Dredging operation intelligent decision optimization method based on multi-source data fusion

Abstract

The invention relates to the technical field of dredging construction, and particularly discloses a dredging operation intelligent decision optimization method based on multi-source data fusion, comprising: collecting and preprocessing multi-source data, the multi-source data comprising general data and ship type exclusive data for different dredger types; and based on the dredger type, constructing and executing a ship type exclusive dual-drive coupling sub-model, fusing the multi-source data, and executing ship type exclusive spectrum coupling analysis and ship type exclusive active anomaly identification. According to the invention, by constructing the ship type exclusive dual-drive coupling sub-model and the data system, the combination of frame unification and ship type specialization is realized, so that the decision system not only can accurately match the unique physical mechanism of the specific ship type, but also can operate under a unified intelligent frame.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of dredging construction, in particular to a dredging operation intelligent decision optimization method based on multi-source data fusion. BACKGROUND

[0002] Dredging operation is an important project to ensure navigation and build land, and its operation decision is crucial in complex marine environment. The traditional dredging operation highly depends on manual experience or static construction scheme, which is difficult to respond to the dynamic changes in the environment in real time, resulting in low operation efficiency and high energy consumption. Therefore, the industry has been seeking to introduce an intelligent decision model, but this immediately encounters a conflict between universality and accuracy in practice: different types of dredging ships (such as drag suction and cutter suction) have completely different physical characteristics and operation mechanisms, and an accurate decision model must be deeply bound to the characteristics of the ship type; however, a too specialized model loses its universality and portability. The existing technology either adopts a one-size-fits-all universal model, but lacks ship type specificity and leads to inaccurate decisions, or adopts a highly customized single ship type model, but the development cost is high and a unified intelligent decision framework cannot be formed.

[0003] In order to make the model intelligent, it must be able to evolve automatically according to real-time multi-source data, that is, to realize the bidirectional driving and fusion of data-model. However, this ingenious adaptive ability also has its limitations:

[0004] On the one hand, when the system tries to respond to changes in the real-time environment by dynamically adjusting the weight, such as energy consumption priority or progress priority, if the environmental data fluctuates near the decision threshold, this high responsiveness will immediately cause the control instructions of the system to oscillate at a high frequency between the two strategies, causing equipment wear and even control failure, making responsiveness and stability incompatible;

[0005] On the other hand, when the system tries to automatically adjust its parameters according to the deviation between the model output and the sensor data, once encountering sensor failure or extreme working conditions, the adaptive mechanism will mistakenly consider these dirty data or abnormal deviations as features that the model needs to learn, and then incorrectly adjust the parameters to cater to these incorrect data, ultimately leading to the pollution and even collapse of the entire decision model. SUMMARY

[0006] The present application aims to at least partially solve one of the problems in the related art. To this end, the purpose of the present application is to propose a dredging operation intelligent decision optimization method based on multi-source data fusion to realize stable and efficient intelligent dredging operation.

[0007] To achieve the above-mentioned purpose, the first aspect of the present application proposes a dredging operation intelligent decision optimization method based on multi-source data fusion, comprising the following steps:

[0008] collect and pre-process multi-source data, the multi-source data including general data and ship type specific data for different dredging ship types;

[0009] based on the dredging ship type, build and execute a ship type specific dual drive coupling sub-model, fuse the multi-source data, and perform ship type specific spectral coupling analysis and ship type specific active anomaly identification;

[0010] based on the results of the multi-source data fusion, perform ship type specific dynamic weight adjustment and ship type specific safety constraints, and call a ship type specific pre-trained working condition library to generate a dredging operation decision;

[0011] perform ship type specific resilience supervision and verification on the execution effect of the dredging operation decision, and generate a ship type specific tamper-proof report;

[0012] based on the execution effect, iteratively optimize the ship type specific dual drive coupling sub-model.

[0013] To achieve the above purpose, the second aspect embodiment of the present application proposes a dredging operation intelligent decision optimization system based on multi-source data fusion, comprising:

[0014] a data acquisition module for collecting and pre-processing multi-source data, the multi-source data including general data and ship type specific data for different dredging ship types;

[0015] a model execution module for building and executing a ship type specific dual drive coupling sub-model based on the dredging ship type, fusing the multi-source data, and performing ship type specific spectral coupling analysis and ship type specific active anomaly identification;

[0016] a decision generation module for performing ship type specific dynamic weight adjustment and ship type specific safety constraints based on the results of the multi-source data fusion, and calling a ship type specific pre-trained working condition library to generate a dredging operation decision;

[0017] a supervision and verification module for performing ship type specific resilience supervision and verification on the execution effect of the dredging operation decision, and generating a ship type specific tamper-proof report;

[0018] an iterative optimization module for iteratively optimizing the ship type specific dual drive coupling sub-model based on the execution effect.

[0019] To achieve the above purpose, the third aspect embodiment of the present application proposes an electronic device comprising a memory, a processor and a computer program stored in the memory, wherein the computer program is executed by the processor to implement the above-mentioned dredging operation intelligent decision optimization method based on multi-source data fusion.

[0020] The dredging operation intelligent decision optimization method based on multi-source data fusion provided by the embodiment of the application realizes the combination of unified framework and ship type by constructing a double-drive coupled sub-model and a data system special for the ship type, so that the decision system can accurately match the unique physical mechanism of a specific ship type and can also run under a unified intelligent framework; more importantly, by introducing a strategy with hysteresis and smooth transition, the control instruction oscillation caused by data noise jitter is eliminated to a certain extent while the sensitive response to environmental changes is realized, so that the responsiveness and stability of the system are taken into account, and the smooth operation of the physical equipment is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 is a flowchart of the dredging operation intelligent decision optimization method based on multi-source data fusion provided by the application;

[0022] Figure 2 is an illustration of extracting wave crests and troughs in a sea wave image sequence in the dredging operation intelligent decision optimization method based on multi-source data fusion provided by the application;

[0023] Figure 3 is a frequency spectrum comparison diagram of equipment deviation and environmental disturbance in the dredging operation intelligent decision optimization method based on multi-source data fusion provided by the application;

[0024] Figure 4 is an average coupling strength factor change curve diagram in the dredging operation intelligent decision optimization method based on multi-source data fusion provided by the application;

[0025] Figure 5 is a construction process and index curve diagram of a sea wave comprehensive risk index in the dredging operation intelligent decision optimization method based on multi-source data fusion provided by the application;

[0026] Figure 6 is a comparison diagram of predicted values and measured values in the dredging operation intelligent decision optimization method based on multi-source data fusion provided by the application;

[0027] Figure 7 is a smooth transition function simulation diagram in the dredging operation intelligent decision optimization method based on multi-source data fusion provided by the application;

[0028] Figure 8 is a running structure diagram of the dredging operation intelligent decision optimization system based on multi-source data fusion provided by the application;

[0029] Figure 9 is a structure diagram of an electronic device provided by the application. DETAILED DESCRIPTION

[0030] Embodiments of the present application are described below in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference signs represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0031] The multi-source data fusion-based dredging operation intelligent decision optimization method, system and electronic device of the embodiments of the present application are described below with reference to the accompanying drawings.

[0032] Embodiment one:

[0033] The present embodiment provides a complete implementation process of a multi-source data fusion-based dredging operation intelligent decision optimization method. The method of the present embodiment is usually executed by a dredging operation intelligent decision optimization system deployed on a dredging ship locally or a remote cloud server. The system can include, in hardware, a multi-source data acquisition sensor, a data acquisition card, an edge computing gateway, a satellite or a 5G communication module for data communication on board; and in software, a data acquisition module, a model execution module, a decision generation module, a supervision verification module and an iterative optimization module.

[0034] As shown in Figure 1 , the complete method process of the present embodiment specifically includes the following five core steps:

[0035] Step S1: Acquisition and preprocessing of multi-source data

[0036] The first step is to comprehensively and in real time acquire all data related to dredging operations. In order to realize subsequent ship type-specific intelligent decision-making, the present application divides data sources into two categories: general data and ship type-specific data.

[0037] 1. Acquisition of general data: General data refers to basic environmental and state data that all types of dredging ships need to refer to during operation. In the present embodiment, the general data at least includes:

[0038] High-precision positioning data: Real-time three-dimensional coordinates (longitude, latitude, elevation) and heading, speed information of the dredging ship are obtained using differential global satellite navigation systems such as RTK-GNSS, including GPS, BeiDou, etc., which is the basis for operation path planning, earthwork calculation and construction area compliance checking.

[0039] Hydrological and geomorphological data: Real-time underwater topographic and geomorphological data and water depth data of the operation area are obtained using a ship-mounted single-beam or multi-beam depth sounder. At the same time, real-time tide level data are obtained through a connected tide station or Tide-Link system, which are used to correct all measured water depths to a unified datum elevation.

[0040] Meteorological and hydrological data: Using the automatic weather station, current meter, etc. on the ship, the wind speed, wind direction, flow rate, flow direction, etc. of the operation point are obtained. These information is crucial for predicting the drift of the ship body and adjusting the power of the propeller.

[0041] Sea wave data: This is one of the key inputs of the invention. Traditional sea wave data, such as buoys and shore-based radars, have the defect of fixed point and slow update. In this embodiment, a more advanced remote sensing measurement method based on unmanned aerial vehicle (UAV) is adopted to obtain sea wave data. Specifically:

[0042] The operator can control an unmanned aerial vehicle equipped with a high-definition or hyperspectral camera to hover or fly over the dredging ship operation area, especially the windward surface. The unmanned aerial vehicle collects a continuous sequence of sea wave images and transmits the image data back to the decision system in real time.

[0043] The sea wave analysis module in the system then processes the received image sequence. For example, computer vision algorithms such as optical flow method, particle image velocimetry (PIV) technique, or semantic segmentation network based on deep learning are used to identify wave crests and troughs in the image and track their movements. In this way, the system can extract key, quantitative sea wave parameters from visual information, including:

[0044] Wave height : refers to the vertical distance between consecutive wave crests and troughs;

[0045] Period : refers to the time interval between the passage of two consecutive wave crests (or troughs) at the same point;

[0046] Sea wave breaking height : refers to the instantaneous wave crest height when the sea wave breaks in shallow water or due to wind action;

[0047] Surge water level : refers to the instantaneous, non-periodic water level rise caused by storms or long-period waves, such as nearshore surges;

[0048] Surge duration : refers to the duration of the above-mentioned surge water level rise.

[0049] These real-time, high-spatial and temporal resolution sea wave parameters obtained by unmanned aerial vehicles are the direct basis for subsequent sea wave comprehensive risk index calculation, and their accuracy is higher than that of traditional methods.

[0050] As Figure 2 The wave crest and trough extraction diagram in the sea wave image sequence is intuitively shown, that is, by processing the sea wave images continuously taken by the unmanned aerial vehicle above the dredging operation area, the positions of each wave crest and trough are identified, and this serves as the basis for subsequent calculation of key parameters such as wave height, period, and risk index.

[0051] Figure 2 The blue main curve simulates the changes in sea surface waveforms after image processing and time series reconstruction, representing the dynamic process of waves rising and falling within one cycle. Red dots mark the wave crests automatically identified by the system; these crests are the highest points in each complete wave, representing the maximum rise height of the wave. Green dots mark the corresponding wave troughs, the lowest points of the wave, reflecting the lowest point of wave energy.

[0052] pass Figure 2 The system can reliably identify the occurrence time and height of each wave crest and trough, facilitating further accurate calculation of wave height and period. This extracted high-precision wave height and period data will be directly used to calculate the wave hazard index, which in turn serves as the input variable for safety constraints in subsequent decision-making modules.

[0053] 2. Collection of ship-specific data: The system must collect key data that reflects the unique operating mechanism of a specific dredging vessel, such as a trailing suction hopper, cutter suction hopper, or grab bucket dredging vessel. This data is the cornerstone for building a ship-specific model.

[0054] (1) For trailing suction hopper dredging vessels (TSHD):

[0055] Mud hopper loading capacity: This is calculated in real time by an array of pressure sensors installed at the bottom of the mud hopper or by monitoring changes in the ship's draft. This data is directly related to the economic loading of the trailing suction hopper operation.

[0056] Sloshing amplitude of mud and sand in mud tank: By monitoring the level sensor or inclinometer installed on the inner wall of the mud tank, the violent sloshing of the mud and sand mixture in the tank can seriously affect the navigation stability of the ship, especially in bad sea conditions.

[0057] This also includes the vacuum level of the rake head, the speed and power of the mud pump, the drag force of the rake arm, and the turbidity of the overflow pipe.

[0058] (2) For CSD (Cut Suction Dredger):

[0059] Sludge discharge pipe pressure: Pressure sensors are installed at the sludge pump outlet and key nodes in the sludge discharge pipeline, such as above-water and underwater joints. This pressure is the driving force for maintaining the long-distance transport of sludge mixtures and is also a core indicator for determining whether the pipeline is blocked.

[0060] Sludge discharge flow rate fluctuation: The instantaneous flow rate is monitored using an electromagnetic flow meter or ultrasonic flow meter installed on the sludge discharge pipe. This embodiment focuses more on its fluctuation value, that is, the variance or standard deviation of the flow rate. Severe flow rate fluctuations usually indicate that the cutter head's digging conditions are unstable, such as encountering hard soil or rocks.

[0061] The rotational speed and torque of the reamer head, the depth of excavation, the position of the transverse trolley, the transverse speed, the anchor cable tension, etc.

[0062] (3) For grab dredger:

[0063] Grab opening degree: monitored by angle sensors installed on the grab hydraulic or cable system;

[0064] Grab lifting speed: obtained by monitoring the encoder or rotational speed sensor of the winch;

[0065] These two parameters are key to defining the efficiency of the grab operation cycle. In addition, the weight of the grab, the roll and pitch angles of the hull, etc. are also included.

[0066] 3. Data preprocessing: The raw data collected, whether general or proprietary, is chaotic and cannot be directly used for modeling. The preprocessing step aims to improve data quality, including:

[0067] Data cleaning: Identify and exclude obvious outliers and null values caused by sensor failure or communication interruption;

[0068] Noise filtering: Apply filters (such as Kalman filtering, moving average filtering) to high-frequency fluctuation signals such as dredge pipe pressure and sediment sloshing amplitude to extract the true trend signal;

[0069] Timestamp alignment: Since different sensors, such as GPS, flow meters, and wave height meters, have different sampling frequencies and delays, all data must be aligned to the same time reference;

[0070] Data normalization: Scale different dimensioned data, such as pressure, flow, and rotational speed, to a uniform interval (such as 0 to 1 or -1 to 1) to facilitate subsequent neural network model input.

[0071] Step S2: Build and execute ship type-specific dual-drive coupling sub-model

[0072] This step calls a ship type-specific dual-drive coupling sub-model that matches the current dredger type, such as a cutter suction dredger. The dual-drive coupling sub-model is built based on dredging-specific physical models and performs deep fusion and analysis on preprocessed multi-source data.

[0073] 1. Connotation of dual-drive coupling:

[0074] Mechanism-driven: The dredging-specific physical model has embedded physical and fluid mechanics formulas that reflect the operating principles of the ship type, used to describe the characteristics of sediment transport, fluid dynamics, and energy consumption for specific ship types. For example, for a cutter suction dredger, the model has embedded formulas for calculating pressure loss of two-phase flow of sediment mixture in the pipeline; for a drag suction dredger, it has embedded physical models for sediment settling in the mud tank.

[0075] Data-driven: The model contains artificial intelligence models trained with historical and real-time data, such as deep neural networks, support vector machines, etc. Specifically, the features of the ship-specific data and the general data that are strongly correlated with the performance indicators of the dredging operation are extracted and fused. The strongly correlated feature parameters include wave height and mud tank loading capacity. For example, a neural network is used to approximate the nonlinear relationship between the cutter torque and the pressure fluctuation of the mud discharge pipe under the same working condition.

[0076] Coupling: The mechanism model and the data model are not independent, but are mutually corrected and coupled. The mechanism model provides a physical constraint boundary for the data model, preventing the data model from producing predictions that violate physical common sense. The data model, on the other hand, modifies the empirical parameters in the mechanism model that are difficult to measure based on real-time data.

[0077] 2. Ship-specific active anomaly recognition:

[0078] During model execution, a key task is active anomaly recognition. Traditional anomaly recognition, such as simple threshold alarms, is very rough. This invention uses an advanced diagnostic method based on spectral coupling analysis. The core idea of this method is that the decline in efficiency or increase in safety risk of a dredging ship, as a complex dynamic system, is often not caused by a single factor, but by the resonance or strong coupling of environmental disturbances and equipment deviations at a certain frequency.

[0079] To quantify the degree of coupling, the system performs the following calculations:

[0080] A. The system obtains the principal component sequence of equipment deviation and the sequence of environmental disturbance;

[0081] B. The system performs Fourier transform FFT on the two time series to convert time-domain signals to frequency-domain signals, obtaining their deviation power spectrum and disturbance power spectrum These two spectra show the distribution of energy at different frequencies ;

[0082] C. At the same time, the system calculates the cross-power spectrum of the two signals. The cross-power spectrum is a complex number whose real and imaginary parts can reflect the correlation or cooperative motion strength of the two signals at a specific frequency ;

[0083] D. The system defines one or more sensitive frequency bands according to the ship type, such as to , from 0.05 Hz to 0.2 Hz;

[0084] E. The system performs integration operation in the sensitive frequency band: calculate the low-frequency co-responding quantity The integration here is for a certain component of the cross-power spectrum, usually the real part or the modulus, which represents the total energy of the device and the environment co-motion in the sensitive frequency band;

[0085] Calculate the total energy of the deviation and the disturbance This represents the average of the energy of the two signals in the same frequency band as a standardized reference;

[0086] F. Finally, calculate the average coupling strength factor This factor is a dimensionless normalized value, such as between 0 and 1, which accurately quantifies the coupling strength of the device deviation and the environmental disturbance in the sensitive frequency band.

[0087] As Figure 3 shows the key technical path of the spectral coupling analysis link, which is used for intelligent identification of whether the device anomaly in dredging operations is affected by the coupling of environmental disturbances.

[0088] Figure 3 The middle red curve represents the spectral distribution of the device deviation signal, and the blue curve represents the spectral distribution of the environmental disturbance signal. It can be observed that in the low-frequency region, especially near zero hertz, both curves have a significant peak value. This phenomenon of strong energy concentration in the same frequency band is an important feature for the system to determine the resonance of the device and the environment. The system calculates the coupling strength factor through such frequency coincidence characteristics, and when the factor exceeds the set ship-specific threshold, for example, 0.7, it is determined to be in a strong coupling state, and immediately triggers operation strategies such as reducing the lateral movement speed or activating the compensation device. Figure 3 The difference area of the red and blue curves on the middle frequency axis also represents the uncoupled frequency band, which helps the system to exclude false alarms caused by incidental fluctuations.

[0089] Finally, the system will continuously monitor the value. When is greater than a coupling factor threshold set for the ship type, such as 0.7 for the heave-torque coupling of the cutter suction type, the system determines that a strong coupling anomaly has occurred. At this time, the system will immediately trigger ship-specific countermeasures. For example, automatically activate the heave compensation system of the cutter bridge, or send an alarm and suggestion to the operator to reduce the lateral movement speed to avoid the resonance frequency.

[0090] As Figure 4 shows the dynamic evolution curve of the average coupling strength factor, which is a key quantitative indicator reflecting the coupling degree of the device deviation and the environmental disturbance obtained through spectral analysis.

[0091] InFigure 4 In the figure, the blue curve represents the fluctuation of coupling strength calculated by the system in real time during the actual operation process, which shows a certain periodic rising and falling rule, and the noise jitter forms a complex change pattern. The red dotted line represents the preset coupling judgment threshold of the dredging ship type, and the set value is 0.7, which is the dividing standard for the system to judge whether it is in a dangerous resonance state.

[0092] From Figure 4 It can be seen that the coupling factor crosses the threshold line and maintains above it for several periods, such as between thirty seconds and fifty seconds, the factor continues to be higher than the judgment value, the system will immediately identify as a strong coupling state in this stage, trigger the dedicated response measures, including reducing the cutter speed or starting the heave compensation mechanism, etc. On the contrary, when the curve is below the threshold, such as in the interval of zero to twenty seconds, the system will consider that there is no significant risk of cooperation between the device and the environment, and the normal operation strategy can be maintained.

[0093] Figure 4 The slow fluctuations and sudden changes in coupling strength also reflect the continuous perception ability and sensitive response mechanism of the steps of the application under dynamic environmental conditions.

[0094] Step S3: generating dredging operation decision

[0095] Based on the results of model fusion and analysis in step S2, the decision generation module of this step will calculate and output a set of optimal operation parameters.

[0096] 1. Decision output example:

[0097] For the drag suction type: output the optimal speed, rake depth, mud pump power, best overflow time, etc.

[0098] For the cutter suction type: output the optimal transverse speed, cutter speed, mud pump speed, trolley step distance, etc.

[0099] 2. Apply ship-specific safety constraints: when generating the above optimal decisions, the system must consider safety as the highest priority constraint. If only efficiency or energy consumption is targeted, it may still operate at high speed in dangerous sea conditions, which is unacceptable.

[0100] This embodiment creatively uses the sea wave comprehensive risk index collected in step S1 and calculated in step S3 as a key safety constraint for decision generation. The calculation process of the index is as follows:

[0101] A. The system first uses the wave height , period , breaking height , surge water level , surge duration extracted from the unmanned aerial vehicle image in step S1. and other parameters.

[0102] B. The system must also be pre-configured with a set of reference thresholds for the current ship type, which are ship type specific, as the tolerance to sea waves is completely different for different sizes and types of ships, including:

[0103] is the preset reference wave height, is the preset reference period. For example, for a large trailing suction hopper dredger, it can be set to 3.0 meters; for a medium trailing suction dredger, it can be only 1.0 meter;

[0104] is the preset reference breaking height, is the preset reference breaking impact distance, is the sea wave breaking impact distance;

[0105] is the preset reference swell water level, is the preset reference swell duration, is the preset reference swell impact distance,

[0106] is the swell impact distance.

[0107] C. The system calculates three sub-item risk evaluation indexes respectively:

[0108] (1) Sea wave risk evaluation index:

[0109]

[0110] This formula evaluates the danger of the regular wave height and period, where represents the relative wave height, represents the relative period, and generally the waves with short periods impact the ship body more violently, and the sum of the two is taken as a measure of danger;

[0111] The function is a sigmoid saturation function, whose output value is smoothly limited in the range of (-1, 1) or (0, 1). When the sea wave is much smaller than the reference value, the risk index is close to 0; when the sea wave is much larger than the reference value, the risk index tends to 1 (maximum danger), but will not increase indefinitely. This conforms to the actual situation that the danger is saturated.

[0112] (2) Sea wave breaking risk evaluation index:

[0113]

[0114] This formula evaluates the danger of breaking wave, which is especially important when operating in near-shore shallow water area. The danger is determined by the relative height (H) and the relative closeness (D) of the breaking wave.

[0115] And is also a Sigmoid function, whose output range is strictly limited in (0, 1), with similar advantages as , i.e. providing smooth, bounded risk evaluation.

[0116] (3) Surge danger evaluation index:

[0117]

[0118] This formula evaluates the danger of surge, which considers three factors of surge: relative water level (H), relative time length (T) and relative closeness (D).

[0119] The Sigmoid function is used for the same reason as above, to ensure the non-linear saturation property of risk evaluation.

[0120] D. Finally, the system calculates the comprehensive sea danger index (D) by weighted average:

[0121]

[0122] The weight coefficients 0.4, 0.3, 0.3 here are calibrated according to a large amount of historical data and expert experience. In this example, the danger of regular wave height and period is considered to be the most important, while breaking and surge each accounts for 30%, which can also be dynamically adjusted according to the ship type and operating area.

[0123] E. Application As safety constraints:

[0124] The calculated D (a value between 0 and 1) is immediately sent to the decision generation module, which must meet , the maximum tolerable danger threshold value specific to the ship type when generating decisions;

[0125] If exceeds the warning threshold, such as 0.6, the decision module must automatically reduce its efficiency target and increase the safety target. For example, the generated decision will automatically include constraints such as reducing the speed by 30%, limiting the mud power to 80%, etc.

[0126] If​​​​​​​ When the index exceeds a certain threshold, such as 0.85, the decision module will override all other optimization objectives and directly generate the highest priority instruction to immediately stop the operation and lift the cutter head to a safe position above the water surface.

[0127] For example, Figure 5 The construction process of the sea wave comprehensive risk index and the index curve are shown, Figure 5 The blue curve in the middle represents the sea wave risk index calculated based on wave height and period, reflecting the intensity of the impact of traditional periodic waves on the hull. When the period is high, it means that the safety of the operation decreases.

[0128] The green curve is the breaking risk index, which reflects the destructive impact of sea waves when they break in shallow water. When the waves are steep and close to the preset warning range, the index rises rapidly.

[0129] The cyan line represents the surge risk index, which quantifies the risk of non-periodic water level rise caused by storms or long-period waves. Its rise generally lags behind the regular wave height, but has a significant impact under low-frequency strong wave conditions.

[0130] Figure 5 The most critical is the red bold curve, which is the sea wave comprehensive risk index constructed by the invention. This curve integrates the above three sub-indexes and is weighted and integrated, forming an important basis for the intelligent decision system to determine whether to adjust the dredging operation parameters. When the index exceeds a certain threshold, such as Figure 5 In the multiple time periods in the figure, the index value reaches or exceeds 0.8, the system will immediately limit parameters such as the depth of the cutter head and the power of the mud pump to ensure the safety of the operation. Conversely, in the stable region where the index is below 0.4, the system can prioritize the progress goal to improve efficiency.

[0131] Figure 5 The fluctuations of various curves are intertwined, but the comprehensive curve maintains a consistent upward or downward trend, demonstrating the effectiveness and practical value of the weighted construction mechanism in maintaining stable response capabilities in a multi-dimensional disturbance environment.

[0132] Step S4: Ship-specific resilience monitoring of the execution effect of dredging operation decision

[0133] In this step, the system not only generates decisions, but also monitors the execution effect of the decisions in real time, and ensures that the system can still maintain resilience, i.e., safely and controllably degrade operation, rather than directly crash, when unexpected situations are encountered.

[0134] The ship-specific resilience monitoring module of the embodiment is specifically used to handle two typical emergency scenarios: environmental compliance and unknown geology.

[0135] 1. Trigger of resilience supervision: The system will monitor two high-level triggers in real time, including:

[0136] Trigger 1: Entering environmental sensitive area

[0137] Monitoring method: The decision server of the system is pre-loaded with the GIS electronic map of the work area, and the environmental sensitive area such as coral reef protection area, important fish spawning area, and drinking water intake area is clearly marked on the map;

[0138] Trigger: The system compares the real-time GNSS positioning of the ship with the GIS map, and once it is found that the work radius or anchoring range of the dredger has invaded the electronic fence of the environmental sensitive area, the resilience supervision is triggered immediately.

[0139] Trigger 2: Encounter with bottom sudden change scene

[0140] Monitoring method: rely on the analysis results of the double-drive coupled sub-model in step S2;

[0141] Trigger: When the model detects that the ship-specific data has a sharp and sustained anomaly, such as the instantaneous surge of the cutter torque to the limit, the fluctuation value of the mud discharge flow is continuously high;

[0142] Difference: This trigger is different from the frequency spectrum coupling, which may only be an efficiency problem, and different from the sensor failure, which is a data reliability problem; here the trigger is clearly directed to the mutation of the physical world, i.e. digging into hard rock, sunken ship, or high-viscosity clay layer that is not shown in the survey report.

[0143] 2. Execution of resilience supervision: Once any of the above triggers is activated, the system will perform the following actions:

[0144] A, combined with real-time sea wave data: when the system executes the supervision action, it does not stop all work at once, but also refers to the sea wave comprehensive risk index calculated in step S3 , which is a high-level safety consideration.

[0145] For example, when encountering a sudden change in bottom (cutter jam), if the sea conditions are very bad at the same time, the system must adopt the most conservative strategy, such as immediately abandoning the soil and lifting the bridge; if the sea conditions are good , the system can try more fine-grained degradation operations, such as reversing the cutter and reducing the speed.

[0146] B, automatically generate ship-specific operation parameter degradation coefficient: the system will generate one or a set of degradation coefficients (multipliers between 0 and 1). This coefficient is ship-specific and scene-dependent.

[0147] C. Adjust the operation decision based on the degradation: these generated degradation coefficients are immediately sent to the decision generation module in step S3. They will be used as the highest priority constraints, overriding the original efficiency optimal or energy optimal targets. The decision module must regenerate and issue the safe degraded operation parameter instructions, such as the instructions limiting the torque and transverse speed, based on these degradation coefficients.

[0148] 3. Supervision certificate: in this step S4, the system also generates a tamper-proof report specific to the ship type. This report details when, where, why, how, and the results of the response. This report uses encryption hashing or blockchain technology to prevent post-factum modification, for verification and audit by the construction party, supervision party, and maritime department.

[0149] Step S5: Iterative optimization of the ship-specific dual-drive coupling sub-model based on the above execution effect

[0150] Since the dredging operation is not one-off, but a long-term process, the system of the embodiment is designed to have the ability of online learning or reinforcement learning, specifically including the following processes:

[0151] 1. Feedback data source: the tamper-proof report generated in step S4 and the verified execution effect data;

[0152] 2. Optimization target: the ship-specific dual-drive coupling sub-model in step S2;

[0153] 3. Optimization process: for example, the system records in step S4 that the frequency of performing degradation operations on bottom conditions in a certain area is very high. The data-driven module of the system will obtain these reports, and it will learn that there is a strong correlation between the GIS coordinates of the specific area and the characteristics of high torque and high pressure fluctuation of the dredged material, and the weights of the neural network model will be updated in the next iterative optimization;

[0154] 4. Effect: when the dredging ship is about to operate near the GIS coordinates of the specific area next time, the model in step S2 can predict that the bottom condition here is hard; therefore, in step S3, the system will no longer blindly pursue high efficiency, but will proactively reduce the cutter speed and transverse speed, thereby avoiding passive triggering of degradation operations and realizing intelligent upgrading from passive resilience supervision to proactive decision-making.

[0155] The embodiment one not only solves the contradiction between ship type universality and accuracy, but also through fine environmental risk quantification, deep working condition coupling diagnosis, and robust resilience supervision strategy, makes the whole dredging operation decision system exhibit unprecedented high safety, high adaptability, and high intelligence level when facing complex, dynamic, and high-risk real marine environment.

[0156] Embodiment two:

[0157] This embodiment two further describes in the step S2 of embodiment one: how to build and implement the model of the specific ship type double drive coupling sub-model, and the specific implementation of the model building details and its core robust self-diagnosis mechanism.

[0158] This embodiment aims to solve the adaptive and stability conflict problem highlighted in the background art, especially when the model encounters sensor failure or extreme working conditions, etc. Data pollution, the system incorrectly adjusts the model parameters, leading to model collapse or decision-making errors.

[0159] I. Architecture and operation of double drive coupling sub-model

[0160] The ship-specific double drive coupling sub-model used in this embodiment is an advanced hybrid modeling method, which combines rigorous physical mechanisms and flexible data-driven capabilities, and realizes mutual supervision and correction between the two.

[0161] 1. Model composition level: the model includes at least the following two levels:

[0162] Mechanism-driven layer: this layer is a mathematical model based on the principles of physics, fluid mechanics, engineering mechanics, etc. of a specific dredging ship type, i.e. dredging-specific physical model. For example, for a cutter suction dredger, this layer includes the energy loss equation of the two-phase flow of mud in the pipeline, the soil resistance calculation formula of the cutter, the force model of the ship body on the transverse system, etc. The main role of the mechanism-driven layer is to provide accurate physical constraints and interpretable baseline predictions;

[0163] Data-driven layer: this layer is a black box model composed of deep neural networks, support vector machines, or reinforcement learning algorithms, which uses a large amount of historical operation data and real-time sensor data for training, and is used to capture complex, nonlinear, high-dimensional relationships that are difficult to accurately describe by mechanism models. For example, the nonlinear relationship between the speed of the mud pump and the actual delivery efficiency in a specific bottom environment.

[0164] 2. Coupling mechanism of the model: during the operation of the model, the mechanism-driven layer and the data-driven layer are not run in isolation, but are dynamically coupled and corrected by a bias analyzer. This coupling mechanism is reflected in the following aspects:

[0165] First, the mechanism-driven layer gives a mechanism prediction value based on the current working condition parameters , such as predicting the pipeline pressure under the current flow rate;

[0166] Second, the data-driven layer gives a data prediction value based on real-time sensor input ;

[0167] Next, the system reads actual measurement values directly from field sensors . The system utilizes mechanism predicted values to constrain data predicted values and calculates the deviation between them , which is defined as the difference between mechanism predicted values and actual measurement values, i.e.

[0168]

[0169] This deviation represents the inherent error of the mechanism model and the degree of deviation between real-time data and physical principles. The core of the dual-drive coupling mechanism is that the model will determine whether the current model parameters need to be adjusted according to the size of this deviation and a preset first deviation threshold , so as to realize the adaptive iterative optimization of the model, including:

[0170] When , it means that the mechanism model and the actual working condition are well fitted, or the deviation is within an acceptable range. At this time, the model keeps the current parameters running;

[0171] When , it means that the model has deviated from the actual situation. At this time, the system triggers the adaptive mechanism to fine-tune some parameters in the data-driven layer or the mechanism-driven layer. For example, update the neural network weights based on the gradient descent algorithm, or correct the friction coefficient in the mechanism model to reduce , so that the model adapts to the new working condition or environment.

[0172] As Figure 6 shows the key comparison logic in the real-time execution process of the dual-drive coupling sub-model, that is, by monitoring the difference between the physical model output and the actual measured value of the device sensor, the deviation is generated to judge whether the model deviates from the actual working condition.

[0173] In Figure 6 ​In this case, the blue curve represents the predicted value output by the mechanism-driven model, and the green curve represents the measured value collected by the system through the field sensor. The trend of the two curves is close in most time periods, indicating that the model has good prediction effect under stable working conditions. However, in some periods, especially between thirty seconds and forty-five seconds, the green curve shows intensified fluctuations and slight phase shift, and the difference between the blue curve is significantly enlarged. The right red curve is the deviation curve constructed according to the difference between the two, which reflects the fitting degree between the model and the field. When the deviation value continues to rise and exceeds the judgment threshold set by the system, the system will trigger the fusion mechanism diagnosis to determine whether the model drift or sensor anomaly. When it is identified that the deviation anomaly belongs to data pollution, the system will suspend the model parameter update and enable the emergency strategy to prevent the model from being misled by false data.

[0174] II. Fusion mechanism diagnosis

[0175] Although the above coupling mechanism achieves adaptability, it has a fatal flaw: when the sensor fails, such as the pressure sensor reading zero or outputting the maximum value, will suddenly become very large, exceeding . The adaptive mechanism will mistakenly think that this is a defect of the model itself, and will crazily adjust the model parameters, eventually leading to model pollution or even collapse.

[0176] This embodiment introduces a high-robustness self-diagnosis and shunt treatment mechanism for this conflict problem, namely the fusion mechanism diagnosis. Based on the original, two additional technical links are introduced: the second deviation threshold (mutation deviation threshold) and deviation continuity monitoring, to distinguish between two essentially different types of anomalies, model drift and data pollution.

[0177] 1. Introduce double threshold and deviation continuity monitoring: based on the original first deviation threshold , this embodiment adds two key monitoring parameters:

[0178] (1) Second deviation threshold : Set a threshold much larger than to determine whether an extreme anomaly has occurred. For example, may be 5% of the predicted value, and may be 30% of the predicted value;

[0179] (2) Deviation continuity judgment : Set a minimum duration to distinguish between transient noise and persistent anomaly. For example, can be set to 5 seconds.

[0180] 2. Mechanism logic and shunt treatment: the system will continuously monitor and its result is handled by splitting into three cases:

[0181] Case 1: normal operation or model fine-tuning, i.e. is small, no more than , then:

[0182] Status: ;

[0183] Diagnosis conclusion: the model is basically matched with the actual working condition;

[0184] Disposal measures: maintain the existing parameters for operation, or carry out a very small amplitude background optimization.

[0185] Case 2: model drift, triggering adaptive adjustment, i.e. is large, but the duration is short or within an acceptable range, then:

[0186] Status: , and the duration is less than ;

[0187] Diagnosis conclusion: it is believed that the model prediction has drifted from the actual situation due to slow changes in the environment or the substrate;

[0188] Disposal measures: trigger the adaptive iterative optimization of the first part of embodiment two. For example, the system will retrain the data-driven layer model parameters based on to make the model realign with the actual working condition. This adjustment is gradual and limited, and will not cause the model to lose control.

[0189] Case 3: data pollution or extreme working condition, triggering resilience disposal, i.e. is huge, and the duration is long, then:

[0190] Status: , or is greater than and the duration exceeds ;

[0191] Diagnosis conclusion: this is the highest level of abnormality. The system immediately determines that there is a sensor failure, communication interruption, extreme external interference, etc., i.e. the data source or the physical working condition itself has an untrustworthy abnormality;

[0192] Disposal measures: this is the most critical part of the embodiment, which ensures the robustness of the model, including the following:

[0193] (1) Isolate abnormal data sources: the system will mark the data that causes as untrustworthy or faulty, and completely shield or not adopt these isolated data in the subsequent decision-making and model optimization process;

[0194] (2) Switch to Emergency Model: The system's decision generation module immediately invokes a preset, static emergency model or safety model. This emergency model no longer relies on current real-time anomaly data, but instead makes conservative decisions based on historical safety data and an expert rule base. For example, the sludge pump speed is forcibly locked at a safe value, and the cutter torque is limited to a safe upper limit.

[0195] (3) Send alarms: The system sends high-level sensor failure or extreme working condition alarms to operators and remote monitoring centers, and prompts the affected sensors or model components, requiring manual intervention for verification.

[0196] 3. Automatic model recovery and takeover: The system continues to monitor in the background during the execution of the emergency model. Once the isolated sensors return to normal, or the extreme external interference disappears, leading to... Falling back to After a period of time, the system will determine that the abnormal operating condition has been resolved. At this point, the system will smoothly switch back from the emergency model to the adaptive model, restoring the dredging operation to an intelligent optimization state.

[0197] The dual-drive coupling model and self-diagnostic mechanism proposed in this second embodiment introduce a dual-bias threshold. and They successfully distinguished technically between two completely different types of substances that both lead to high... Anomalies: Model drift (correctable) and data contamination (requires isolation).

[0198] Adaptability to model drift: through This ensures that the model can continuously and gradually adjust itself as parameters such as substrate and environment change slowly, maintaining the system's adaptability and high efficiency.

[0199] Robustness to data contamination: through This ensures that in the event of sudden high-deviation events such as sensor failure, the system can immediately perform data isolation and emergency takeover, effectively preventing erroneous data from contaminating or poisoning the core model. This completely solves the technical problem of adaptive systems being prone to collapse under harsh working conditions in the background technology, and greatly improves the reliability and safety of the dredging operation decision system.

[0200] In summary, the technical solution of Embodiment 2 endows the entire system with a high degree of self-diagnosis capability and resilience for fault-tolerant operation in a dynamic and uncertain marine environment.

[0201] Example 3:

[0202] In the background art, intelligent decision-making systems face a serious technical problem of conflict between responsiveness and stability: on the one hand, the system needs to respond sensitively to changes in the external environment to ensure safety; on the other hand, if environmental signals, such as sea wave danger indexes, fluctuate or oscillate at a high frequency near a certain decision threshold, a too sensitive system will instead lead to catastrophic high-frequency switching of control instructions, causing severe wear and tear of equipment or control failure.

[0203] The purpose of the present embodiment is to completely solve the above conflict through a delicate weight adjustment strategy, so that the system can both respond sensitively to real environmental changes and remain stable to meaningless noise fluctuations, and when it must switch strategies, ensure that the process is smooth and non-impulsive, which specifically includes the following contents:

[0204] I. Establishment of the relationship between the objective function and the weight

[0205] In dredging operations, the decision-making system always faces at least two core optimization goals that conflict with each other: one is the energy consumption goal, and the other is the progress goal.

[0206] 1. Definition of the objective function: the system internally defines at least two optimization objective functions:

[0207] Energy consumption objective function : This function aims to minimize the unit cost of the operation. is a composite function that includes: the total power consumption of key equipment such as the main mud pump, water flushing pump, cutter head, and traversing winch of a cutter suction dredger; it can also be the fuel consumption per unit of time or unit of earthwork; in some cases, it can also include a depreciation model of the equipment, as high energy consumption operation is usually accompanied by high wear and tear;

[0208] Progress objective function : This function aims to maximize the efficiency of the operation. is also a composite function that includes: the amount of excavated earthwork per unit of time. For example, it can be calculated by the product of the flow rate and concentration of the discharged mud; for a drag suction dredger, it can be the loading efficiency per drag or the number of daily cycles.

[0209] These two objectives are naturally conflicting in physics. For example, to improve , the operator will usually increase the speed of the mud pump and the power of the cutter, but this will inevitably lead to a sharp rise in .

[0210] 2. Introduction of the weight relationship: in order to find an optimal balance point between the conflicting goals of energy consumption and progress, the present embodiment introduces a dynamic weight.

[0211] The final optimization of the system's comprehensive objective function is the weighted sum of the two sub-objective functions, the system assigns a weight to the energy consumption objective function and a weight to the progress objective function At any moment, the two weights must satisfy a normalization constraint, i.e.

[0212]

[0213] This constraint indicates that if is set to 0.8, then must automatically be 0.2. This indicates that the system’s current highest priority is energy saving and safety, because high energy consumption often comes with high risk, while progress is in the secondary position. Conversely, if is set to 0.9, then must be 0.1, indicating that the system is in a rush mode.

[0214] 3. Introduction of trigger factor: in the traditional static scheme, and may be set to fixed values before the entire construction phase begins, but this approach cannot cope with dynamic marine environments.

[0215] A key point of the present invention is that and are dynamically adjusted. The system introduces a trigger factor that is used to determine the specific values of and in real time.

[0216] Optionally, the trigger factor directly comes from the sea wave comprehensive risk index , which quantifies the risk level of the current working environment, and the logical relationship between the weight , is clear:

[0217] When is extremely low (calm sea), it means that the environmental risk is very small. At this time, the system should boldly pursue efficiency, setting (progress weight) to a high value and (energy / safety weight) to a low value;

[0218] When is extremely high (severe sea conditions), it means that the environmental risk is extremely great. At this time, the system must prioritize safety and forcibly set to a high value. For example, by reducing energy consumption to reduce equipment load and risk, while Must be set to low value.

[0219] II. Solve the problem of jitter: high stability hysteresis and smooth transition mechanism

[0220] Will be adjusted as a trigger factor and , although the responsiveness is achieved, but also brought the oscillation problem described above. If the calculated value due to sensor noise or instantaneous fluctuations in the sea wave, just around a certain set of single-point threshold, such as 0.6 back and forth (0.59, 0.61, 0.58, 0.62…) jump, the system will command and crazy switching between high and low values. This high-frequency weight oscillation will be directly translated into control instruction oscillation, which is unbearable for any physical system, it will cause huge mechanical impact and safety hazards.

[0221] In order to completely solve this conflicting technical problem, the invention introduces a double high stability mechanism consisting of hysteresis threshold and smooth transition.

[0222] Mechanism one: double threshold-based hysteresis control

[0223] This embodiment does not use a single decision threshold, but sets two different thresholds, namely:

[0224] Onset threshold : This is a higher dangerous activation threshold;

[0225] Off threshold : This is a lower dangerous deactivation threshold.

[0226] The two thresholds must satisfy the relationship in value.

[0227] For example, in this embodiment, is set to 0.7, and is set to 0.5. Through these two thresholds, the system defines three explicit state intervals and switching logic:

[0228] State one: safety and progress priority zone ( );

[0229] Scenario: when is below 0.5, indicating that the sea conditions are very safe;

[0230] Action: the system sets (progress weight) to a high value. For example .

[0231] State Two: Dangerous vs Safe Priority Zone ( );

[0232] Scenario: When is higher than 0.7, it indicates that the sea condition has entered the dangerous level;

[0233] Action: The system sets Energy Consumption / Safety Weight) to a high value. For example .

[0234] State Three: Hysteresis Hold Zone ( );

[0235] Scenario: When is exactly between 0.5 and 0.7;

[0236] Action: The system does not perform any new switching action, but maintains the state before switching.

[0237] If the system is rising from the Safe Zone ( ) to 0.6, the system will maintain the Progress Priority state ( );

[0238] If the system is falling from the Dangerous Zone ( ) to 0.6, the system will maintain the Safety Priority state ( ).

[0239] And as for how the hysteresis mechanism solves the jitter problem, here we can simulate a jitter scenario:

[0240] Jumping around 0.6, for example, 0.59, 0.61, 0.58, 0.62… Since all these jitter values (0.58 to 0.62) fall within the hysteresis hold zone between (0.5) and (0.7); therefore, no matter oscillates at a high frequency within this interval, the system will ignore these noises and maintain its state before entering this region, whether it is safety priority or progress priority. Therefore, the weights and will remain stable and no switching will occur, and the oscillation of control commands is completely eliminated.

[0241] Only when the sea wave is really over (0.7), the system will switch to safety priority; only when the sea wave is really below (0.5), the system will switch back to progress priority. This hysteresis loop design perfectly balances the responsiveness to real signals and stability to noise signals.

[0242] Mechanism Two: Smooth Transition Function

[0243] The hysteresis mechanism above solves the problem of when to switch, but not how to switch. If the system decides to switch at the very moment , then will jump from 0.1 to 0.8, which is a mathematical step. This step change in weight will also cause a step change in control command, which is a sudden large change that will cause a large impact on the mechanical system, which is something to be avoided in engineering. To solve this impact problem, this embodiment introduces a smooth transition function.

[0244] The smooth transition function ensures that when the system decides to switch the weight, i.e. crosses or , the change in weight and is gradual and smooth, rather than instantaneous.

[0245] For example, when the system detects that has crossed (0.7), and needs to switch from its current value of 0.1 to the target value of 0.8, the system will not immediately execute ;

[0246] Instead, the system will start a transition timer, and over the next pre-set time period, it will gradually change using a function, for example by:

[0247] Linear transition: the value of is linearly increased from 0.1 to 0.8 over 10 seconds. For example, it is increased by every second;

[0248] S-shaped function transition: using the form , where is an S-shaped function that smoothly varies between 0 and 1. The advantage of the S-shaped function is that at the start and end of the transition, the rate of change (derivative) is close to zero, which makes the transition slow at the start and end, and fast in the middle, achieving the best smoothness and minimal impact on the physical device.

[0249] And based on the above implementation, while is being smoothly increased, is being smoothly decreased at the same time, to always satisfy the constraint . For example, at seconds, has smoothly transitioned to 0.45, then must be exactly equal to .

[0250] The embodiment establishes the adaptive basis of decision-making through the weight relationship of , and the trigger factor composed of . More importantly, it fundamentally solves the stability problem of control oscillation through the hysteresis double-threshold / mechanism; at the same time, it solves the smoothness problem of switching impact through the smooth transition function mechanism. Ultimately, in the embodiment, the system obtains a high-level decision-making brain that can not only respond sensitively to real danger, but also intelligently filter false noise, and stably protect physical equipment. It perfectly solves the conflict between responsiveness and stability in the background technology, making the dredging operation obtain the crucial control stability and mechanical safety on the road to intelligentization and unmanned.

[0251] As Figure 7 shows the smooth transition function simulation diagram. Figure 7 The upper part shows the dynamic change process of the energy consumption weight and the progress weight, where the red curve represents the energy consumption weight, and the blue curve represents the progress weight. The changes of the two curves strictly follow the constraint condition that the sum of the weights is always one. The lower part shows the change of the trigger factor, i.e., the sea wave comprehensive danger index, and the hysteresis interval composed of the set opening threshold and closing threshold.

[0252] From Figure 7 , it can be clearly observed that when the trigger factor value crosses the opening threshold represented by the red dashed line from the safe interval, the system starts to smoothly transition from the progress priority mode to the safety priority mode. At this time, the energy consumption weight gradually increases from a low level while the progress weight decreases accordingly. The entire transition process lasts for a preset time period and the change curve presents an S-shaped feature, ensuring that the change rate is minimal at the beginning and end of the weight adjustment, effectively avoiding the equipment impact problem caused by weight mutation in traditional systems. Similarly, when the trigger factor value crosses the closing threshold represented by the blue dashed line from the danger interval, the system returns to the progress priority mode in the same smooth way.

[0253] It is particularly noteworthy that in the hysteresis retention area between the two threshold values of the trigger factor, although there are normal fluctuations in environmental parameters, the system weight remains stable, which proves that the hysteresis control mechanism adopted by the invention can effectively filter environmental noise jitter and prevent harmful oscillation of control commands near the high-risk threshold. The entire weight adjustment process not only ensures the sensitive response of the system to real environmental danger changes, but also ensures the stability and continuity of the control commands, solving the technical problem that responsiveness and stability cannot be compatible, and providing a reliable decision-making basis for intelligent control of dredging operations.

[0254] Embodiment Four:

[0255] As Figure 8 shown, corresponding to the method embodiments described above, the application also proposes a dredging operation intelligent decision optimization system based on multi-source data fusion, comprising:

[0256] a data acquisition module for acquiring and preprocessing multi-source data, the multi-source data including general data and ship type exclusive data for different dredging ship types;

[0257] a model execution module for constructing and executing a ship type exclusive dual-drive coupling sub-model based on the dredging ship type, fusing the multi-source data, and performing ship type exclusive spectrum coupling analysis and ship type exclusive active anomaly recognition;

[0258] a decision generation module for performing ship type exclusive dynamic weight adjustment and ship type exclusive safety constraint based on the results of the multi-source data fusion, and calling a ship type exclusive pre-trained working condition library to generate a dredging operation decision;

[0259] a supervision and verification module for performing ship type exclusive resilience supervision and verification on the execution effect of the dredging operation decision, and generating a ship type exclusive tamper-proof report;

[0260] an iterative optimization module for iteratively optimizing the ship type exclusive dual-drive coupling sub-model based on the execution effect.

[0261] The system realizes the combination of unified framework and ship type specificity by constructing a ship type exclusive dual-drive coupling sub-model and data system, so that the decision system can accurately match the unique physical mechanism of a specific ship type and run under a unified intelligent framework. In particular, the system introduces a strategy with hysteresis and smooth transition, which not only realizes the sensitive response to environmental changes, but also to some extent eliminates the control instruction oscillation caused by data noise jitter, so that the responsiveness and stability of the system are taken into account, ensuring the smooth operation of the physical equipment.

[0262] Embodiment Five:

[0263] Corresponding to the above embodiments, the application also proposes an electronic device.

[0264] As Figure 9 shown is a structural schematic diagram of an electronic device in the application, the electronic device 100 comprising a processor 101 and a memory 103. The processor 101 and the memory 103 are connected, such as through a bus 102. Optionally, the electronic device 100 can also comprise a transceiver 104. It should be noted that in actual application, the transceiver 104 is not limited to one, and the structure of the electronic device 100 does not constitute a limitation on the embodiments of the application.

[0265] The processor 101 can be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA, or other programmable logic device, transistor logic device, hardware component, or any combination thereof. It can implement or execute various exemplary logical blocks, modules, and circuits described in conjunction with the present disclosure. The processor 101 can also be a combination that implements a computing function, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

[0266] The bus 102 can include a path that transmits information between the above-mentioned components. The bus 102 can be a PCI bus or an EISA bus, etc. The bus 102 can be divided into an address bus, a data bus, a control bus, etc. For convenience of representation, Figure 9 In the figure, only one thick line is used, but it does not mean that there is only one bus or only one type of bus.

[0267] The memory 103 is used to store a computer program corresponding to the above-mentioned embodiment of the method for intelligent decision optimization of dredging operation based on multi-source data fusion. The computer program is controlled and executed by the processor 101. The processor 101 is used to execute the computer program stored in the memory 103 to realize the content shown in the above-mentioned method embodiment.

[0268] The electronic device 100 includes but is not limited to mobile terminals such as notebook computers, PADs (tablet computers), and the like, and fixed terminals such as desktop computers and the like. Figure 9 The electronic device 100 shown is only an example and should not limit the function and use range of the embodiments of the present application.

[0269] Although the embodiments of the present application have been shown and described above, it should be understood that the above-mentioned embodiments are exemplary and should not be understood as limiting the present application. Those skilled in the art can make changes, modifications, replacements, and variations to the above-mentioned embodiments within the scope of the present application.

Claims

1. A method for intelligent decision optimization of dredging operation based on multi-source data fusion, characterized in that, Includes the following steps: Collect and preprocess multi-source data, including general data and vessel-specific data for different dredging vessel types; Based on the dredging vessel type, a vessel-specific dual-drive coupling sub-model is constructed and executed to fuse the multi-source data and perform vessel-specific spectral coupling analysis and vessel-specific active anomaly identification. Based on the results of the multi-source data fusion, ship-specific dynamic weight adjustment and ship-specific safety constraints are performed, and ship-specific pre-trained working condition library is called to generate dredging operation decisions. The effectiveness of the dredging operation decisions is monitored and verified using ship-specific resilience, and a ship-specific tamper-proof report is generated. Based on the execution results, the ship-specific dual-drive coupling sub-model is iteratively optimized.

2. The method of claim 1, wherein, The types of dredging vessels include trailing suction hopper dredging vessels, cutter suction dredging vessels, and grab bucket dredging vessels. The ship type-specific data includes: For the aforementioned trailing suction hopper dredging vessel, the vessel-specific data includes at least the mud hopper loading capacity and the mud sloshing amplitude within the mud hopper; For the aforementioned cutter suction dredging vessel, the vessel-specific data shall include at least the sludge discharge pipe pressure and sludge discharge flow rate fluctuation values; For the grab-type dredging vessel, the vessel-specific data includes at least the grab opening and closing degree and the grab lifting speed.

3. The method of claim 1, wherein, The general data includes ocean wave data; The sea wave data includes wave height extracted from sea wave images collected by a drone , period , sea wave breaking height , surge water level , and surge duration ​ The method also includes calculating a comprehensive wave hazard index based on the wave data. And used for the ship type-specific safety constraints; The comprehensive wave hazard index ; in, The wave hazard assessment index. The wave breakage hazard assessment index. It is a wave surge hazard assessment index.

4. The method according to claim 3, characterized in that, The wave hazard assessment index ,in To preset the reference wave height, For preset reference period; The wave breakage risk assessment index ,in To preset the reference breaking height, The distance was affected by the breaking waves. To preset the reference fracture impact distance, It is a natural constant; The wave surge hazard assessment index ,in To preset the reference surge level, To preset the reference surge duration, Distance affected by surge This is a preset reference surge impact distance.

5. The method according to claim 1, characterized in that, The construction and execution of the ship-specific dual-drive coupling sub-model includes: Mechanism layer construction: Based on the dredging-specific physical model, the dredging-specific physical model is used to describe the sediment transport, hydrodynamic and energy consumption characteristics of a specific dredging vessel type, and introduces vessel type-specific correction terms for the dredging vessel type; Constructing a data layer: Extracting and fusing feature parameters that are strongly correlated with dredging performance indicators from the ship type-specific data and the general data, the strongly correlated feature parameters include wave height and mud hopper loading capacity; Establish a fusion mechanism: calculate the deviation between the output value of the mechanism layer and the feature vector of the data layer, and automatically adjust the preset parameters in the dredging-specific physical model when the deviation is greater than a preset deviation threshold.

6. The method according to claim 5, characterized in that, The establishment of the fusion mechanism also includes: Set a sudden deviation threshold that is greater than the preset deviation threshold; When the deviation is greater than the sudden deviation threshold, or when the deviation is still greater than the preset deviation threshold after one or more automatic adjustments, the automatic adjustment is paused and the fusion mechanism diagnosis is triggered. The fusion mechanism diagnosis includes: performing consistency verification on the multi-source data that constitutes the feature vector of the data layer; If the consistency check determines that there is an abnormal data source, then the abnormal data source is isolated, and the data reconstruction model is enabled to generate alternative data for the calculation of the deviation, and the automatic adjustment is restored. If the consistency verification determines that the data source is consistent, then the dredging-specific physical model is deemed invalid, the automatic adjustment is frozen, and an emergency response model is called from the ship-specific pre-trained operating condition library to replace the dredging-specific physical model.

7. The method according to claim 1, characterized in that, The ship-type specific spectral coupling analysis includes: For each dredging vessel type, its sensitive frequency band is adapted; the principal component sequence of equipment deviation and the environmental disturbance sequence are obtained, and the deviation power spectrum is calculated respectively. Disturbance power spectrum and cross-power spectrum ; Calculate the low-frequency cooperative response within the sensitive frequency band. and the total energy of deviation and disturbance ; Calculate the average coupling strength factor , and when the stated If the coupling factor exceeds the ship type-specific threshold, it is determined to be a strong coupling, and ship type-specific countermeasures are implemented.

8. The method according to claim 3, characterized in that, The ship type-specific dynamic weight adjustment includes: Set energy consumption weight With schedule weight ,in ; For the aforementioned dredging vessel type, a vessel-specific triggering factor is set, which is composed of parameters from the vessel-specific data and the wave data. When the ship type-specific triggering factor is triggered, dynamic adjustment is made. and Weight allocation.

9. The method according to claim 8, characterized in that, A start threshold and an exit threshold are set for the ship type-specific triggering factor, wherein the exit threshold is lower than the start threshold; The dynamic adjustment and The weight allocation specifically includes: When the value of the ship type-specific trigger factor rises and exceeds the activation threshold, the weight allocation is adjusted to the first target weight reorganization. When the value of the ship type-specific trigger factor decreases and falls below the exit threshold, the weight allocation is adjusted to the second target weight reorganization. And when the weight allocation adjustment is triggered, the and The value is based on a preset smooth transition function and gradually changes to the first or second target weight reorganization within a preset transition time period to avoid oscillations in the work status caused by sudden weight changes.

10. The method according to claim 3, characterized in that, The specific resilience monitoring for this ship type includes: Regarding the aforementioned dredging vessel type, when the dredging vessel is detected to have entered an environmentally sensitive area or encountered a sudden change in the seabed sediment; By combining the real-time wave data, a downgrade coefficient for the ship type's specific operating parameters is automatically generated; The operational parameters in the dredging operation decision are adjusted based on the downgrade coefficient of the operational parameters.

Images