dot_max_final Maximum acceleration a _ddot_max_final and the maximum Jerk j _max_final As a boundary constraint, the current actual opening degree u is planned in real time. _cur Smooth transition to target u ref The execution trajectory u of (t) _traj_raw (t).

[0176] Specifically, the algorithm executes once within a scrolling window (e.g., every 100ms):

[0177] Calculate the current opening degree u _cur Target u at the next moment ref_next The displacement Δu.

[0178] Based on the magnitude of Δu, and j _max_final ,a _ddot_max_final ,u _dot_max_final Given the constraints, solve for the duration of each stage of the S-curve.

[0179] Scenario 1 (long distance): The trajectory consists of seven stages (acceleration, uniform acceleration, deceleration, uniform speed, acceleration and deceleration, uniform deceleration, and deceleration).

[0180] Scenario 2 (Medium distance): The trajectory cannot reach the maximum speed, and the constant speed segment Tv is zero.

[0181] Scenario 3 (short distance): The trajectory cannot reach the maximum acceleration, and the uniform acceleration / deceleration segment (Ta) is also zero.

[0182] Generate trajectory segments. Based on the solved time profile, within the current control cycle (e.g., 100ms), discretely generate sequences of acceleration a(t), velocity v(t), and position u(t), which serve as the original execution trajectory u. _traj_raw Part of (t).

[0183] Perform rolling updates, using the opening at the end of the trajectory segment as the u value for the next cycle. _cur and for u ref_next Perform feedforward fine-tuning (e.g., to compensate for steady-state errors introduced by the notch filter), and then repeat the process.

[0184] In a preferred embodiment, to improve the robustness of trajectory solving, this step also includes a fast re-solution mechanism. If the S-shaped trajectory generator suffers from overly stringent constraints (e.g., j...), _max_final (Extremely small) or failure to solve within the specified time due to notch filter phase issues (feas)_flag If u = 0, the system will not stop, but will immediately perform a fast re-solution, for example, temporarily further reducing u. _dot_max_final or j _max_final (For example, reduce by 10%), recalculate, and ensure that a safe and feasible trajectory can always be generated.

[0185] Example 6: A detailed description of the specific process of synthesizing the final form constraint table, and a description of the final form constraint table used. _table_final It is formed by the fusion of multiple sources of risks and constraints.

[0186] Step 6.1, couple the basic constraint table and the feedback constraint factor to synthesize the enhanced constraint table, and couple the basic constraint table and the feedback constraint factor to synthesize the final constraint table (in this step, this is manifested as synthesizing an intermediate enhanced constraint table), specifically including: applying the feedback constraint factor as a multiplicative strengthening factor; applying the multiplicative strengthening factor to at least one constraint value in the basic constraint table to dynamically tighten the constraint of that constraint value.

[0187] Basic limit table limit _table_base SRM derived from closed-form spectral risk is based on feedforward risk η _band (t) is generated. Feedback constraint factor κ _fb (t) is derived from transient event triggering and parameter upper bound ET-OPU based on transient event J(t) and force upper bound F. _peak_plus generate.

[0188] This step couples these two risks. Specifically, the feedback form-limiting factor κ... _fb (t) (value ≥ 0) is used as a penalty term or reinforcement factor, acting on the basic bounded table limit in a multiplicative or divisive manner. _table_base The limit values ​​in the table are used to obtain the enhanced bounding table limit. _table_star .

[0189] For example, the coupling formula is as follows:

[0190] u _dot_max_star (t)=u _dot_max_base (t) / (1+ζ1*κ _fb (t));

[0191] j _max_star (t)=j _max_base (t) / (1+ζ2*κ _fb (t));

[0192] notch _param_star.depth =notch _param_base.depth +δ*κ _fb (t);

[0193] u_max_star (t)=u _max_base (t)*(1-ζ3*κ _fb (t));

[0194] Where ζ1, ζ2, ζ3, and δ are all positive calibrable coefficients. It can be understood that when the feedback risk κ... _fb When (t) is 0, the enhanced bounded table limit is used. _table_star equals the limit table _table_base When κ _fb When (t)>0, the limit table is enhanced. _table_star The speed, Jerk, and maximum opening limits are tightened (reduced), and the notch filter depth is increased, thus suppressing feedback risk.

[0195] Step 6.2: Integrate external constraints to synthesize the final form constraint table.

[0196] Based on step 6.1, the coupling of the base constraint table and the feedback constraint factor to synthesize the final constraint table also includes: obtaining the external robust time window constraints and the physical upper limit of the actuator; when synthesizing the final constraint table, performing point-by-point value fusion on the constraints defined by the base constraint table, the feedback constraint factor, the robust time window constraints and the physical upper limit of the actuator.

[0197] This fusion is based on the enhanced form constraint table generated in step 6.1. _table_star (It has been coupled with the basic bounded table limit) _table_base and κ _fb (This was done)

[0198] Obtaining constraints:

[0199] physical upper limit of the actuator hardware _limit_table Specifically, these are the inherent, insurmountable limits of the valve and its servo system (e.g., maximum physical velocity u). _hw_dot_max ).

[0200] Robust time window constraint schedule _θ (t), specifically, refers to time-varying constraints issued by external systems (such as tidal surge warning and scheduling systems), for example, forcing u to... _dot_max It must not exceed 1% / min.

[0201] Point-by-point value fusion: Specifically, by taking the minimum value (min()) point by point, all the above constraints are fused to obtain the final constraint table limit. _table_final (t). An exemplary fusion formula is as follows:

[0202] u _max_final =min(u_max_star ,schedule _θ .u _max (t),u _hw_max );

[0203] u _dot_max_final =min(u _dot_max_star ,schedule _θ .u _dot_max (t),u _hw_dot_max );

[0204] j _max_final =min(j _max_star ,schedule _θ .j(t),j _hw_max );

[0205] limit _table_final (t) ensures that the execution trajectory will not violate any of the feedforward risk, feedback risk, physical upper limit and external scheduling constraints, and realizes a safe envelope under multiple constraints.

[0206] Step 6.3: Generate robust time window constraints.

[0207] Preferably, the method further includes: acquiring tidal bore monitoring data through a tidal bore monitoring unit; predicting the arrival time and risk level of the tidal bore based on the tidal bore monitoring data; and generating robust time window constraints based on the prediction results.

[0208] This step is the robust time window constraint schedule in step 6.2. _θ A preferred source of (t).

[0209] Tidal bore monitoring units (such as tide gauges and radar current meters) deployed at downstream river mouths (such as the entrance to Qinzhou Bay of the Pinglu Canal) are used to acquire tidal bore monitoring data in real time.

[0210] This data is reported to the operational safety center. Based on this data, the predictive model within the center (e.g., a hydrodynamic model or an AI model) predicts the time T when the tidal bore will arrive at the lock gate of the ship. _arrival And risk level (e.g., tidal bore height, maximum flow velocity).

[0211] Automatically generate a schedule based on the prediction results. _θ (t). For example, in [T _arrival -10min,T _arrival Within a time window of +10 minutes, schedule _θ .u _dot_max (t) is set to a very low value (e.g., 0.5% / min), or a pause command is issued directly (u _dot_max=0), and then send the constraint to step 6.2 for fusion.

[0212] Example 7 provides a preferred and additional monitoring and degradation implementation method, which describes in detail the trajectory generation process and performs final arbitration and shaping on the execution trajectory.

[0213] Step 7.1: Evaluate and transition the four-state machine.

[0214] The final form constraint table, frequency band energy ratio, and valve opening plan are integrated to generate the execution trajectory. This also includes: activating a four-state machine with normal, slow opening, pause, and emergency stop modes; the four-state machine performs unidirectional main loop transitions between the four states based on the threshold conditions of the frequency band energy ratio and the threshold conditions of the feedback form constraint factor to output the current state; the final generation of the execution trajectory is controlled by the current state.

[0215] Enable a state machine with four states: {NORMAL, SOFT_OPEN, PAUSE, EMERGENCY_STOP}.

[0216] Input: The state machine transitions (switches) are mainly based on two input signals: η _band (t) (feedforward risk), κ _fb (t) (Feedback risk).

[0217] Threshold and logic: for η _band and κ _fb Set the threshold with hysteresis respectively (e.g., H). _η_hi >H _η_lo ,K _fb_hi >K _fb_lo ) and minimum residence time T _dwell .

[0218] Unidirectional main loop transition: The state machine has strict degradation priority:

[0219] EMERGENCY_STOP>PAUSE>SOFT_OPEN>NORMAL.

[0220] Enter (degrade): If κ _fb (t)≧Feedback limiting factor high threshold (entry threshold) K _fb_hi (For example, extremely high feedback risk), or failure to solve the reported trajectory (feas) _flag If η = 0), then immediately proceed to EMERGENCY_STOP. Otherwise, if η = 0, then proceed to EMERGENCY_STOP. _band (t)≧High threshold of frequency band energy ratio (entry threshold) H _η_hi or κ _fbIf (t) is at a moderate level, then at least SOFT_OPEN will be entered. If the risk persists or η _band If it is extremely high, then it will enter PAUSE mode.

[0221] Exit (Resume): Must be when η _band (t)≦H _η_lo And κ _fb (t)≦K _fb_lo And the dwell time in the current downgraded state is ≥ T _dwell Only then is pressing:

[0222] The sequence is EMERGENCY_STOP->PAUSE->SOFT_OPEN->NORMAL, with each level going backwards. Here, lo is an abbreviation for low, and H... _η_lo K _fb_lo "Equal to" refers to the low threshold of the corresponding parameter, i.e., the exit threshold.

[0223] Preferably, the exit logic is also subject to external constraints. For example, if the current state is on a schedule... _θ (t) (e.g., tidal surge) Within a strong constraint window, the state machine is prohibited from rolling back from PAUSE to SOFT_OPEN, ensuring that external constraints are executed first.

[0224] The state machine outputs the current state (next) in each control cycle. _state (t).

[0225] Step 7.2: Remapping and trajectory shaping based on state parameters.

[0226] The final generation of the execution trajectory is controlled by the current state, specifically including: configuring parameter remapping ratio tables for the four states respectively; selecting the corresponding parameter remapping ratio table according to the current state; applying the corresponding parameter remapping ratio table to remap the Jerk, acceleration and velocity limits in the final constraint table; and performing secondary shaping and trimming of the execution trajectory based on the remapped limits.

[0227] Configure the multiplier table state _gain Among them, the adjustment multiplier gain is the adjustment multiplier, such as the speed limit multiplier g. _udot , acceleration upper limit multiplier g _uddot Jerk upper limit multiplier g _j .

[0228] NORMAL state: gain = {g _j =1.0,g _uddot =1.0,g _udot =1.0}.

[0229] SOFT_OPEN state: gain = {g_j =0.5g _uddot =0.6g _udot =0.6} (Example value).

[0230] PAUSE state: gain = {g _j =0.0,g _uddot =0.0,g _udot =0.0} (Forces velocity and acceleration to zero, maintaining the current position).

[0231] EMERGENCY_STOP state: gain points to a preset table of fixed limits for fast and safe shutdown. _slew .

[0232] Select and remap, based on the next output from step 7.1. _state (t), select the corresponding gain. Multiply this gain by the limit. _table_final The corresponding limit values ​​in the table are used to obtain the operation limit table op. _table (t). For example, if the state is SOFT_OPEN, then op _table .u _dot_max =limit _table_final .u _dot_max *0.6.

[0233] Secondary shaping and trimming will transform the original execution trajectory u _traj_raw (t), and the operation bounded table op _table The Jerk, acceleration, and velocity limits in (t) (which have been remapped by the state machine gain) are compared and trimmed a second time.

[0234] The final output trajectory u _traj (t) represents the result after the feedforward risk η _band Feedback risk κ _fb External constraints schedule _θ Physical limit hardware _limit and state machine _gain A highly secure execution trajectory with five-fold envelope constraints.

[0235] Preferably, the present invention may further include consistency and safety bypass steps, independent of the aforementioned four-state machine, as a last line of defense for continuous monitoring:

[0236] Tracking error e _u , including the final u _traj (t) feedback of the actual valve position u _meas The difference in (t).

[0237] Implement health status_flag (Such as servo drive alarms, communication timeouts). If e _u Exceeding limits or health _flag If an exception occurs, the fallback will be triggered immediately. _flag =1, force bypass all the above trajectory generation logic, and set the control command Ctrl. _cmd (t) Covers a preset, condition-independent conservative safety table. _table (For example, always close the valves slowly at a rate of 0.1% / min and 0.5% / min, regardless of the circumstances), to ensure the system has safety redundancy in the event of end-point failure.

[0238] Example 8 describes how the system uses real-time data to update the feedforward model and feedback threshold online during operation.

[0239] The method also includes: performing a closed-loop update step to generate updated feedforward kernel parameters and updated trigger threshold parameters based on the execution trajectory and historical measurement data; wherein, the updated feedforward kernel parameters are used for the step of extrapolating hydrodynamic spectrum risks based on valve opening plans; and the updated trigger threshold parameters are used for the step of identifying transient events and quantifying and generating feedback constraint factors based on mooring tension and local flow velocity predictions.

[0240] Specifically, this closed-loop update step is executed after the end of a control cycle (e.g., a complete valve opening or closing process), or in batches after multiple control cycles, utilizing the execution trajectory u issued in this cycle. _traj (t) and historical measurement data (e.g., flow) collected synchronously during this period. raw (t) and F(t)).

[0241] The update steps specifically include:

[0242] Step 8.1: Update the feedforward kernel parameters, used to update the feedforward kernel h. r Specifically, the execution trajectory u of this cycle _traj (t) (as input) and local flow velocity v _meas_hist (t) (as output) serves as the new data pair, used to update the Recursive Least Squares (RLS) algorithm. Using the recursive formula of the RLS algorithm, the covariance matrix and kernel vector are updated based on the new data pair, yielding the updated feedforward kernel parameters. _param_next (i.e., the updated h) r Updated parameters (e.g., h) r (λ,β,L) will be written into the kernel of the nuclear bank. _bank In the middle, it is called in the next control cycle, so that the feedforward model can track the slow time-varying hydrodynamic characteristics.

[0243] Step 8.2: Update the trigger threshold parameters. Specifically, the system retrieves the event logs for the current period. _log The log records the trigger time and peak J of the transient event. _peak and the corresponding feedback factor κ _fb Information such as this. The system (e.g., the running security hub) uses statistical analysis of this log to calculate the false alarm rate for the current period (e.g., events were triggered but...). _fb The threshold parameter is either always 0 or has a false negative rate (e.g., F(t) has clearly exceeded the limit but J(t) has not triggered). Based on this statistical result, the parameters used in the process of identifying transient events based on the joint criterion are fine-tuned to obtain the updated trigger threshold parameter threshold. _next For example, if the false alarm rate is too high, the entry threshold J should be appropriately increased. _hi Or increase the hysteresis interval (J) _hi -J _lo If the triggering is too slow (insufficient lead time), then appropriately reduce J. _hi Alternatively, the weights of k1 and k2 can be adjusted. The updated threshold will be used in the next control cycle call to improve the accuracy of transient event identification.

[0244] Example 9 provides an implementation process in a specific scenario.

[0245] Suppose a large water-saving lock is performing a water-saving filling operation. Initial conditions: The valve is currently in the closed state (u _cur =0%). Nominal (maximum) valve opening speed u _dot_nom =5% / min, nominal Jerk j _nom =0.5% / min 3 The maximum allowable tension F of the mooring system _allow =30kN. The mooring tension F(t) initially stabilizes at 10kN. Valve opening plan u plan (t): The system plans to use u within the next 60 seconds. _dot_nom Open at a constant speed of 5% / min to 5%.

[0246] The calculation process is as follows:

[0247] Step 9.1: Conduct feedforward risk simulation.

[0248] Obtain the feedforward kernel: Assuming online identification has converged, the kernel h of the current opening interval (small opening) is... r (Discretized) as [0.3, 0.5, 0.2].

[0249] Predicted flow rate: u plan (t) (ramp signal) and h rPerform convolution to obtain the local velocity prediction v hat (t).

[0250] Spectrum estimation: for v hat (t) Perform short-time Fourier transform (STFT) analysis. Assuming that at t = 10s, the analysis results show energy concentration in the critical frequency band (0.2-2Hz), the frequency band energy ratio η is calculated. _band (10s) = 0.15. (0.15 indicates a medium feedforward risk).

[0251] Mapping basis form table:

[0252] Assume the mapping coefficients γ = 2.0 and ρ = 1.5.

[0253] Calculate u _dot_max_base =u _dot_nom / (1+γ*η _band )=5.0 / (1+2.0*0.15)=3.85% / min.

[0254] Calculate j _max_base =j _nom / (1+ρ*η _band )=0.5 / (1+1.5*0.15)=0.408% / min 3 .

[0255] Meanwhile, short-time Fourier transform (STFT) analysis yielded a spectral peak f0 = 0.8 Hz, with depth determined by η. _band The mapping is -5dB.

[0256] SRM Output: Limit _table_base ={u _dot =3.85, j=0.408, notch _f0 =0.8Hz,...}.

[0257] Step 9.2: Identify transient events.

[0258] At t=15s, due to the slight swaying of the ship in the lock chamber, the tension F(t) of a certain mooring hook suddenly jumps from 10kN to 22kN, and then oscillates around 25kN.

[0259] Construct a criterion, assuming α = 0.5, at t = 15.01s, |D 0.5 The calculated value of F(t)∣ surges to 1.2. S _w The calculated value of wavelet sparsity (t) soars to 0.9. Assume k1 = 0.6, k2 = 0.5. The joint criterion J(t) = 0.6 * 1.2 + 0.5 * 0.9 = 1.17.

[0260] Identify events and set a threshold J _hi =0.8, J _lo =0.4, minimum residence time T _min =50ms. J(t)=1.17≧J _hi Assume this state lasts for 60ms (>T). _min ET-OPU trigger: At t = 15.06s, the transient event is confirmed. _flag =1.

[0261] In response to the event, update the equivalent stiffness k _line Get k _line_plus Combined with k _line_plus and v in step 9.1 hat (t) (velocity prediction), solve the hydrodynamic model. The upper limit peak value of the short-term force F is derived. _peak_plus = 42kN. Upper limit of comparison: F _peak_plus =42kN>F _allow =30kN. Quantization factor: κ _fb =(F _peak_plus / F _allow )-1=(42 / 30)-1=0.4.

[0262] Step 9.3: Perform trajectory integration and execution.

[0263] Coupled form limit table: Get limit _table_base (u) _dot =3.85) and κ _fb =0.4. Assume the coupling coefficient ζ1 = 1.5. Calculate u. _dot_max_star =u _dot_max_base / (1+ζ1*κ _fb = 3.85 / (1+1.5*0.4) = 2.41% / min. This yields the enhanced constraint table limit. _table_star .

[0264] Fusion constraint: Assume there is no external schedule at this time. _θ Constraints, physical upper limit u _hw_dot_max =10% / min. u _dot_max_final =min(u _dot_max_star ,u _hw_dot_max =min(2.41,10) = 2.41% / min. This gives the limit. _table_final .

[0265] State machine evaluation: The state machine detected κ. _fb =0.4 (feedback risk) exceeds K _fb_hi(e.g., 0.2). The state machine immediately switches from NORMAL to SOFT_OPEN.

[0266] Parameter remapping: the scaling table for the SOFT_OPEN state _gain In the middle, the speed multiplier g _udot =0.6. Remapping limit: op _table .u _dot_max =u _dot_max_final *g _udot =2.41*0.6=1.45% / min.

[0267] Trajectory Generation: S-shaped Jerk trajectory generator, with u ref (t) (shaped by a notch filter at f0 = 0.8 Hz) is the target value, at j _max (Already κ) _fb and state _gain (Double tightening) and u _dot_max Under the strict constraint of 1.45% / min, the final execution trajectory u is generated. _traj (t).

[0268] Despite the original plan plan (t) is the 5% / min turn-on speed, but due to SRM (η) _band =0.15) and ET-OPU (κ) _fb The combined effect of (=0.4) and arbitration through the state machine (SOFT_OPEN) determines the actual execution trajectory u of the valve. _traj (t) is smoothly limited to below the maximum speed of 1.45% / min, and the 0.8Hz component in its spectrum is suppressed, avoiding further deterioration of the mooring tension F(t) and ensuring safe passage through the lock.

[0269] According to one aspect of this application, the steps for online identification of lightweight feedforward kernels (aperture -> local velocity) can also be:

[0270] Extract the historical execution trajectory from the previous control cycle from the alignment data packet. _traj_hist (t) and local flow velocity v _meas_hist (t). Using recursive least squares with a forgetting factor, impulse response kernel h of length L (3-5 taps) is identified. r (Header) (r represents the openness interval label).

[0271] For example, the formula is processed as follows:

[0272] Minimize Σ _k (v _meas_hist (k)-Σ _i h r (i)•u_traj_hist (ki)) 2 +β•∣∣h r || 2 The forgetting factor λ controls the time-varying weights.

[0273] Obtain kernel parameters _param ={h r ,λ,β,L,interval label r}, used for feedforward inference and risk mapping.

[0274] According to one aspect of this application, the generation process of the three-segment S-shaped jerk trajectory can also be as follows:

[0275] Obtaining the Spectrum Shaping Reference u _ref (t), Current valve opening u _cur The final bounded form table limit _table_final The maximum velocity u in (t) _dot_max_final Maximum acceleration a _ddot_max_final ,Maximum Jerk j _max_final Time step dt.

[0276] For each control moment, calculate the valve opening u. _cur Target spectrum shaping reference u at the next time step _ref_next Find the shortest time, constrained Jerk S-shaped trajectory. Solve for the parameters of the three-segment structure (acceleration, uniform acceleration, deceleration):

[0277] Calculate the displacement Δu = u _ref_next -u _cur ;

[0278] Calculate the theoretical acceleration time Tj = a under the jerk constraint. _ddot_max_final / j _max_final ;

[0279] At maximum speed u _dot_max_final Maximum acceleration a _ddot_max_final Maximum Jerk j _max_final Under the common constraints, solve for the durations Tj1, Ta, Tj2, and Tv of each segment (if Δu is too small to enter the uniform velocity segment, solve for the segment with reduced order as "no Tv solution").

[0280] Using a closed-loop time profile (example, with the first stage jerk=+j _max_final ),

[0281] Acceleration a(t) = j _max_final *t,0≦t <Tj1;

[0282] Velocity v(t) = ∫a(t)dt, displacement u(t) = ∫v(t)dt;

[0283] Other stages are spliced ​​together according to symmetrical or speed-limited / acceleration-limited saturation conditions;

[0284] In the discrete domain, the execution trajectory segment u is generated with a step size of dt. _seg (t:k->k+H), and perform saturation clipping;

[0285] The end point of the trajectory segment is used as the next period u _cur The new value for the scrolling window H can be 0.5-1.0s.

[0286] Obtain execution trajectory u _traj_raw (t) will be remapped according to the state during the remapping and trajectory shaping process; at the same time, the trajectory feasibility flags (feas) will be obtained. _flag (If the solution fails or there is a constraint conflict, the value is 0), which is used for safe bypass judgment.

[0287] According to one aspect of this application, the four-state machine evaluation (entry / exit / hysteresis / minimum dwell) process can also be:

[0288] Obtain the frequency band energy ratio η _band (t), feedback limiting factor κ _fb (t), trajectory feasibility markers _flag Optional event flags _flag The current state (t-1) and the timestamp.

[0289] Configuration state set {NORMAL,SOFT_OPEN,PAUSE,EMERGENCY_STOP} and threshold / hysteresis / minimum dwell time parameter table:

[0290] η threshold: H _η_hi >H _η_lo ;

[0291] κ threshold: K _fb_hi >K _fb_lo ;

[0292] Minimum dwell time: T _dwell_normal ,T _dwell_soft ,T _dwell_pause ,T _dwell_estop ;

[0293] Priority: EMERGENCY_STOP > PAUSE > SOFT_OPEN > NORMAL;

[0294] One-way main loop: NORMAL => SOFT_OPEN => PAUSE => EMERGENCY_STOP, reverse loop only returns level by level when the exit threshold and minimum dwell time are met.

[0295] For example, the transfer rule:

[0296] If fees _flag =0 or κ _fb (t)≧K _fb_hi If so, it will immediately enter EMERGENCY_STOP;

[0297] Otherwise if η _band (t)≧H _η_hi or event _flag =1, then at least SOFT_OPEN will be entered; if η _band High and consistently exceeding T _hold or κ _fb (t)≧K _fb_mid Then enter PAUSE;

[0298] Exit rule: When η _band (t)≦H _η_lo And κ _fb (t)≦K _fb_lo And the current state dwell time is ≥ the corresponding minimum dwell time, and the system will back down step by step according to the unidirectional main loop;

[0299] Optional strongly constrained window (if schedule is enabled) _θ ): Prevents PAUSE from falling back to SOFT_OPEN.

[0300] Get the next state _state (t) and the state dwell timer well _counter It is used for parameter remapping and instruction encoding.

[0301] This application provides an active suppression scheme that does not employ a fixed valve curve. Instead, it predicts local flow velocity online using a feedforward kernel and performs short-time spectral estimation on this prediction to extrapolate the critical frequency band energy η of the valve plan online. _band And the spectral peak f0. Based on this, the system generates a basic conformal table containing notch filter parameters through closed-loop mapping, and performs spectral shaping on the opening plan before execution. This achieves pre-suppression of spectral risks and solves the problems of feedforward blindness and spectral risk.

[0302] This application abandons the fixed threshold alarm and constructs a joint criterion J(t) by calculating the fractional derivative and wavelet sparsity of the mooring tension F(t). This criterion is highly sensitive to the precursors of signal abrupt changes (rather than just amplitude), and through hysteresis triggering logic, it can identify micro-transient events before the peak force is fully formed. The short-term upper bound of the force F is derived from this criterion. _peak_plus To quantify the feedback factor κ _fb This has enabled a shift from post-event alarms to pre-event prediction and intervention, solving the problems of feedback lag and inability to capture micro-transient precursors of loosening and tightening.

[0303] This application will provide feedback on the limiting factor κ. _fb As a multiplicative strengthening factor, the constraints of the basic form constraint table (feedforward risk) are dynamically tightened; on the other hand, the four-state machine simultaneously considers the energy η of the danger band. _band and feedback limiting factor κ _fb The two inputs are arbitrated. Ultimately, the two risk factors work together to progressively constrain the trajectory (e.g., from 5% / min to 1.45% / min), resolving the issue of lack of coordination.

[0304] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details of the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all fall within the protection scope of the present invention.

Claims

1. A method for safe and coordinated control of ship lockage of a large water-saving ship lock, characterized in that, The method comprises the following steps: Based on the valve opening plan to deduce the hydrodynamic spectrum risk, the local flow rate prediction, the frequency band energy ratio and the base limit shape table are obtained; Based on the mooring tension and the local flow rate prediction, the transient event is identified and the feedback limit shape factor is quantified; Coupling the base limit shape table and the feedback limit shape factor to synthesize the final limit shape table, combining the final limit shape table, the frequency band energy ratio and the valve opening plan to generate the execution trajectory; Wherein, the frequency band energy ratio refers to the proportion of the spectrum energy in the preset dangerous frequency band to the total frequency band energy, which is determined by performing short-time spectrum estimation on the local flow rate prediction; The base limit shape table at least includes: maximum jerk limit value, maximum acceleration limit value, maximum speed limit value, and notch filter parameter, which is converted by a closed mapping function; Based on the mooring tension and the local flow rate prediction, the transient event is identified and the feedback limit shape factor is quantified, including: Analyzing the mooring tension signal to construct a joint criterion; Identifying the transient event based on the trigger condition of the joint criterion; In response to the transient event, the short-term stress upper limit is deduced combined with the local flow rate prediction; Comparing the short-term stress upper limit with the preset allowable tension upper limit to quantitatively obtain the feedback limit shape factor.

2. The method of claim 1, wherein, Analyzing the mooring tension signal to construct a joint criterion, including: Calculating the fractional derivative of the mooring tension signal; Evaluating the wavelet sparsity of the mooring tension signal; Weighted combination of fractional derivative and wavelet sparsity to construct joint criterion.

3. The method of claim 1, wherein, Identifying the transient event based on the trigger condition of the joint criterion, including: Set the entry threshold and the exit threshold, wherein the entry threshold is higher than the exit threshold to form a hysteresis interval; Continuously monitor and compare the joint criterion with the entry threshold; When the joint criterion is not less than the entry threshold, and the state duration meets the minimum residence time, the transient event is identified.

4. The method of claim 1, wherein, In response to the transient event, the short-term stress upper limit is deduced combined with the local flow rate prediction, and the feedback limit shape factor is quantified, including: In response to the identified transient event, update the equivalent mechanical parameters of the mooring system; Based on the updated equivalent mechanical parameters and the local flow rate prediction, the hydrodynamic equivalent model is constructed and solved to deduce the short-term stress upper limit; According to the comparison result of the short-term stress upper limit and the preset allowable tension upper limit, the feedback limit shape factor is determined.

5. The method of claim 1, wherein, Based on the valve opening plan to deduce the hydrodynamic spectrum risk, the local flow rate prediction, the frequency band energy ratio and the base limit shape table are obtained, including: Apply the feedforward kernel to the valve opening plan to obtain the local flow rate prediction; Perform short-time spectrum estimation on the local flow rate prediction to determine the frequency band energy ratio; Convert the frequency band energy ratio to the base limit shape table by a closed mapping function; The base limit shape table at least includes: maximum jerk limit value, maximum acceleration limit value, maximum speed limit value, and notch filter parameter.

6. The method of claim 5, wherein, The feedforward kernel is obtained by an online identification method, which includes: Obtain the historical execution trajectory and the corresponding local flow rate in the field; Apply the recursive least squares algorithm with forgetting factor to identify the feedforward kernel based on the historical execution trajectory and the local flow rate in the field.

7. The method of claim 5, wherein, The determination of the notch filter parameter includes: Perform short-time spectrum estimation on the local flow rate prediction to extract the spectrum peak of the dangerous frequency band; Determine the center frequency of the notch filter parameter based on the spectrum peak of the dangerous frequency band; Based on the band energy proportion, dynamically set the bandwidth and depth of the notch filter parameters.

8. The method of claim 1, wherein, Fusing the final limit-shaping table, the band energy proportion and the valve opening plan, generating the execution trajectory, including: Applying the feedback limit-shaping factor as a multiplicative reinforcement factor; Applying the multiplicative reinforcement factor to at least one limit value in the base limit-shaping table, dynamically tightening the constraint of the limit value, synthesizing the final limit-shaping table; Applying the notch filter parameters contained in the final limit-shaping table to the spectrum shaping of the valve opening plan, obtaining the spectrum shaping reference; Applying the three-section S-shaped limit jerk trajectory generator, generating the execution trajectory based on the spectrum shaping reference under the constraints of the jerk, acceleration and speed limits defined by the final limit-shaping table.

9. A ship lockage safety coordination detection device for a large water-saving ship lock, characterized by, Comprise: A global state perception module for providing real-time data required by the operation safety hub for decision-making; An operation safety hub module for receiving and processing data from the global state perception module; A collaborative safety control module for outputting control signals according to the instructions of the operation safety hub; The global state perception module and the collaborative safety control module are connected with the operation safety hub module for data and control signal connection; The device is used to implement the method steps of any one of claims 1-8.

Images