Guideway water level linkage prediction method

Abstract

This invention discloses a method for predicting water levels in navigation channels, comprising: constructing a background field of water levels at control sections and quantifying the natural attenuation time parameter of water level fluctuations to establish an effective calculation time window. By bidirectionally filtering historically executed gate openings and future planned gate openings within the time window, superimposed wave components are generated using time-shift alignment and amplitude accumulation based on wave propagation characteristics. This invention employs a phase-aware and nonlinearly corrected convolutional superposition model to calculate the phase interference and nonlinear effects of multiple gate opening fluctuations, coupling the background field to generate a predicted water level along the navigation channel. This invention solves the problem of accurate water level prediction under complex superposition of multiple gate openings, improving the intelligence level of navigation safety early warning.

Description

Technical Field

[0001] This invention belongs to the field of lock control technology, and in particular to a method for predicting the water level of a navigation channel. Background Technology

[0002] Large navigation hubs are typically equipped with multi-line locks, and their frequent water release operations generate unsteady currents and water level fluctuations in the downstream approach channels. These fluctuations directly affect the safety of navigation and the efficiency of lock passage, especially in long, restricted approach channels where the cumulative effect of these fluctuations can lead to ships running aground or experiencing mooring difficulties. Therefore, accurately predicting water level fluctuations along the approach channels can optimize lock scheduling, ensure navigation safety, and improve channel capacity.

[0003] Currently, existing technical solutions for studying the water level in navigation channels include: offline simulation using one-dimensional or multi-dimensional numerical models, or extrapolation of the water surface line based on the assumption of steady flow; in terms of wave fluctuation calculation, analytical formulas based on long-wave theory are used, assuming a fixed waveform for a single lock discharge, and water level changes under multiple lock operations are estimated through simple linear superposition. Furthermore, some scheduling systems rely solely on current measured water levels or trend extrapolation for decision-making, lacking the ability to predict hydrodynamic processes in the short term.

[0004] Furthermore, existing technologies struggle to address the complex hydrodynamic coupling issues arising from the combined operation of multiple locks, primarily due to three aspects: nonlinear superposition failure, poor parameter adaptability, and a disconnect between prediction and scheduling. Therefore, further research and innovation are needed to resolve these problems in existing technologies. Summary of the Invention

[0005] The purpose of this invention is to provide a method for predicting the water level of navigation channels in conjunction with water levels, in order to solve one of the problems existing in the prior art.

[0006] According to one aspect of this application, a method for predicting pilotway water levels includes:

[0007] Analyze the hydrological evolution characteristics of the downstream control section of the navigation channel and construct a background field of the control section water level describing the impact of the hub's discharge on the downstream water level benchmark.

[0008] The memory effect of hydrodynamics in the pilot channel is quantified to determine the natural decay time parameter of water level fluctuations, and an effective calculation time window covering historical retrospective periods and future prediction periods is constructed based on the current prediction time.

[0009] By utilizing the effective calculation time window, bidirectional time-series filtering is performed on the multi-source lock operation information to extract the effective historical lock operations and effective future planned lock operations that fall within the effective calculation time window.

[0010] For each of the selected valid historical gate openings and valid future planned gate openings, time-shift alignment and amplitude accumulation based on wave propagation characteristics are performed to generate superimposed wave components that reflect the disturbance effect of multiple gate openings.

[0011] The superimposed wave components are coupled to the background water level field of the control section to generate a linked water level prediction process for each section along the navigation channel.

[0012] Beneficial effects: This invention solves the problem of water level prediction under complex superposition of multiple lock operations, improving the intelligence level of navigation safety early warning. In other words, this invention addresses the complex hydrodynamic coupling scenario of multi-lock joint operation, solving three major problems: nonlinear superposition failure, poor parameter adaptability, and disconnect between prediction and scheduling. It achieves accurate water level prediction under complex superposition of multiple lock operations, enhances the adaptability of parameters to complex operating conditions, establishes a linkage mechanism between prediction results and scheduling decisions, and improves the intelligence and practicality of navigation safety early warning. The related technical effects will be described in detail below with reference to specific embodiments. Attached Figure Description

[0013] Figure 1 A flowchart of a navigation channel water level linkage prediction method provided in an embodiment of this application.

[0014] Figure 2 A flowchart for performing time-shift alignment and amplitude accumulation based on wave propagation characteristics is provided for embodiments of this application.

[0015] Figure 3 The flowchart provided in the embodiments of this application shows the generation of superimposed wave components reflecting the disturbance effect of multiple gate discharges under the first calculation mode.

[0016] Figure 4 This is a flowchart illustrating the phase characteristics of each gate's arrival at the target prediction section, provided as an embodiment of this application.

[0017] Figure 5 A flowchart illustrating a nonlinear correction rule configured to control waveform evolution, provided in an embodiment of this application. Detailed Implementation

[0018] 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.

[0019] It should be noted that the terms "first," "second," etc., 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 embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0020] To address the aforementioned issues, the applicant conducted in-depth searches and analyses, and discovered:

[0021] Specifically, the traditional linear superposition method ignores the nonlinear deformation of waves in shallow water and the resonance amplification effect when multiple wave phases meet, resulting in large prediction errors when the wave amplitude is large or the frequency is close. Furthermore, the existing model uses fixed attenuation parameters and static time windows, which cannot dynamically adjust the memory duration according to real-time water conditions, resulting in insufficient robustness. In addition, the prediction results are often only used as a passive reference, and a reverse closed-loop optimization mechanism based on risk quantification has not been established, so it cannot directly guide the dynamic adjustment of the gate opening plan.

[0022] To solve these problems, combined with Figures 1 to 5 The present invention will be specifically described through the following embodiments.

[0023] Example 1: A method for predicting the water level of a navigation channel is provided to solve the problem of constructing a unified calculation boundary from multi-source heterogeneous data, and to provide standardized input for subsequent fluctuation superposition.

[0024] As an example, the present invention can be implemented by performing the following steps:

[0025] Step S101: Analyze the hydrological evolution characteristics of the downstream control section of the navigation channel and construct a background field of the control section water level describing the impact of the hub's discharge on the downstream water level benchmark.

[0026] In this embodiment, the system establishes a communication connection with the hub control center to obtain historical hydrological data of the downstream control section of the pilot channel. The background water level field of the control section refers to the benchmark water level process after eliminating short-period lock fluctuation interference. The system constructs this background field using a pilot channel entrance gate area control water level prediction model.

[0027] As one possible implementation, the system reads the real-time discharge flow data Q(t) (unit: cubic meters / second) and uses the empirical formula Z... _m(t) = 0.013 × Q(t) 2 / 3 +35.27, calculate the background field data Z of the water level at the control section. _m (t). The coefficients 0.013 and the constant 35.27 in this formula are specific navigation channel parameters obtained based on regression analysis of historical measured data. Through this step, the system outputs background field data of the control section water level in continuous time series form, which serves as the reference for subsequent overlays.

[0028] Step S102: Quantify the memory effect of the hydrodynamics of the pilot channel, determine the natural decay time parameter of water level fluctuations, and construct an effective calculation time window covering the historical retrospective period and the future prediction period based on the current prediction time.

[0029] In this embodiment, the memory effect reflects the persistence of long wave propagation and attenuation within the pilot channel. To quantify this effect, the system determines the natural attenuation time parameter T of water level fluctuations based on historical monitoring data. _d Specifically, the system identifies the wave process caused by a single lock discharge, and records the duration of the process when the measured wave height decays to no more than 5% of the initial peak value or below the set minimum effective wave height threshold (e.g., 0.05 meters).

[0030] Based on statistical analysis, this embodiment will use T _d The timeframe is set to 4 hours. Based on this, the system constructs an effective calculation time window, which covers the period from the current time t back to T. _d The historical period (i.e., t-4 hours) and the future period extending backward to the forecast cutoff time (e.g., t+4 to 16 hours).

[0031] Step S103: Utilize the effective calculation time window to perform bidirectional time-series filtering on the multi-source lock operation information, and extract the effective historical lock operations and effective future planned lock operations that fall within the effective calculation time window.

[0032] In this embodiment, the system performs bidirectional data filtering. For historical data, the system extracts data from the hub centralized control and monitoring system within [tT]. _d The system calculates the actual number of lock operations within the time period [t, t+predicted duration]. For future data, the system extracts the planned lock operations within the time period [t, t+predicted duration] from the navigation scheduling system. For example, if the current time is 0, the system will identify n historical lock operations of Lock No. 1 within -4 hours, and k planned lock operations within the future predicted time period. The selected lock operations constitute the effective input set for subsequent fluctuation calculations.

[0033] Step S104: For each of the selected valid historical gate openings and valid future planned gate openings, perform time-shift alignment and amplitude accumulation based on wave propagation characteristics to generate a superimposed wave component that reflects the disturbance effect of multiple gate openings.

[0034] Accordingly, in the first calculation mode, superimposed fluctuation components are generated by measuring distance, calculating lag, calling waveforms, translation and scaling, and linear superposition.

[0035] In other embodiments, generating superimposed wave components may also include:

[0036] For each independent lock discharge in the lock operation information, based on the time of occurrence, the preset single fluctuation process triggered by the single discharge is time-shifted to obtain the individual fluctuation component of that lock discharge.

[0037] All the obtained individual fluctuation components are linearly superimposed to form a superimposed fluctuation component.

[0038] Step S105: Couple the superimposed wave component to the background field of the water level at the control section to generate a linkage water level prediction process for each section along the navigation channel.

[0039] In this embodiment, the system will control the background field Z of the cross-section water level. _m (t) and superimposed wave component H _p Linear coupling is performed between (x, t). The specific calculation formula can be described as follows:

[0040] Z(x,t)=Z _m (t)+H _p (x, t).

[0041] The above Z(x, t) represents the coordinated water level prediction process for each section along the navigation channel in the future. In other words, x corresponds to the section location, and t corresponds to the time node.

[0042] Another example is that lock operation information can consist of historical monitoring data from the hub centralized control and monitoring system and future scheduling data from the navigation scheduling plan system;

[0043] Historical monitoring data, including the actual opening time, measured working head, and operating mode of the gates that have been executed, is used to determine the fluctuation triggering origin of valid historical gates that have been executed in the two-way time series screening.

[0044] Future scheduling data, including the planned opening time of the gates to be executed, the expected working head, and scheduling constraints, is used to deduce the potential fluctuation timing of effective future planned gates in two-way timing screening.

[0045] Accordingly, the actual opening time and measured operating head in historical monitoring data can be used to calculate the wave intensity and determine the initial phase and initial energy of the wave (i.e., the origin of wave excitation). The expected operating head and scheduling constraints in future scheduling data can be used to extrapolate the specific time points at which future gate operations may trigger waves (i.e., the potential wave sequence).

[0046] For example, the system will read the working head (e.g., 20 meters) and operating mode (e.g., single or double valve opening) for each gate operation, and use them as index keys for subsequent calls to the corresponding standard waveform from the single gate response function library.

[0047] Another example provides a different method for predicting navigation channel water levels, specifically:

[0048] The process of determining the background water level at the preset control section within the navigation channel;

[0049] Based on the lock operation information, a superimposed wave component representing the impact of multiple lock discharges is generated;

[0050] By coupling the background water level process with the superimposed fluctuation component, the linked water level prediction of the target section along the navigation channel is derived.

[0051] Another example is that lock operation information can also include:

[0052] Historical lock operation records prior to the current forecast time, and future lock operation plans after the current forecast time;

[0053] When generating superimposed wave components, a unified response model is used to process the historical lock operation records and the lock operations in the future lock operation plan.

[0054] In other words, historical lock operation records correspond to historical monitoring data, and future lock operation plans correspond to future scheduling data.

[0055] Example 2 describes an optional implementation of a fast prediction model based on the principle of linear superposition, suitable for scenarios with small amplitude or high computational speed requirements. Specifically, it includes:

[0056] Step S201: Calculate the spatial transmission distance from the location of the lock wave source to any target predicted section along the pilot channel.

[0057] In this step, the system spatially discretizes the pilotway. For each calculated section x along the pilotway, the system calculates the distance along the path between it and the lock outlet (wave source), i.e., the spatial transmission distance L. _x This distance is a fundamental geometric parameter for calculating the propagation delay of waves.

[0058] Step S202: Calculate the propagation time lag of the wave in the pilot channel based on the spatial transmission distance, and perform phase correction on the wave response of the effective historical executed gates and the effective future planned gates to achieve time translation alignment.

[0059] Accordingly, the system calculates the wave velocity c using the long-wave velocity formula c=sqrt(g×h), where g is the acceleration due to gravity and h is the water depth at the cross-section. The propagation time lag τ is also considered. _x Calculated as spatial transmission distance L _x Divide by the wave speed c. For an activation time of T. e The time when the wave reaches section x is T. e +τ _x The system replaces the independent variable t of the fluctuation function with (tT) e -τ _x This achieves rigid translation of the waveform on the time axis, completing phase correction and alignment.

[0060] Step S203: Quantify the spatial energy dissipation of the wave propagation along the spatial transmission distance, use the spatial energy dissipation to determine the intensity attenuation ratio of each gate wave response when it reaches the target prediction section, and complete the amplitude accumulation.

[0061] In this embodiment, spatial energy dissipation reflects the energy loss of the wave during propagation. The system uses a function that includes a distance term to quantify this indicator. For example, the intensity attenuation ratio can be expressed as (1-(x / L)). 2 ), where x is the current cross-section distance and L is the total length of the navigation channel. The system uses this ratio to scale the initial wave height generated at the source, calculates the remaining amplitude when the wave reaches the target cross-section, and realizes amplitude accumulation calculation.

[0062] Step S204: Call the preset single-wave basic waveform and use the decay function based on the exponential law to map the spatial energy dissipation into a linear decay coefficient.

[0063] In this embodiment, the system presets a single-wave basic waveform H. _s (x, t). Specifically, this waveform can be described by the following formula: H _s (x, t) = 0.85 × 1.15 × H 0.5 ×B×(gh) 0.5 ×(1-x 2 / L 2 )×exp(-1.054×ν×L×t / (ghB))×sin(π×(gh) 0.5 ×t / (2×L));

[0064] In this context, the parameter in the exponential decay term exp(-1.054×ν×L×t / (ghB)) is the mapped linear decay coefficient, comprehensively reflecting the dissipation effects in both time and space. ν represents the kinematic viscosity of the water (unit: m³). 2 / s), sin(π×(gh) 0.5×t / (2×L)) represents the periodic oscillation factor (dimensionless), 0.85 and 1.15 correspond to empirical correction coefficients (dimensionless), B corresponds to the average width of the pilot channel, h represents the average water depth of the pilot channel (unit: m), and (gh) 0.5 This represents the speed of wave propagation (unit: m / s), that is, the speed at which gravitational waves propagate in shallow water, (1-x 2 / L 2 ) represents the position decay factor (dimensionless), π corresponds to pi, and H represents the baseline fluctuation amplitude (unit: m), which is the maximum height when the fluctuation initially occurs.

[0065] Step S205: Perform rigid time translation on the basic waveform of the single fluctuation of each gate according to the propagation time lag, and perform amplitude scaling on the translated waveform according to the linear attenuation coefficient to obtain the linear response sequence of each gate.

[0066] In this embodiment, for each valid gate i, the system sets its specific operating parameters (opening time ΔT). ei That is, the opening time of gate i; the head H) is substituted into the basic waveform formula. Specifically, when t-ΔT ei When ≥0, calculate H _sei (x, t); otherwise 0. This generates a complete linear response sequence of the gate at the target section x over time. That is, H _sei This represents the effective future plan gate.

[0067] Step S206: Perform a linear superposition operation on the linear response sequences of all valid historical executed gates and valid future planned gates to obtain the superimposed fluctuation components.

[0068] In this embodiment, the linear response sequences of all n historical gates and k future planned gates are summed. The specific calculation formula is as follows:

[0069] H _p (x, t) = ∑(H) _pei )+∑(H _sei ).

[0070] The above summation result H _p (x, t) represents the superimposed fluctuation component reflecting the disturbance effect of multiple gate discharges. Alternatively, H... _pei H _sei These correspond to valid historical executed gates and valid future planned gates, respectively.

[0071] Example 3 describes a second calculation mode suitable for complex working conditions, which solves the prediction accuracy problem under large amplitude working conditions through phase sensing and nonlinear correction.

[0072] In the second calculation mode, the superimposed wave component reflecting the disturbance effect of multiple gate discharges is generated through a convolution superposition process of multiple gate phase sensing and nonlinear correction. This process includes:

[0073] Step S301: Construct a single gate response function library containing standard waveform characteristics under different operating conditions, and configure nonlinear correction rules for controlling waveform evolution.

[0074] In this embodiment, the system constructs a single-gate-time response function library offline. This library stores standardized single-gate-time fluctuation response sequences, indexed by operating head, gate operation mode, and gate chamber number. Specifically, the system calculates standard response data under different heads (e.g., 20m, 30m). Furthermore, the system configures nonlinear correction rules, including linear thresholds (e.g., 0.05) and nonlinear thresholds (e.g., 0.1), as well as corresponding correction factor calculation formulas.

[0075] Based on this, nonlinear correction rules for controlling waveform evolution are configured, specifically including:

[0076] For example, the relative wave height ratio is obtained by calculating the ratio of the instantaneous wave amplitude of a single gate response to the local water depth data at the target predicted section.

[0077] In this embodiment, the system reads the instantaneous wave amplitude A and the local water depth h of a certain gate section, and calculates the relative wave height ratio r = A / h. This serves as the physical basis for determining whether nonlinear correction needs to be initiated.

[0078] Accordingly, the relative wave height ratio is compared with preset linear and nonlinear thresholds, and the response state of each gate is divided into linear, weakly nonlinear, or strongly nonlinear regions to form a wave linearity interval.

[0079] Accordingly, the system compares the relative wave height ratio r with a threshold. If r < 0.05, it is in the linear region; if 0.05 ≤ r < 0.1, it is in the weak nonlinear region; if r ≥ 0.1, it is in the strong nonlinear region.

[0080] Furthermore, for response states in the weak or strong nonlinear regions, a piecewise nonlinear correction factor is generated, which includes amplitude enhancement, period stretching, or attenuation rate adjustment, to nonlinearly reconstruct the standard waveform in the single-gate response function library.

[0081] Accordingly, for the weakly nonlinear region, the system adopts a factor (1+γ) _1 ×r 2 Amplitude correction. For the strongly nonlinear region, the system not only corrects the amplitude but also utilizes formula T. _new =T×(1-γ _2 ×r) and β _new =β×(1+γ_3 ×r) adjusts the period and decay rate to reconstruct a nonlinear waveform that conforms to the dynamics of large waves. In other words, T _new β _new Characterizing the adjusted period and decay rate, γ _1 γ _2 γ _3 This is the corresponding correction factor.

[0082] Step S302: Based on the propagation time lag, analyze the phase characteristics of each gate's arrival at the target prediction section, and combine this with nonlinear correction rules to generate multi-gate convolutional response kernel parameters adapted to the current operating conditions; wherein, analyzing the phase characteristics of each gate's arrival at the target prediction section can be performed in the following manner:

[0083] Optionally, based on the navigation channel depth and topographic conditions in the hydrological evolution characteristics, the characteristic oscillation period of the target prediction section is calculated, and the arrival phase angle data of each effective historical executed gate and effective future planned gate when reaching the target prediction section is calculated in combination with the propagation time lag.

[0084] In this embodiment, the system calculates the characteristic oscillation period T based on the cross-sectional water depth h and the terrain. _n For each gate pass, the system calculates the phase angle φ = ((tt) upon reaching the target section. _start -τ) / T _n τ × 360 degrees is used to describe the phase state when the wave arrives. Alternatively, τ is the propagation time lag, t _start This refers to the actual opening time of a single gate.

[0085] Optionally, gates with similar phases are grouped into at least one phase cluster, and the number of gates and the degree of phase concentration within each phase cluster are counted. In other words, based on the arrival phase angle data, gates with similar phases are grouped into at least one phase cluster, and the number of gates and the degree of phase concentration within each phase cluster are counted.

[0086] Furthermore, the system sets a phase difference threshold (e.g., 30 degrees). The system iterates through the arrival phase angles of all gates and groups gates with phase differences less than this threshold into a phase cluster. The system then counts the number of gates N within each cluster. _k and phase standard deviation σ _φ , used to quantify the degree of phase concentration. Where N _k The k-phase cluster number in the sequence.

[0087] Optionally, based on the number of gates and the degree of phase concentration, a phase enhancement coefficient is generated for each phase cluster, and this (phase enhancement coefficient) is used as part of the kernel parameters of the multi-gate convolution response to perform weighted enhancement on the fluctuation response belonging to the same phase cluster.

[0088] Accordingly, the system calculates the phase enhancement coefficient C. _phase The calculation formula can be:

[0089] C _phase =1+α×(N _k / N _0 )×(1-σ _φ / σ _max ).

[0090] Where α is the intensity coefficient, N _0 As the baseline quantity, σ _max This represents the upper limit of the phase standard deviation.

[0091] It should be understood that this coefficient, as part of the convolution response kernel, is used to amplify the amplitude of wave clusters with highly overlapping phases, simulating a resonance effect.

[0092] Step S303: Under the control of the multi-gate convolution response kernel parameters, the single-gate response function library is called to perform convolution operations and nonlinear synthesis on each valid historical executed gate and valid future planned gate to obtain the superimposed fluctuation components.

[0093] In this embodiment, the system processes the waveforms of each gate using the generated convolutional response kernel parameters (including phase enhancement coefficients and nonlinear correction factors). The individual waveforms are nonlinearly reconstructed and then weighted and superimposed according to phase clusters. The synthesized sequence is the superimposed wave component.

[0094] Example 4: An exemplary scheme describing the online adaptive adjustment process of the prediction parameters, wherein the natural decay time parameter of water level fluctuations is dynamically updated based on prediction feedback, and the update process includes:

[0095] Accordingly, measured water level data of the cross-section along the navigation channel are obtained, and time-series alignment and numerical comparison are performed with the corresponding time-linked water level prediction process to generate a prediction error sequence.

[0096] Accordingly, the system collects the measured water level Z in real time. _obs , and the predicted water level Z _pred Alignment. The system calculates the error sequence e(t) = Z. _pred (t)-Z _obs (t).

[0097] Furthermore, statistical analysis is performed on the prediction error sequence to generate error assessment data. Based on the deviation between the error assessment data and the preset target error level, the natural decay time parameter of water level fluctuation is corrected online.

[0098] In this embodiment, the system calculates the mean square error E of the error sequence. _current Using the update formula T _d_new =T_d_old +k×(E _target -E _current For the decay time parameter T _d Corrections are made. For example, when the error is too large and indicates that historical residual effects have not been eliminated, the system automatically increases T. _d value.

[0099] In other words, T _d_new T _d_old These represent T before and after the update, respectively. _d k represents the correction factor, E _target This represents the preset target error level.

[0100] Based on this, the effective calculation time window for the next moment is reconstructed by using the corrected water level fluctuation natural decay time parameter, so as to iteratively optimize the subsequent bidirectional time series screening process.

[0101] Accordingly, the system will update T _d (For example, 4.2 hours) This is immediately applied to the calculation at the next moment, redefining the length of the effective calculation time window. This ensures that the selected historical gate set always matches the current hydrodynamic memory characteristics.

[0102] Example 5 describes an optional method for quantitatively assessing the resonance effect caused by high-frequency operation of multiple locks and for automatically identifying key sensitive areas for navigation. This solves the problem of accurately locating high-risk areas from the entire approach channel under complex hydrodynamic conditions, providing a basis for subsequent early warning. Optionally, the method includes:

[0103] Step S501: Calculate the time interval between adjacent gates in the valid historical executed gates and valid future planned gates.

[0104] Accordingly, the system extracts the time-sorted sequence of gate opening times from the multi-gate operation sequence data. For any two adjacent gates i and j in the sequence (with opening times t and j respectively), _i and t _j The system calculates the absolute value of its time difference Δt. _ij =|t _i -t _j This calculation not only applies to adjacent gate pairs, but can also be extended to any pair of gate pairs within a certain time window to capture non-adjacent but periodically regular operating rhythms.

[0105] Step S502: Obtain the characteristic oscillation period of the pilot channel and calculate the period deviation between the gate interval and an integer multiple of the characteristic oscillation period.

[0106] In this embodiment, the system reads the characteristic oscillation period T of each section of the pilot channel. _nFurthermore, the system analyzes the gate interval Δt. _ij With characteristic period T _n The matching relationship is determined by the system. Specifically, the system calculates the minimum difference between the interval and an integer multiple of the feature period (Я=1, 2, 3...). The calculation formula can be expressed as:

[0107] Deviation=min(|Δt _ij -Я×T _n |). The smaller this deviation, the closer the excitation frequency is to the natural frequency of the pilotway, and the greater the possibility of resonance. In the above formula, Я represents a positive integer coefficient, and Deviation represents the period deviation.

[0108] Step S503: Construct a resonance risk index based on the period deviation to reflect the degree of matching between the excitation period and the inherent period, and evaluate the resonance amplification effect of the pilot channel under multiple gate actions.

[0109] Accordingly, the system constructs a resonance risk index R. _res To quantify the degree of matching, the system introduces an excitation cycle similarity function S. _ij The calculation formula can be described as follows:

[0110] S _ij =max _ Я(exp(-|Δt _ij -Я×T _n | / σ _t ));

[0111] Where, σ _t This is a time-scale parameter that controls the rate of decay. The formula shows that when the time interval is close to an integer multiple of the characteristic period, S... _ij Approaching 1.

[0112] Furthermore, the system combines the gate amplitude with S _ij The weighted average is then used to calculate the comprehensive resonance risk index of the cross-section, as follows:

[0113] R _res (x)=∑(S _ij ×(A _i +A _j ) / (2×h));

[0114] In other words, A _i A _j These represent the fluctuation amplitudes of the i-th and j-th gates, respectively.

[0115] It should be understood that this indicator comprehensively reflects the frequency matching degree and excitation intensity; the higher the value, the stronger the resonance amplification effect.

[0116] Building on this, the method also includes identifying key areas that have a significant impact on navigation safety, including:

[0117] Optionally, the maximum wave amplitude and spatial water level gradient of each section along the navigation channel are extracted based on the linkage water level prediction process, and the maximum relative wave height and wave steepness data are calculated by combining local water depth data.

[0118] In this embodiment, the system utilizes the linked water level prediction process Z(x,t) to extract the maximum water level fluctuation amplitude A at each cross-section throughout the entire time period. _max (x). Further calculation of the maximum relative wave height r _max (x)=A _max (x) / h(x); where h(x) represents the water depth at section x.

[0119] Furthermore, the system calculates the spatial difference of water level between adjacent sections, i.e., the spatial water level gradient Slope(x,t)=(Z(x+dx,t)-Z(x,t)) / dx, and extracts its maximum value as the fluctuation steepness data.

[0120] In this step, the above parameters characterize the intensity and steepness of the fluctuation, respectively.

[0121] Optionally, the maximum relative wave height, wave steepness data and resonance risk indicators can be weighted and combined to construct a sensitive section assessment index that reflects shallow water effects, waveform structure and resonance probability.

[0122] For example, the system constructs a sensitive section evaluation index S _sens (x). This index is a weighted composite of multiple physical factors, and the calculation formula is:

[0123] S _sens (x)=w _1 ×r _max (x)+w _2 ×Slope _max (x)+w _3 ×R _res (x);

[0124] Where w _1 w _2 w _3 The preset weighting coefficients (e.g., 0.4, 0.3, 0.3 respectively) mean that the Slope _max (x) corresponds to the steepness of the fluctuation.

[0125] Based on this, this index can reflect the cross-section's effects due to shallow water (relatively large wave height), waveform structure (large steepness), and resonance probability (R). _res (High) and the comprehensive risks faced.

[0126] Optionally, a set of sensitive sections can be selected based on the sensitive section assessment indicators, and sensitive weight configuration data with high risk weights can be generated for the sections in the sensitive section set to enhance the risk expression of the area in subsequent safety assessments.

[0127] For example, the system sets a sensitivity threshold θ _s When the evaluation index S of a certain section _sens (x) When the threshold is exceeded, it is marked as a sensitive section and included in the sensitive section set. For sections within the set, the system generates sensitive weight configuration data, for example, assigning a weighting factor greater than 1 (such as 1.5), while assigning a factor of 1.0 to non-sensitive sections. This configuration data will be used in the subsequent calculation of the air traffic safety index, so that high-risk areas are highlighted in the global assessment.

[0128] Example 6 describes how to construct an indicator system for quantifying navigation risk levels. This system transforms physical quantities such as water depth, resonance, and wave patterns into a unified dimensionless index, directly serving navigation decision-making. Specifically, the construction process is as follows:

[0129] Step S601: Calculate the difference between the predicted water depth and the minimum safe water depth stipulated for navigation in the linkage water level prediction process to obtain the water depth safety margin. Then, use a nonlinear mapping function to convert the water depth safety margin into a water depth risk component.

[0130] In this embodiment, the system calculates and predicts the water depth H. _pred (x, t) = Z(x, t) - Z _bed (x), where Z _bed Let H be the riverbed elevation. Next, calculate the water depth safety margin M(x, t) = H. _pred (x, t) - H _min_safe , where H _min_safe The minimum safe water depth is defined. The system uses a nonlinear function to convert the margin into a water depth risk component R. _depth This is used to highlight the risk of low water levels.

[0131] Alternatively, its mapping formula is: R _depth (x, t) = exp(-M(x, t) / M _0 ), where M _0 This is the characteristic margin constant. This formula causes the risk component to increase exponentially and sharply when the safety margin M approaches 0 or becomes negative, serving as a strong warning.

[0132] Step S602: Integrate the resonance risk index and volatility steepness data, and use sensitive weight configuration data to weight and amplify the sections located in the sensitive section set to generate structural risk components.

[0133] In this embodiment, the system normalizes the fluctuation steepness data to R. _steep (x, t) normalizes the resonance risk index to R _res_norm (x). Next, the data W is configured using sensitive weights. _sens (x) Weight the two factors together to generate the structural risk component R. _struct (x, t), its calculation formula can be:

[0134] R _struct (x, t) = W _sens (x)×(w _a ×R _res_norm (x)+w _b ×R _steep (x, t)). This component focuses on reflecting dynamic risks beyond simple insufficient water depth, such as difficulty in ship maneuvering or the risk of grounding. Above, w _a w _b These are the corresponding weighting coefficients.

[0135] Step S603: Weight the water depth risk component and the structural risk component to generate a navigation safety index covering the prediction period and the space along the pilotway.

[0136] For example, the system synthesizes the two components mentioned above to obtain the navigation safety index J(x,t)=λ _1 ×R _depth (x, t) + λ _2 ×R _struct (x, t). Where λ _1 and λ _2 λ is the weighting coefficient. _1 Larger values ​​(e.g., 0.7) are used to emphasize the dominant role of water depth safety. Above, J(x, t) is a matrix covering the entire spatiotemporal domain; larger values ​​represent higher navigation risks, providing a unified quantitative benchmark for subsequent early warning classification and scheduling optimization.

[0137] Example 7 describes the closed-loop decision-making process of optimizing the scheduling scheme based on prediction results. It demonstrates how the present invention upgrades a predictive system to a proactive control system. As an example, this example performs the following steps:

[0138] Step S701: With minimizing the cumulative value or extreme value of the navigation safety index within the prediction period as the optimization objective, and under the premise of satisfying the navigation efficiency constraint, construct an optimization task for the future lock operation plan.

[0139] In this embodiment, the system constructs an optimization objective function J. _total J is defined as the weighted sum of the navigation safety indices of all cross-sections along the route during the prediction period. _total =∑ _x∑ _t (J(x,t)), or take the maximum value J _max =max(J(x,t)). This is used to make J... _total or J _max minimize.

[0140] Furthermore, constraints are set, including: the adjustment range of gate opening time (e.g., the original scheduled time + / - 30 minutes), and the minimum safe interval between adjacent gates (e.g., not less than 40 minutes), so that navigation efficiency is not excessively affected.

[0141] Step S702: Use the optimization task to iteratively search and adjust the planned opening time or operation interval of the lock to be executed in the future lock operation plan, and simultaneously trigger the linkage water level prediction process and the update calculation of the navigation safety index.

[0142] Specifically, the system employs heuristic search algorithms (such as hill climbing or genetic algorithms) to solve the aforementioned optimization task. The system generates candidate gate opening time series; utilizes multi-gate convolutional response kernel parameters to quickly recalculate the superimposed fluctuation components under the candidate sequence; updates the linked water level prediction process and navigation safety index; and calculates a new objective function value. If the new objective function value is superior, the candidate sequence is retained. This process is rapidly iterated in the computer until the optimal solution is found or the termination condition is met.

[0143] Step S703: Obtain a scheduling scheme that meets the optimization objective and generate gate adjustment suggestions that include recommended opening times and draft control requirements.

[0144] Specifically, after the iterative search is completed, the system outputs the optimal gate opening schedule that minimizes the risk to navigation. Based on this, the system generates specific natural language suggestions, such as suggesting that gate number 2 be delayed by 15 minutes to avoid the resonance peak caused by gate number 1.

[0145] Furthermore, based on the optimized lowest predicted water level, the system reverse-engineers the maximum permissible draft of the vessel and generates draft control recommendations (e.g., the maximum draft of the vessel during this period is limited to 3.5 meters), thus achieving a complete closed loop from prediction to decision-making.

[0146] According to one aspect of this application, it also includes weighted synthesis of the water depth risk component and the structural risk component to generate a navigation safety index covering the prediction period and the space along the pilotway.

[0147] Specifically, the system establishes risk grading standards to translate the generated air traffic safety index into specific early warning actions. The system presets multiple level thresholds, such as a focus threshold T. _watch and warning threshold T _warnThe system iterates through each security index J(x, t) in the spatiotemporal matrix. When J(x, t) is less than T... _watch When T is active, mark it as safe; when T is active, mark it as safe. _watch ≤J(x,t) <T _warn When J(x, t) ≥ T, it is marked as a warning level; when J(x, t) ≥ T _warn When this occurs, it is marked as a severe warning level.

[0148] For areas identified as sensitive sections, the system will forcibly upgrade their risk level (for example, even if the index is in the warning range, it will be directly upgraded to a severe warning) to ensure the safety redundancy of critical areas.

[0149] According to another aspect of this application, since exceeding the limit at a single point or moment may be computational noise, the system executes a continuous segment identification algorithm. On the time axis, if a cross section is in a severe warning state for more than 3 consecutive time steps (e.g., 15 minutes), it is merged into a risk time window. On the spatial axis, if multiple adjacent cross sections are simultaneously in a risk state, they are merged into a risk river segment.

[0150] Based on this, the system generates structured early warning information packages, with the format of time interval-spatial interval-risk level-dominant risk type (such as insufficient water depth or resonance risk). This approach significantly reduces scattered alarms and provides dispatchers with a comprehensive risk overview.

[0151] According to another aspect of this application, the multi-gate convolution response kernel parameters (including phase cluster structure, nonlinear correction rule, phase enhancement coefficient, and adaptive decay time T) _d These parameters are core system assets. The system packages these parameters, adds a timestamp and version number (e.g., Kernel_V20231027_1400), and persistently stores them in the database. During rapid optimization iterations, the optimization algorithm directly calls the latest version of the parameter package for millisecond-level recalculation, without retraining or recalibrating. This ensures consistency of the core hydrodynamic parameters used throughout the entire process from prediction and early warning to optimization.

[0152] In some embodiments, parts of the methods of the present invention can also be implemented in the following manner:

[0153] Accordingly, the following information is obtained in real time from the lock control system / scheduling system: historical lock information, i.e., the actual opening time, working head, and operation mode of the multi-line locks in the pilot channel within a period of time (e.g., 4 hours) before the current time; and planned lock information, i.e. the expected locks within the next few hours (e.g., 4-16 hours) provided by the scheduling plan, as well as their working head and estimated running time.

[0154] Based on this, historical gate sequence and predicted gate sequence are formed.

[0155] Furthermore, the water level prediction of the control section is carried out by using the existing water level prediction model of the pilot channel entrance area (including hydraulic calculation model and / or data-driven model) to obtain the water level process of the pilot channel control section within the future prediction window.

[0156] Furthermore, the water level fluctuation is considered as a linear superposition of responses excited by multiple independent lock discharge processes, and the superposition result constitutes the total fluctuation process. This linear superposition model is applicable to working conditions with shallow water characteristics, long pilot channel structure and well-defined gate area control section, and satisfies the long wave propagation condition.

[0157] Furthermore, the fluctuation process of a single lock session is calculated (by calling an existing module). The running time, effective head, and initial water level of the pilot channel for each lock session are input into the existing rapid calculation module for fluctuations along the restricted long pilot channel to obtain the fluctuation process of that lock session at any point along the pilot channel.

[0158] Based on this, multi-gate time shifting and superposition processing is performed. For historical and predicted gate openings, the following steps are taken: Based on the natural decay time of the fluctuations, the portion of the fluctuations that still has an impact on the current or future time is extracted. The fluctuation process corresponding to each gate opening is shifted along the time axis according to its discharge time, so that all fluctuation processes use the prediction window as a unified reference. At any position x and any time t, the fluctuation processes of all effective gate openings are superimposed to obtain the composite fluctuation component caused by all gate openings, i.e.:

[0159] ;

[0160] Where i is the gate index.

[0161] Furthermore, the pilot channel water level linkage prediction uses the predicted water level results of the control section as downstream boundary conditions. By superimposing these conditions, the predicted water level process for each section along the channel can be obtained, i.e.:

[0162]

[0163] Based on this, the prediction method further includes: reading T based on the ship passage plan and the duration of each stage of lock operation. d The lock's discharge time is one hour later; T is the time T before the current moment. d All water release operations within the hour are read and their fluctuation contribution at time t is summed; for expected water release operations within the forecast period, it is determined whether they have already affected time t, and the summation calculation is performed; historical fluctuations and forecast fluctuations are summed to form a complete fluctuation term; the fluctuation term is superimposed on the water level benchmark value of the control section to obtain the final water level prediction; as time progresses, the scheduling information and water release time are dynamically corrected, and the above steps are repeated to complete the rolling prediction.

[0164] Furthermore, the navigation safety margin assessment and early warning, based on the design draft requirements of each section, minimum safe water depth, and early warning level thresholds, utilizes the water level prediction process to identify risks, which can further trigger early warnings of insufficient water depth, suggestions to limit vessel draft, and suggestions to adjust lock schedules.

[0165] As an example, historical monitoring data shows that the inherent natural decay time of fluctuations during lock operation is 4 hours. This means that the water level fluctuations in the navigation channel are actually the superposition of fluctuations caused by lock discharge within the 4 hours preceding the current moment and fluctuations caused by lock discharge within the predicted period following the current moment. The lock discharge valve actuation time within the 4 hours preceding the current moment can be obtained from real-time centralized control data, while the actuation time of the lock discharge valve within the predicted period needs to be determined based on the actual lock operation schedule and other factors.

[0166] Optionally, if the current time is 0, then the subsequent predicted time period is t. Within the 4 hours prior to the current time, Lock No. 1 operated n times, and the time difference between the corresponding valve opening times and the current time is Δt. e1 , Δt e2 , …, Δt en Lock No. 2 has been operated m times, and the time difference between the corresponding valve opening time and the current time is Δt. s1 , Δt s2 , …, Δt sm The prediction period is that Lock No. 1 will operate k times, and the time difference between the corresponding valve opening time and the current time is ΔT. e1 ΔT e2 , …, ΔT ek Lock No. 2 will operate p times, and the time difference between the corresponding valve opening time and the current time is ΔT. s1 ΔT s2 , …, ΔT sp .

[0167] Optionally, the water level prediction at the control section can be achieved by using an existing water level prediction model for the navigation channel entrance area to obtain the water level process at the navigation channel control section. The specific calculation method can be as follows:

[0168] ;

[0169] in, This refers to the outflow from the hub.

[0170] In the example, the formula for calculating the single lock fluctuation process of Lock No. 1 in the pilot channel can be:

[0171] ;

[0172] Formula for calculating the single lock fluctuation process of Lock No. 2 in the pilot channel:

[0173] ;

[0174] Where 6 and 0.3 can be correction coefficients / empirical parameters, and φ is the correction term.

[0175] Furthermore, the operating time, effective head, and initial water level of the approach channel for each lock operation are input into the existing rapid calculation module for fluctuations along the restricted long approach channel to obtain the fluctuation process η at any point along the approach channel L for that lock operation. i (x, t).

[0176] Based on this, the time shift and superposition of multiple lock operations are used to calculate and predict the lock discharge fluctuation process within the predicted time period.

[0177] When t-ΔT ei <0 (ΔT) ei H represents the time difference between the valve opening time corresponding to lock No. 1 within the predicted time period and the current time (where i is the lock index). _sei (x, t) represents the effective future planning gate, that is:

[0178]

[0179] When t-ΔT ei When ≥0,

[0180] ;

[0181] When t-ΔT si <0 (ΔT) si H represents the effective future planned number of locks 2 when the valve opening time corresponding to lock 2 is within the predicted time period and the current time (where i is the lock number index). _pei (x, t), that is:

[0182]

[0183] t-ΔT si When ≥0,

[0184] ;

[0185] Where, Δt si The time difference between the valve opening time and the current time within 4 hours before the current time of Lock No. 2, i is the lock index, 50, 0.3, and 6 are correction coefficients / empirical parameters, and π / 0.3 is the correction term.

[0186] Based on this, the fluctuation process is calculated (by summing the results of the previous two steps).

[0187] ;

[0188]

[0189] Above, H _pei (x, t), H _psi (x, t) represent the valid historical lock operations performed within the previous 4 hours of the current time for locks 1 and 2, respectively.

[0190] As time progresses, the accuracy of determining the timing of lock discharge is improved, the time difference data in the above steps are corrected, and the above steps are repeated to improve prediction accuracy.

[0191] For example, the superposition effect is illustrated by the measured fluctuation data from 16:00 on April 11, 2023 to 4:00 on April 12, 2023. During this period, No. 1 lock discharged water 7 times and No. 2 lock discharged water 12 times. This period is the time when the locks are frequently in operation and the period with the most complex fluctuation process.

[0192] Based on the wave decay characteristics described above, when studying the actual water level fluctuation process, the influence of fluctuations formed by multiple releases from the lock in the preceding period that have not yet decayed at the current moment must be considered. The fluctuations will decay approximately 4 hours after they occur. Therefore, this study mainly focuses on the fluctuation process after 20:00 on April 11th. Fluctuations occurring before 16:00 had already decayed by 20:00 and do not affect fluctuations after 20:00. A comparison of the measured and calculated fluctuations in the approach channel near the pier downstream of Lock No. 1 and at the middle section of the approach channel during this period shows that the absolute error of wave height is small in most operating conditions, and the predicted results are in good agreement with the measured results.

[0193] In another example, some methods of the present invention can also be carried out in the following manner:

[0194] Given a predicted start time, select from historical lock operation records data that are T years or more before the current predicted time. _d Historical lock discharge events within a given time period are selected, and planned lock discharge events within the predicted time period after the current predicted time are filtered from the future lock operation plan data. All events are integrated into multi-lock operation sequence data in chronological order, where each record contains information such as gate number, opening time, operation mode, and working head.

[0195] Furthermore, for each gate record in the multi-gate operation sequence, its working head and operating mode are used as indexes to call the single-gate response function library to obtain the corresponding standard single-gate response sequence. The standard response is then scaled and phase-corrected according to the actual water depth and the operating conditions of the hub to obtain the condition-corrected single-gate response sequence for that gate at each cross section.

[0196] At each prediction time, for gates within the time range covered by the multi-gate time window parameters, a time shift operation is performed to shift the response sequence of each gate relative to its opening time, aligning it with the time axis of the prediction time, thus forming a set of time-shifted response sequences with the prediction time as a reference.

[0197] The process of obtaining the multi-lock time window can be as follows: extract information such as the opening time, operation mode and working head of each lock from historical lock operation records and future lock operation plans, combine them with the background field of water level at the control section, determine the natural decay time parameter used to intercept the effective historical lock influence, and form a multi-lock time window.

[0198] Furthermore, by utilizing the topographic and water depth data of the pilot channel and the long-wave period information contained in the single lock response, the characteristic oscillation period of each cross section along the route is calculated. At each predicted time and on each cross section, for each lock in the time-shifted response sequence set, based on the characteristic oscillation period and the lock opening time, the arrival phase angle data corresponding to that cross section and that time is calculated to obtain the lock-phase correspondence.

[0199] Furthermore, based on the arrival phase angle data, gates with phase differences less than a preset phase difference threshold are grouped into the same phase cluster, forming phase cluster structure data to characterize different phase combinations. Statistical characteristics such as the number of gates and the degree of phase concentration are calculated for each phase cluster.

[0200] Optionally, for each lock, each section, and each predicted time, based on the time-shifted response sequence set and the topography and water depth data of the pilot channel, the ratio of the instantaneous wave amplitude to the water depth at that section and lock is calculated to obtain the relative wave height ratio data.

[0201] Optionally, based on the comparison of the relative wave height ratio data with preset linear interval thresholds and weak nonlinear interval thresholds, the response of each gate is divided into a linear region, a weak nonlinear region, and a strong nonlinear region, and different correction strategies are adopted for each region. That is, in the linear region, the original single gate response is maintained; in the weak nonlinear region, a secondary enhancement factor is applied to the response amplitude; and in the strong nonlinear region, the response amplitude, equivalent period, and equivalent attenuation coefficient are corrected simultaneously to generate a continuous nonlinear corrected single gate response sequence in each interval.

[0202] Accordingly, for each phase cluster, the corresponding phase enhancement coefficient data is calculated based on the number of gates in the cluster, the degree of phase concentration, and the relative position of the phases in the characteristic period. The phase enhancement coefficient data is used to reflect the amplification effect produced when multiple gates are superimposed in close phases.

[0203] Next, at each cross section and at each prediction time, the nonlinear modified single gate response sequences belonging to the same phase cluster are added together to obtain the intra-cluster wave superposition sequence of that phase cluster. Then, the intra-cluster wave superposition sequence is multiplied by the corresponding phase enhancement coefficient data to obtain the phase enhancement wave sequences of different phase clusters.

[0204] Based on this, the phase enhancement wave sequences of all phase clusters are superimposed at the same cross section and the same prediction time to obtain the multi-gate total wave superposition sequence at that cross section and time, which can be used to form multi-gate wave prediction data along the entire path and prediction period.

[0205] Accordingly, when real-time or recently updated monitoring water level data is available, the difference sequence between the predicted and measured values ​​is extracted from the predicted data of multiple gate fluctuations along the route and the monitoring water level sequence of the corresponding time period. Error statistics within a certain time window, such as mean square error or root mean square error, are calculated to form error assessment data for characterizing the current convolution stacking accuracy.

[0206] Furthermore, based on the deviation between the error assessment data and the target error level, a predefined update rule is used to adjust the natural decay time parameter T. _d Make fine adjustments, for example, using T _d_new =T _d_old +k_Td×(E _target -E _current An update strategy in the form of ) is used to obtain a new adaptive natural decay duration parameter.

[0207] Furthermore, the updated natural decay duration parameter, adaptive phase enhancement coefficient data, nonlinear correction parameter, and phase cluster construction rule are summarized into a set of persistent multi-gate convolutional response kernel parameters for repeated use and continuous iteration in subsequent predictions.

[0208] In summary, this invention employs a phase-sensing and nonlinear correction convolutional superposition technique. By analyzing the phase characteristics of each gate's arrival at the prediction section and constructing a phase cluster, the scheme introduces a phase enhancement coefficient to quantify the amplification effect when multiple waves are in phase. Furthermore, based on the relative wave height ratio, piecewise nonlinear corrections are applied to the waveform (such as correcting wave amplitude and equivalent period). This overcomes the limitations of traditional linear superposition, captures shallow water nonlinear effects and resonance risks, and improves prediction accuracy under complex conditions.

[0209] In addition, a parameter adaptive update mechanism based on prediction feedback was constructed. By calculating the error statistics between the predicted water level and the measured water level in real time, the system can correct the natural decay time parameter (T) online. _dBased on this, the model can sense changes in environmental boundaries such as water level and flow velocity, dynamically adjust the length of the effective calculation time window, eliminate the cumulative error caused by fixed parameters, and ensure the long-term robustness of the system under different seasons and water conditions.

[0210] Based on this, a closed-loop optimization decision-making model based on the navigation safety index was established. The scheme not only outputs predicted water levels but also further integrates water depth margin, resonance risk, and wave steepness to construct a quantified safety index. Using this index as the objective function, the optimal future gate start time is searched in reverse. This achieves a leap from passive prediction to active control, solving the problem of balancing navigation safety and scheduling efficiency.

[0211] 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 predicting navigation channel water levels in conjunction with navigation channel water levels, characterized in that, include: Analyze the hydrological evolution characteristics of the downstream control section of the navigation channel and construct a background field of the control section water level describing the impact of the hub's discharge on the downstream water level benchmark. The memory effect of hydrodynamics in the navigation channel is quantified, the natural decay time parameter of water level fluctuation is determined, and an effective calculation time window covering historical retrospective periods and future prediction periods is constructed based on the current prediction time. By utilizing the effective calculation time window, bidirectional time-series filtering is performed on the multi-source lock operation information to extract the effective historical lock operations and effective future planned lock operations that fall within the effective calculation time window; For each of the selected valid historical gate openings and valid future planned gate openings, time-shift alignment and amplitude accumulation based on wave propagation characteristics are performed to generate superimposed wave components that reflect the disturbance effect of multiple gate openings. The superimposed wave components are coupled to the background field of the water level at the control section to generate a linked water level prediction process for each section along the navigation channel. In the second calculation mode, the superimposed wave component reflecting the disturbance effect of multiple gate discharges is generated through a convolutional superposition process of multi-gate phase sensing and nonlinear correction. This process includes: constructing a single-gate response function library containing standard waveform characteristics under different operating conditions, configuring nonlinear correction rules for controlling waveform evolution; analyzing the phase characteristics of each gate reaching the target prediction section based on the propagation time lag, and generating multi-gate convolutional response kernel parameters adapted to the current operating conditions by combining the nonlinear correction rules; under the control of the multi-gate convolutional response kernel parameters, calling the single-gate response function library to perform convolution operations and nonlinear synthesis on each effective historical executed gate and effective future planned gate to obtain the superimposed wave component. The natural attenuation time parameter of water level fluctuation is dynamically updated based on prediction feedback. The update process includes: acquiring measured water level data along the pilot channel section, aligning and comparing the data with the corresponding time-series water level prediction process to generate a prediction error sequence; performing statistical analysis on the prediction error sequence to generate error assessment data; and correcting the natural attenuation time parameter of water level fluctuation online based on the deviation between the error assessment data and the preset target error level; and using the corrected natural attenuation time parameter of water level fluctuation to reconstruct the effective calculation time window for the next time moment and iteratively optimizing the subsequent bidirectional time series filtering process.

2. The method according to claim 1, characterized in that, The lock operation information consists of historical monitoring data from the hub centralized control and monitoring system and future scheduling data from the navigation scheduling plan system. Historical monitoring data includes the actual opening time, measured working head, and operating mode of the gates that have been executed. The fluctuation triggering origin of the valid historical gates that have been executed is determined through two-way time series screening. Future scheduling data includes the planned opening time of the gates to be executed, the expected working head, and scheduling constraints. The potential fluctuation time series of effective future planned gates is deduced through two-way time series screening.

3. The method according to claim 1, characterized in that, Perform time-shift alignment and amplitude accumulation based on wave propagation characteristics, including: Calculate the spatial transmission distance from the location of the lock wave source to any target section along the pilot channel; Based on the spatial transmission distance, the propagation time lag of the wave in the pilot channel is calculated, and the wave response of the effective historical executed gates and the effective future planned gates is phase corrected to achieve time shift alignment. Based on the spatial transmission distance, the spatial energy dissipation of the wave propagation is quantified. The intensity attenuation ratio of each wave response when it reaches the target prediction section is determined by the spatial energy dissipation, and the amplitude accumulation is completed.

4. The method according to claim 3, characterized in that, In the first calculation mode, a superimposed wave component reflecting the disturbance effect of multiple gate discharges is generated, specifically including: Call the preset single-wave basic waveform and use the decay function based on the exponential law to map the spatial energy dissipation into a linear decay coefficient; Based on the propagation time lag, a rigid time shift is performed on the basic waveform of the single fluctuation of each gate, and the amplitude scaling is performed on the shifted waveform based on the linear attenuation coefficient to obtain the linear response sequence of each gate. For the linear response sequences of all valid historical executed gates and valid future planned gates, perform a linear superposition operation to obtain the superimposed fluctuation components.

5. The method according to claim 1, characterized in that, Analyze the phase characteristics of each gate's arrival at the target prediction section, including: Based on the navigation channel water depth and topographic conditions in the hydrological evolution characteristics, the characteristic oscillation period of the target prediction section is calculated. Combined with the propagation time lag, the arrival phase angle data of each effective historical executed gate and effective future planned gate when reaching the target prediction section are calculated. Gates with similar phases are grouped into at least one phase cluster, and the number of gates and the degree of phase concentration within each phase cluster are counted. Based on the number of gates and the degree of phase concentration, a phase enhancement coefficient is generated for each phase cluster. This coefficient is then used as part of the kernel parameters of the multi-gate convolution response to perform weighted enhancement on the fluctuation response belonging to the same phase cluster.

6. The method according to claim 1, characterized in that, Configure nonlinear correction rules for controlling waveform evolution, including: The relative wave height ratio is obtained by calculating the ratio of the instantaneous wave amplitude of a single gate response to the local water depth data at the target predicted section. The relative wave height ratio is compared with the preset linear and nonlinear thresholds, and the response state of each gate is divided into a linear region, a weak nonlinear region, or a strong nonlinear region to form a wave linearity range. For response states in the weak or strong nonlinear regions, a piecewise nonlinear correction factor is generated, which includes amplitude enhancement, period stretching, or attenuation rate adjustment, to nonlinearly reconstruct the standard waveform in the single-gate response function library.

7. The method according to claim 1, characterized in that, The method also includes identifying resonance risks based on the lock's operating frequency, specifically: Calculate the time interval between adjacent gates in the valid historical executed gates and valid future planned gates; Obtain the characteristic oscillation period of the pilot channel and calculate the period deviation between the gate interval and an integer multiple of the characteristic oscillation period; Based on the period deviation, a resonance risk index reflecting the degree of matching between the excitation period and the inherent period is constructed to assess the resonance amplification effect of the pilotway under multiple gate actions.

8. The method according to claim 7, characterized in that, The method also includes identifying key areas that have a significant impact on navigation safety, specifically: Based on the linkage water level prediction process, the maximum wave amplitude and spatial water level gradient of each section along the navigation channel are extracted, and the maximum relative wave height and wave steepness data are calculated by combining local water depth data respectively. By weighting and combining the maximum relative wave height, wave steepness data and resonance risk indicators, a sensitive section assessment index reflecting shallow water effect, waveform structure and resonance possibility is constructed. A set of sensitive sections is selected based on the sensitive section assessment indicators. Sensitive weight configuration data with high risk weights is generated for the sections in the sensitive section set to enhance the risk expression of the area in subsequent safety assessments.

Images