Method and system for judging ship motion situation through ship 3 to 6 posture information

Abstract

The application discloses a kind of by ship 3 to 6 posture information determination ship movement situation method and system, comprising: acquisition 3 to 6 posture core data, environment and control data, after pre-processing, calculate 3 to 6 posture dynamic coupling coefficient and dynamic confidence degree;Based on this, construct weighted optimization objective function and constraint condition, obtain optimized attitude parameter by solving;With parameter, coupling coefficient and confidence degree, train LSTM network to build adaptive swing risk short-term prediction model, for predicting future ship swing risk grade and posture;According to risk grade, posture, ship type and sea state query decision rule base, generate structured suggestion, collect execution feedback and be used for model updating;The application solves the reliability and efficiency problem of data acquisition, provides high-quality, continuous, synchronous data basis for subsequent processing;Solve the problem of isolated parameter optimization, fixed and rigid coupling coefficient, realize high-precision, strong adaptation attitude parameter optimization.

Description

Technical Field

[0001] This invention relates to the field of ship navigation safety and intelligent sensing technology, and in particular to a method and system for judging the motion status of a ship through three-axis and six-attitude information. Background Technology

[0002] Ship attitude information in three directions and six states (roll angle, pitch angle, roll rate, pitch rate, heave displacement, and heave velocity) is a crucial indicator of navigation stability, directly impacting cargo safety (container overturning, bulk cargo displacement), crew comfort (seasickness rate), and ship maneuverability (rudder effectiveness, power response). However, existing ship attitude monitoring and roll risk warning technologies still suffer from the following problems in practical applications, failing to meet the demands for high precision, high reliability, and full-scenario navigation.

[0003] 1. Parameter acquisition is simplistic and redundant, resulting in both low accuracy and low reliability:

[0004] The current approach relies on a single sensor (such as a marine gyroscope that only measures roll / pitch angles), which is susceptible to interference from the marine environment (wind and wave impact, ship vibration) and generates data noise: when wind and waves suddenly occur, the roll angle of the gyroscope may jump by ±3°, and ship vibration causes fluctuations in the speed of the helical instrument, resulting in an error rate of over 5% for a single data source;

[0005] Although multi-source acquisition is introduced, the data acquisition interruption rate is high when the sensor fails (such as IMU disconnection under level 6 waves), and the fixed 100Hz sampling frequency causes a waste of computing power in calm sea conditions and insufficient dynamic capture in severe sea conditions.

[0006] 2. Parameter optimization is isolated and the coupling coefficient is fixed, resulting in poor adaptability to changing sea states:

[0007] Existing technologies employ "single-point filtering" (such as Kalman filtering to process the roll angle separately), ignoring the physical coupling relationship between attitude parameters: under oblique wave conditions, pitching intensifies rolling, and heave exceeding 2m leads to an increase in roll angular velocity, resulting in parameter deviation due to isolated optimization; although previous solutions considered the coupling relationship, they used fixed coupling coefficients and did not correlate real-time wave height and speed: when the wave height increases from 2m to 4m, the fixed coefficient increases the roll optimization error from 0.5% to 3%, and at a speed of 18kn, the pitch angular velocity deviation exceeds 2.5° / s, making it unsuitable for changing sea state scenarios. Summary of the Invention

[0008] This invention provides a method and system for determining the motion status of a ship using 3-axis and 6-attitude information, in order to overcome the aforementioned technical problems.

[0009] To achieve the above objectives, the technical solution of the present invention is as follows:

[0010] A method for determining a ship's motion status using 3-axis, 6-attitude information includes:

[0011] S1: Collect core data of the ship in three directions and six attitudes, environmental interference data and ship control data, and preprocess the collected data to obtain processed data; the core data in three directions and six attitudes are roll angle, pitch angle, roll angular velocity, pitch angular velocity, heave displacement and heave velocity.

[0012] S2: Based on the processed data, obtain the dynamic coupling coefficients and dynamic confidence scores for 3-axis 6-pose motion; the dynamic coupling coefficients are used to describe the coupling strength between each motion direction.

[0013] S3: Construct a 3-axis 6-pose weighted optimization objective function based on the processed data and dynamic confidence, construct constraints based on the 3-axis 6-pose dynamic coupling coefficient, solve the 3-axis 6-pose weighted optimization objective function under the constraints, and obtain the optimized 3-axis 6-pose attitude parameters.

[0014] S4: Construct a prediction model based on the LSTM network, input the optimized 3-axis 6-pose attitude parameters, dynamic coupling coefficients and dynamic confidence into the prediction model for training and updating, obtain an adaptive LSTM short-term prediction model for sway risk, and use the model to predict the ship sway risk level and corresponding attitude data for a preset duration in the future and output them.

[0015] S5: Based on the sway risk level, corresponding attitude data, ship type and current sea state, query the preset scenario-based decision rule base, generate structured decision suggestions, collect the execution effect of the structured decision suggestions to generate feedback data, and update the adaptive LSTM sway risk short-term prediction model based on the feedback data.

[0016] Furthermore, based on the processed data, the dynamic coupling coefficients of the three directions and six attitudes are obtained. These dynamic coupling coefficients include the roll-pitch coupling coefficient, the heave-roll angular velocity coupling coefficient, the pitch-heave coupling coefficient, the roll-heave velocity coupling coefficient, and the pitch-roll angular velocity coupling coefficient, as detailed below:

[0017] The roll-pitch coupling coefficient is constructed, and its expression is as follows:

[0018]

[0019] in, The roll-pitch coupling coefficient is... For the roll-pitch foundation coupling strength under calm sea conditions, For wave height weighting under roll-pitch coupling, For real-time wave height, The speed weighting under roll-pitch coupling. For speed, The saturation threshold for roll-pitch coupling;

[0020] The heave-roll angular velocity coupling coefficient is constructed, and its expression is as follows:

[0021]

[0022] in, The heave-roll angular velocity coupling coefficient is... The basic coupling strength of heave-roll angular velocity under calm sea conditions; The heave-roll angular velocity coupling saturation threshold; Wave height weighting under heave-roll angular velocity coupling;

[0023] The pitch-heave coupling coefficient is constructed, and its expression is as follows:

[0024]

[0025] in, The pitch-heave coupling coefficient is... For the foundation coupling strength under calm sea conditions, Wave height weighting under pitch-heave coupling; The saturation threshold for pitch-heave coupling;

[0026] The roll-heave velocity coupling coefficient is constructed, and its expression is as follows:

[0027]

[0028] in, The roll-heave velocity coupling coefficient. For the basic coupling strength of roll-heave velocity under calm sea conditions, For wave height weighting under roll-heave velocity coupling, The speed weighting under roll-heave speed coupling. The saturation threshold for roll-heave velocity coupling;

[0029] The pitch-roll angular velocity coupling coefficient is constructed, and its expression is as follows:

[0030]

[0031] in, The pitch-roll angular velocity coupling coefficient. The basic coupling strength of the pitch-roll angular velocity under calm sea conditions. The wave height weighting under the coupling of pitch and roll angular velocities. The speed weighting under the coupling of pitch and roll angular velocities. The saturation threshold for the pitch-roll angular velocity coupling.

[0032] Furthermore, dynamic confidence is obtained, which includes environmental interference coefficient and sensor confidence, as detailed below:

[0033] The environmental interference coefficient is constructed, and its expression is as follows:

[0034]

[0035] in, The environmental interference coefficient is... Increase the ratio of sensor error. This is the maximum interference threshold for the sensor;

[0036] The confidence score for 3-axis 6-pose projection is expressed as follows:

[0037]

[0038] in, For the 3-axis 6-position, the first Confidence of each attitude parameter For the first The sensor's factory-set basic accuracy for each parameter.

[0039] Furthermore, a 3-axis 6-pose weighted optimization objective function is constructed based on the processed data and dynamic confidence levels, and constraints are constructed based on the 3-axis 6-pose dynamic coupling coefficients, including:

[0040] After normalizing the processed data according to the preset range of each attitude parameter, a 3-axis 6-attitude weighted optimization objective function that minimizes the weighted error is constructed based on the normalized data and dynamic confidence level. Its expression is as follows:

[0041]

[0042] in, Describe the objective function. This represents the total number of 3-axis, 6-pose orientations. Indicates the first The normalized and optimized attitude parameters Indicates the first The processed data;

[0043] Constraints are constructed based on 3-axis 6-attitude dynamic coupling coefficients. These constraints include roll-pitch constraints, heave-roll angular velocity constraints, pitch-heave constraints, roll-roll velocity constraints, and pitch-roll angular velocity constraints, as detailed below:

[0044] The expression for the roll-pitch constraint is:

[0045]

[0046] in, This indicates the optimized roll angle. Indicates the processed roll angle. This indicates the optimized pitch angle. Indicates the processed pitch angle;

[0047] The expression for the heave-roll angular velocity constraint is:

[0048]

[0049] in, This represents the optimized ship roll rate. This indicates the processed ship roll rate. This represents the optimized heave displacement of the ship. This indicates the heave displacement of the ship after processing.

[0050] The expression for the pitch-heave constraint is:

[0051]

[0052] in, This indicates the optimized pitch angle. Indicates the processed pitch angle;

[0053] The expression for the roll-sway velocity constraint is:

[0054]

[0055] in, This represents the optimized heave velocity. This indicates the processed heave velocity;

[0056] The expression for the pitch-roll angular velocity constraint is:

[0057]

[0058] in, This represents the optimized pitch rate. This indicates the processed pitch angular velocity.

[0059] Furthermore, the SLSQP optimization algorithm is used to solve the 3-axis 6-pose weighted optimization objective function according to the constraints, and the optimized 3-axis 6-pose attitude parameters are obtained.

[0060] Furthermore, the prediction model includes a three-layer LSTM network and a Softmax classification output layer connected in sequence;

[0061] The three-layer LSTM network receives the optimized 3-axis 6-pose attitude parameters, dynamic coupling coefficients, and dynamic confidence scores, performs time-series processing on the input data, and obtains the predicted attitude data.

[0062] The Softmax classification output layer is used to receive the predicted attitude data, and maps the predicted attitude data to the corresponding four neurons through the weight matrix and bias. The Softmax activation function outputs a four-dimensional risk probability vector, which corresponds to four ship roll risk levels: safe, caution, danger, and emergency.

[0063] Furthermore, the adaptive LSTM swing risk short-term prediction model is updated based on the feedback data, including:

[0064] If the feedback data indicates that the prediction error of the adaptive LSTM swing risk short-term prediction model exceeds the preset threshold or the confidence level is lower than the preset threshold for n consecutive times, the first two layers of the three-layer LSTM network are frozen, and the third layer LSTM network and the output layer are updated using an incremental learning mechanism.

[0065] Based on the same inventive concept, a system for judging the motion status of a ship through 3-axis 6-attitude information is also proposed, including: a multi-source redundant data acquisition and synchronization module, a dynamic noise preprocessing module, a dynamic coupling collaborative optimization module, an adaptive risk prediction module, and a scenario-based decision-making and feedback closed-loop module.

[0066] The multi-source redundant data acquisition and synchronization module is used to acquire core data of the ship in three directions and six attitudes, environmental interference data and ship operation data at a dynamic sampling frequency, and to perform redundancy verification and time synchronization on the acquired data to obtain synchronized raw time-series data.

[0067] The dynamic noise preprocessing module is used to perform dynamic noise preprocessing on the synchronized original time-series data to obtain a clean dataset.

[0068] The dynamic coupling collaborative optimization module is used to obtain the dynamic coupling coefficients and dynamic confidence of 3-axis 6-pose based on the clean dataset, construct a 3-axis 6-pose weighted optimization objective function based on the clean dataset and dynamic confidence, construct constraints based on the 3-axis 6-pose dynamic coupling coefficients, solve the 3-axis 6-pose weighted optimization objective function under the constraints, and obtain the optimized 3-axis 6-pose attitude parameters.

[0069] The adaptive risk prediction module is used to input the optimized 3-axis 6-pose attitude parameters, dynamic coupling coefficient and dynamic confidence into the updated adaptive LSTM short-term sway risk prediction model for prediction, so as to obtain the sway risk level and corresponding attitude data for a preset duration in the future.

[0070] The scenario-based decision-making and feedback closed-loop module is used to query the scenario-based decision-making rule base according to the sway risk level, ship type and current sea state, generate structured decision suggestions, and generate feedback data based on the decision execution effect to update the adaptive risk prediction module.

[0071] Beneficial effects: This invention provides a method for determining the motion state of a ship using 3-axis and 6-attitude information, which has the following advantages:

[0072] 1. It solves the problems of reliability and efficiency in data acquisition, providing a high-quality, continuous, and synchronous data foundation for subsequent processing: Through multi-source acquisition and synchronization, continuous and time-synchronized raw multi-dimensional data with an error rate of less than 5% is obtained. This ensures the quality of input data from the source and avoids data interruption caused by the failure of a single sensor.

[0073] 2. It solves the problems of isolated parameter optimization and rigid, fixed coupling coefficients, achieving high-precision, highly adaptable attitude parameter optimization:

[0074] Based on multi-source data, dynamic coupling coefficients and confidence levels correlated with real-time sea conditions (wave height, speed) were constructed, and a constrained optimization objective function was built. High-precision collaborative optimization of 3-axis and 6 attitude parameters was achieved. This invention can adapt to different sea conditions, and its fast real-time solution speed provides accurate input for subsequent risk prediction. Attached Figure Description

[0075] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0076] Figure 1 A flowchart of the method for determining the motion state of a ship using 3-axis 6-attitude information provided by the present invention;

[0077] Figure 2 This is a data processing flowchart of the present invention;

[0078] Figure 3 A schematic diagram of a ship's 3-axis, 6-attitude early warning interface using this invention for early warning purposes;

[0079] Figure 4 This invention provides a system block diagram for determining a ship's motion status using 3-axis and 6-attitude information. Detailed Implementation

[0080] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0081] This embodiment provides a method for determining a ship's motion status using 3-axis, 6-attitude information, such as... Figure 1 As shown, it includes:

[0082] S1: Collect core data of the ship's three-axis six-attitudes, environmental interference data, and ship control data, and preprocess the collected data to obtain processed data; the core data of the three-axis six-attitudes are roll angle, pitch angle, roll angular velocity, pitch angular velocity, heave displacement, and heave velocity.

[0083] S2: Based on the processed data, obtain the dynamic coupling coefficients and dynamic confidence scores for 3-axis 6-pose motion; the dynamic coupling coefficients are used to describe the coupling strength between each motion direction.

[0084] S3: Construct a 3-axis 6-pose weighted optimization objective function based on the processed data and dynamic confidence, construct constraints based on the 3-axis 6-pose dynamic coupling coefficient, solve the 3-axis 6-pose weighted optimization objective function under the constraints, and obtain the optimized 3-axis 6-pose attitude parameters.

[0085] S4: Construct a prediction model based on the LSTM network, input the optimized 3-axis 6-pose attitude parameters, dynamic coupling coefficients and dynamic confidence into the prediction model for training and updating, obtain an adaptive LSTM short-term prediction model for sway risk, and use the model to predict the ship sway risk level and corresponding attitude data for a preset duration in the future and output them.

[0086] S5: Based on the sway risk level, corresponding attitude data, ship type and current sea state, query the preset scenario-based decision rule base, generate structured decision suggestions, collect the execution effect of the structured decision suggestions to generate feedback data, and update the adaptive LSTM sway risk short-term prediction model based on the feedback data.

[0087] Specifically, such as Figure 2As shown, the overall steps of this invention are "multi-source redundant acquisition → dynamic noise preprocessing → dynamic coupling collaborative optimization → adaptive risk prediction → scenario-based decision feedback". It addresses data source reliability issues through multi-source redundancy and dynamic sampling strategies; introduces a dynamic coupling coefficient and confidence-weighted optimization model associated with real-time sea conditions (wave height H, speed v) to solve the problems of isolated parameter optimization and poor adaptability of fixed coefficients; uses an LSTM-based adaptive time-series prediction model and incremental learning mechanism to achieve high-precision short-term early warning and solve the long-term model drift problem; constructs a scenario-based decision rule base bound to ship type and sea condition, and a "decision-effect" feedback loop to solve the problems of decision generalization and non-iterative optimization, addressing the shortcomings of existing technologies in reliability, adaptability, long-term stability, and scenario-based application, achieving high-precision monitoring of 3-axis 6-attitude parameters and accurate early warning of sway risk. Through data flow connection (the output of the previous step becomes the input of the next step), a new feedback link of "decision effect → model update" is added to ensure long-term adaptability to ship operation dynamics.

[0088] In a specific embodiment, the following scheme is used to collect core data of the ship in three directions and six attitudes, environmental disturbance data, and ship handling data, and to preprocess the collected data to obtain the processed data:

[0089] S11. Four types of data are collected through a distributed sensor network. The sampling frequency is switched based on the real-time wave height triggered by the weather station, as detailed below:

[0090] Attitude core data, including roll angle, pitch angle, roll angular velocity, and pitch angular velocity, are collected through primary and backup IMUs at the following sampling frequencies:

[0091] Wave height < 2m: 5Hz; 2m ≤ wave height ≤ 6m: 10Hz; Wave height > 6m: 15Hz;

[0092] Meanwhile, the accuracy of the data collected by the primary and backup IMUs is ±0.1° for angle and ±0.01° / s for angular velocity;

[0093] Heave data, including heave displacement and heave velocity, is collected using a heave meter at the same frequency as the IMU.

[0094] Environmental disturbance data, including wave height and wind speed, were collected through a weather station at the following frequency:

[0095] Wave height < 2m: 0.5Hz; Wave height ≥ 2m: 1Hz;

[0096] Ship control data, including speed, is collected via the ECS interface at a frequency of 1Hz.

[0097] S12. Preprocess the collected data as follows:

[0098] 1. Redundancy verification: When the deviation between the primary and backup IMU data is ≤ ±0.1°, the average value is taken; when the deviation is > ±0.1°, the judgment is made in combination with the angle between the wave direction and the heading (the primary IMU in the ship is given priority if the angle is > 60°). The system automatically switches to the backup equipment when the primary equipment fails.

[0099] 2. Time synchronization: GPS time synchronization (accuracy ±1ms) is used to configure a unified timestamp for all data, solving the problem of time sequence misalignment;

[0100] 3. Data storage: InfluxDB is used for storage;

[0101] 4. Perform data cleaning on the synchronized data to obtain processed data:

[0102] When the wave height is less than 2m, the sea is calm. Based on 30 sets of calm sea state experiments, noise elimination and data trend preservation are balanced, and the mean filtering method is finally used for data cleaning.

[0103] When the wave height is between 2m and 4m, it falls under medium to high sea state. The Kalman filter method is used, where the process noise covariance... ; Observation noise covariance Q is calculated based on sensor noise density, and R is obtained by statistically analyzing the variance of observation errors from 20 sets of medium-high sea state experiments.

[0104] When the wave height is greater than 4m, it is in a severe sea state. Through comparison of 15 sets of severe sea state experiments, the db4 wavelet basis has the best effect in preserving high-frequency features. Therefore, wavelet denoising is adopted.

[0105] This invention ensures data continuity through redundancy verification, balances acquisition quality and resource utilization through dynamic sampling, and provides data support for subsequent modules through time synchronization and standardized storage. It provides high-quality raw data with an error rate significantly lower than 5%, and the real-time wave height (H) and speed (v) collected provide the core data source for the calculation of dynamic coupling coefficients.

[0106] Accurate denoising is achieved for different sea state noise characteristics, removing interference (such as impulse noise and vibration noise) and missing values ​​from the original data, and outputting "clean dataset + data quality result" (data quality result = deviation between clean data and standard attitude instrument data). This solves the problem of "data distortion caused by environmental interference", avoids noise transmission that leads to deviation of subsequent optimization parameters, and further improves optimization accuracy.

[0107] In a specific embodiment, based on the processed data, the dynamic coupling coefficients and dynamic confidence scores for 3-axis 6-pose motion are obtained; the scheme for using the dynamic coupling coefficients to describe the coupling strength between each motion direction is as follows:

[0108] The dynamic coupling coefficients in the three directions and six attitudes include the roll-pitch coupling coefficient, the heave-roll angular velocity coupling coefficient, the pitch-heave coupling coefficient, the roll-roll velocity coupling coefficient, and the pitch-roll angular velocity coupling coefficient, as detailed below:

[0109] The roll-pitch coupling coefficient is constructed, and its expression is as follows:

[0110]

[0111] in, The roll-pitch coupling coefficient is... For the roll-pitch foundation coupling strength under calm sea conditions, For wave height weighting under roll-pitch coupling, For real-time wave height, The speed weighting under roll-pitch coupling. For speed, The roll-pitch coupling saturation threshold is defined as H=0.5m in calm sea conditions.

[0112] In this embodiment, The value of 0.35 is the statistical average of 100 sets of experiments (H=0.5m, v=5kn). The value of 0.04 is derived from experimental statistics; for every 1 meter increase in wave height, the impact increases by 0.04. It is 0.02. The value is 0.85, based on the "roll-pitch coupling saturation threshold" calibrated from 100 sets of ship attitude coupling experiments. In the experiment, The maximum measured value is 0.83, and 0.85 is taken as the safety upper limit—this covers the coupling peak under extreme operating conditions while avoiding optimization errors caused by data fluctuations. When H is less than 0.5m, it is calculated as 0.5m; when the calculated result is greater than 0.85, it is taken as 0.85.

[0113] The heave-roll angular velocity coupling coefficient is constructed, and its expression is as follows:

[0114]

[0115] in, The heave-roll angular velocity coupling coefficient is... The basic coupling strength of heave-roll angular velocity under calm sea conditions; The heave-roll angular velocity coupling saturation threshold; Wave height weighting under heave-roll angular velocity coupling;

[0116] In this embodiment, It is 0.65; It is 0.04. for Based on the "heave-roll angular velocity coupling saturation threshold" calibrated from 100 sets of ship attitude coupling experiments, in the experiment, The maximum measured value is 0.88 (corresponding to the extreme sea state of H=8m), and 0.9 is taken as the safety upper limit - which fully covers the coupling peak value of extreme working conditions, and also reserves a small amount of redundancy space to avoid optimization errors caused by fluctuations in wave height data; when H is less than 0.5m, it is calculated as 0.5m, and when H is greater than 4m, the calculation result is taken as 0.9.

[0117] The pitch-heave coupling coefficient is constructed, and its expression is as follows:

[0118]

[0119] in, The pitch-heave coupling coefficient is... For the foundation coupling strength under calm sea conditions, Wave height weighting under pitch-heave coupling; The saturation threshold for pitch-heave coupling;

[0120] In this embodiment, It is 0.4. It is 0.03. The value is 0.8; when H < 0.5m, it is calculated as 0.5m, and when the calculation result is greater than 0.8, it is taken as 0.8.

[0121] The roll-heave velocity coupling coefficient is constructed, and its expression is as follows:

[0122]

[0123] in, The roll-heave velocity coupling coefficient. For the basic coupling strength of roll-heave velocity under calm sea conditions, For wave height weighting under roll-heave velocity coupling, The speed weighting under roll-heave speed coupling. The saturation threshold for roll-heave velocity coupling;

[0124] In this embodiment, It is 0.3. It is 0.02. It is 0.01. The value is 0.75; H < 0.5m is calculated as 0.5m, v < 5kn is calculated as 5kn, and the value is 0.75 when the calculation result is greater than 0.75.

[0125] The pitch-roll angular velocity coupling coefficient is constructed, and its expression is as follows:

[0126]

[0127] in, This is the pitch-roll angular velocity coupling coefficient, achieving full coupling constraint in the angular velocity dimension. The basic coupling strength of the pitch-roll angular velocity under calm sea conditions. The wave height weighting under the coupling of pitch and roll angular velocities. The speed weighting under the coupling of pitch and roll angular velocities. The saturation threshold for the pitch-roll angular velocity coupling;

[0128] In this embodiment, It is 0.32. It is 0.025. It is 0.015. The value is 0.78; if H is less than 0.5m, it is calculated as 0.5m; if v is less than 5kn, it is calculated as 5kn; if the calculation result is greater than 0.78, it is taken as 0.78.

[0129] The dynamic confidence level is obtained, which includes the environmental interference coefficient and the sensor confidence level, as detailed below:

[0130] The environmental interference coefficient is constructed, and its expression is as follows:

[0131]

[0132] in, The environmental interference coefficient is... Increase the ratio of sensor error. This is the maximum interference threshold for the sensor;

[0133] In this embodiment, The value is 0.05, which is obtained by statistically analyzing the relationship between wave height and sensor error (for every 1m increase in wave height, the interference increases by 5%). The value of 0.3 represents the "maximum environmental interference limit that the sensor can withstand". In the experiment, the maximum measured value of Ke was 0.28 (corresponding to the extreme sea state of H=8m). 0.3 was taken as the safety limit.

[0134] The confidence score for 3-axis 6-pose projection is expressed as follows:

[0135]

[0136] in, For the 3-axis 6-position, the first The confidence level of each attitude parameter ensures that highly reliable data carries higher weight. For the first The sensor's factory-set basic accuracy for each parameter.

[0137] In this scheme, based on the physical laws of ship motion, the dynamic coupling coefficient used to describe the degree of mutual influence between two motion directions is first calculated, and then the confidence level of the parameters (i.e. the reliability of the data) is adjusted accordingly. The attitude influence in three directions is considered to solve the problem of poor adaptability between isolated parameter optimization and fixed coupling coefficient, thereby improving the accuracy of attitude parameter optimization. It provides highly reliable input for prediction, and the 0.1s solution time meets the real-time requirements. The constraints are in line with the physical laws of ship motion.

[0138] In a specific embodiment, a 3-axis 6-pose weighted optimization objective function is constructed based on the processed data and dynamic confidence level. Constraints are constructed based on the 3-axis 6-pose dynamic coupling coefficients. The 3-axis 6-pose weighted optimization objective function is solved under the constraints to obtain the optimized 3-axis 6-pose attitude parameters.

[0139] After normalizing the processed data according to the preset range of each attitude parameter, a 3-axis 6-attitude weighted optimization objective function that minimizes the weighted error is constructed based on the normalized data and dynamic confidence level. Its expression is as follows:

[0140]

[0141] in, Describe the objective function. This represents the total number of 3-axis, 6-pose orientations. Indicates the first The normalized and optimized attitude parameters Indicates the first The processed data;

[0142] Constraints are constructed based on 3-axis 6-attitude dynamic coupling coefficients. These constraints include roll-pitch constraints, heave-roll angular velocity constraints, pitch-heave constraints, roll-roll velocity constraints, and pitch-roll angular velocity constraints, as detailed below:

[0143] The expression for the roll-pitch constraint is:

[0144]

[0145] in, This indicates the optimized roll angle. Indicates the processed roll angle. This indicates the optimized pitch angle. Indicates the processed pitch angle;

[0146] The expression for the heave-roll angular velocity constraint is:

[0147]

[0148] in, This represents the optimized ship roll rate. This indicates the processed ship roll rate. This represents the optimized heave displacement of the ship. This indicates the heave displacement of the ship after processing. The "absolute value of optimization deviation" is used to measure the degree of optimization deviation of the heave displacement. The maximum allowable optimization deviation of the roll rate is defined as follows: the greater the change in heave displacement and the stronger the coupling strength, the greater the allowable optimization deviation of the roll rate (which conforms to the physical law that "heave drives roll rate").

[0149] The expression for the pitch-heave constraint is:

[0150]

[0151] in, This indicates the optimized pitch angle. Indicates the processed pitch angle;

[0152] The expression for the roll-sway velocity constraint is:

[0153]

[0154] in, This represents the optimized heave velocity. This indicates the processed heave velocity; The "absolute value of the optimization deviation" is used to measure the degree of deviation between the optimization result and the basic data, and to avoid the negative deviation from offsetting the effect.

[0155] The expression for the pitch-roll angular velocity constraint is:

[0156]

[0157] in, This represents the optimized pitch rate. This indicates the processed pitch angular velocity;

[0158] The SLSQP optimization algorithm is used to solve the 3-axis 6-pose weighted optimization objective function according to the constraints, and the optimized 3-axis 6-pose attitude parameters are obtained.

[0159] In a specific embodiment, a prediction model is constructed based on an LSTM network. The optimized 3-axis 6-pose attitude parameters, dynamic coupling coefficients, and dynamic confidence scores are input into the prediction model for training and updating, resulting in an adaptive LSTM short-term prediction model for roll risk. The scheme for using this model to predict and output the ship roll risk level and corresponding attitude data for a preset duration in the future is as follows:

[0160] The prediction model comprises a three-layer LSTM network and a Softmax classification output layer connected in sequence.

[0161] The three-layer LSTM network receives the optimized 3-axis 6-pose attitude parameters, dynamic coupling coefficients, and dynamic confidence scores, performs time-series processing on the input data, and obtains the predicted attitude data.

[0162] The Softmax classification output layer is used to receive the predicted attitude data, and maps the predicted attitude data to the corresponding four neurons through the weight matrix and bias, and outputs a four-dimensional risk probability vector through the Softmax activation function. The four-dimensional risk probability vector corresponds to four ship roll risk levels: safe, caution, danger and emergency.

[0163] The optimized 3-axis 6-pose attitude parameters, dynamic coupling coefficients, and dynamic confidence scores are input into the prediction model for training and updating, resulting in an adaptive LSTM short-term prediction model for swing risk.

[0164] LSTM networks are commonly used neural networks in predicting data. Those skilled in the art know how to use data to train and incrementally update LSTM networks. Therefore, this embodiment does not describe the specific training and update steps in detail, but only describes the update scenario and update method. Those skilled in the art can update the network according to the provided update method and obtain the required model.

[0165] If the feedback data indicates that the prediction error of the adaptive LSTM swing risk short-term prediction model exceeds the preset threshold or the confidence level is lower than the preset threshold for n consecutive times, the first two layers of the three-layer LSTM network are frozen, and the third layer LSTM network and the output layer are updated using the incremental learning mechanism.

[0166] Specifically, the adaptive update method is shown in Table 1:

[0167] Table 1

[0168]

[0169] The risk prediction levels are defined as shown in Table 2:

[0170] Table 2

[0171]

[0172] The adaptive LSTM short-term prediction model for ship sway risk obtained in this scheme can make short-term predictions of ship sway risk within a preset time period, improve the stability and accuracy of the prediction results, and provide a basis for subsequent scenario-based decision-making.

[0173] Incremental learning ensures long-term reliability, solving the problems of lagging risk prediction, lack of model updates, and insufficient long-term reliability.

[0174] In a specific embodiment, the following scheme involves querying a pre-defined scenario-based decision rule base based on the sway risk level, corresponding attitude data, ship type, and current sea state to generate structured decision suggestions, collecting the execution effect of the structured decision suggestions to generate feedback data, and updating the adaptive LSTM sway risk short-term prediction model based on the feedback data:

[0175] The list of scenario-based decision rules constructed in this scheme is shown in Table 3:

[0176] Table 3

[0177]

[0178] The feedback loop process is as follows:

[0179] 1) Decision effect monitoring: After the decision is executed, the deviation between the attitude parameters and the expected target is monitored in real time (such as whether the roll reduction has been reduced to the expected range); and the feedback data is used as a new dataset to update the model in real time.

[0180] 2) Feedback and Adjustment: When the deviation is greater than 10%, the decision is automatically adjusted, and the "Decision-Effect" data is stored in the case library;

[0181] 3) Rule base update: When the number of cases in a certain scenario is ≥20, the optimal decision suggestions are re-counted to achieve iterative optimization.

[0182] This solution transforms prediction results into structured decisions tailored to ship type and sea state, achieving closed-loop optimization through feedback, thus resolving the issues of "generalized decision-making" and "iteration without feedback." The early warning results are as follows: Figure 3 As shown, the structured recommendations cover common operating conditions, shortening operation response time and reducing operational risks.

[0183] This embodiment also provides a system for determining a ship's motion status based on its three-axis, six-attitude information, such as... Figure 4 As shown, it includes: a multi-source redundant data acquisition and synchronization module, a dynamic noise preprocessing module, a dynamic coupling collaborative optimization module, an adaptive risk prediction module, and a scenario-based decision-making and feedback closed-loop module;

[0184] The multi-source redundant data acquisition and synchronization module is used to acquire core data of the ship in three directions and six attitudes, environmental interference data and ship operation data at a dynamic sampling frequency, and to perform redundancy verification and time synchronization on the acquired data to obtain synchronized raw time-series data.

[0185] The dynamic noise preprocessing module is used to perform dynamic noise preprocessing on the synchronized original time-series data to obtain a clean dataset.

[0186] The dynamic coupling collaborative optimization module is used to obtain the dynamic coupling coefficients and dynamic confidence of 3-axis 6-pose based on the clean dataset, construct a 3-axis 6-pose weighted optimization objective function based on the clean dataset and dynamic confidence, construct constraints based on the 3-axis 6-pose dynamic coupling coefficients, solve the 3-axis 6-pose weighted optimization objective function under the constraints, and obtain the optimized 3-axis 6-pose attitude parameters.

[0187] The adaptive risk prediction module is used to input the optimized 3-axis 6-pose attitude parameters, dynamic coupling coefficient and dynamic confidence into the updated adaptive LSTM short-term sway risk prediction model for prediction, so as to obtain the sway risk level and corresponding attitude data for a preset duration in the future.

[0188] The scenario-based decision-making and feedback closed-loop module is used to query the scenario-based decision-making rule base according to the sway risk level, ship type and current sea state, generate structured decision suggestions, and generate feedback data based on the decision execution effect to update the adaptive risk prediction module.

[0189] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for determining a ship's motion status using 3-axis, 6-attitude information, characterized in that, include: S1: Collect 3-axis 6-attitude core data, environmental disturbance data, and ship control data, and preprocess the collected data to obtain processed data; the 3-axis 6-attitude core data are roll angle, pitch angle, roll angular velocity, pitch angular velocity, heave displacement, and heave velocity. S2: Based on the processed data, obtain the dynamic coupling coefficients and dynamic confidence scores for 3-axis 6-pose motion; the dynamic coupling coefficients are used to describe the coupling strength between each motion direction. S3: Construct a 3-axis 6-pose weighted optimization objective function based on the processed data and dynamic confidence, construct constraints based on the 3-axis 6-pose dynamic coupling coefficient, solve the 3-axis 6-pose weighted optimization objective function under the constraints, and obtain the optimized 3-axis 6-pose attitude parameters. S4: Construct a prediction model based on the LSTM network, input the optimized 3-axis 6-pose attitude parameters, dynamic coupling coefficients and dynamic confidence into the prediction model for training and updating, obtain an adaptive LSTM short-term prediction model for sway risk, and use the model to predict the ship sway risk level and corresponding attitude data for a preset duration in the future and output them. S5: Based on the sway risk level, corresponding attitude data, ship type and current sea state, query the preset scenario-based decision rule base, generate structured decision suggestions, collect the execution effect of the structured decision suggestions to generate feedback data, and update the adaptive LSTM sway risk short-term prediction model based on the feedback data.

2. The method for determining a ship's motion status using 3-axis 6-attitude information according to claim 1, characterized in that, Based on the processed data, the dynamic coupling coefficients of 3-axis and 6-attitude are obtained. The dynamic coupling coefficients of 3-axis and 6-attitude include the roll-pitch coupling coefficient, the heave-roll angular velocity coupling coefficient, the pitch-heave coupling coefficient, the roll-roll velocity coupling coefficient, and the pitch-roll angular velocity coupling coefficient, as detailed below: The roll-pitch coupling coefficient is constructed, and its expression is as follows: in, The roll-pitch coupling coefficient is... For the roll-pitch foundation coupling strength under calm sea conditions, For wave height weighting under roll-pitch coupling, For real-time wave height, The speed weighting under roll-pitch coupling. For speed, The saturation threshold for roll-pitch coupling; The heave-roll angular velocity coupling coefficient is constructed, and its expression is as follows: in, The heave-roll angular velocity coupling coefficient is... The basic coupling strength of heave-roll angular velocity under calm sea conditions; The heave-roll angular velocity coupling saturation threshold; Wave height weighting under heave-roll angular velocity coupling; The pitch-heave coupling coefficient is constructed, and its expression is as follows: in, The pitch-heave coupling coefficient is... For the foundation coupling strength under calm sea conditions, Wave height weighting under pitch-heave coupling; The saturation threshold for pitch-heave coupling; The roll-heave velocity coupling coefficient is constructed, and its expression is as follows: in, The roll-heave velocity coupling coefficient. For the basic coupling strength of roll-heave velocity under calm sea conditions, For wave height weighting under roll-heave velocity coupling, The speed weighting under roll-heave speed coupling. The saturation threshold for roll-heave velocity coupling; The pitch-roll angular velocity coupling coefficient is constructed, and its expression is as follows: in, The pitch-roll angular velocity coupling coefficient. The basic coupling strength of the pitch-roll angular velocity under calm sea conditions. The wave height weighting under the coupling of pitch and roll angular velocities. The speed weighting under the coupling of pitch and roll angular velocities. The saturation threshold for the pitch-roll angular velocity coupling.

3. The method for determining a ship's motion status using 3-axis 6-attitude information according to claim 2, characterized in that, The dynamic confidence level is obtained, which includes the environmental interference coefficient and the sensor confidence level, as detailed below: The environmental interference coefficient is constructed, and its expression is as follows: in, The environmental interference coefficient is... Increase the ratio of sensor error. This is the maximum interference threshold for the sensor; The confidence score for 3-axis 6-pose projection is expressed as follows: in, For the 3-axis 6-position, the first Confidence of each attitude parameter For the first The sensor's factory-set basic accuracy for each parameter.

4. The method for determining a ship's motion status using 3-axis 6-attitude information according to claim 3, characterized in that, A 3-axis, 6-pose weighted optimization objective function is constructed based on the processed data and dynamic confidence level. Constraints are constructed based on the 3-axis, 6-pose dynamic coupling coefficients, including: After normalizing the processed data according to the preset range of each attitude parameter, a 3-axis 6-attitude weighted optimization objective function that minimizes the weighted error is constructed based on the normalized data and dynamic confidence level. Its expression is as follows: in, Describe the objective function. This represents the total number of 3-axis, 6-pose orientations. Indicates the first The normalized and optimized attitude parameters Indicates the first The processed data; Constraints are constructed based on 3-axis 6-attitude dynamic coupling coefficients. These constraints include roll-pitch constraints, heave-roll angular velocity constraints, pitch-heave constraints, roll-roll velocity constraints, and pitch-roll angular velocity constraints, as detailed below: The expression for the roll-pitch constraint is: in, This indicates the optimized roll angle. Indicates the processed roll angle. This indicates the optimized pitch angle. Indicates the processed pitch angle; The expression for the heave-roll angular velocity constraint is: in, This represents the optimized ship roll rate. This indicates the processed ship roll rate. This represents the optimized heave displacement of the ship. This indicates the heave displacement of the ship after processing. The expression for the pitch-heave constraint is: in, This indicates the optimized pitch angle. Indicates the processed pitch angle; The expression for the roll-sway velocity constraint is: in, This represents the optimized heave velocity. This indicates the processed heave velocity; The expression for the pitch-roll angular velocity constraint is: in, This represents the optimized pitch rate. This indicates the processed pitch angular velocity.

5. The method for determining a ship's motion status using 3-axis 6-attitude information according to claim 1, characterized in that, The SLSQP optimization algorithm is used to solve the 3-axis 6-pose weighted optimization objective function according to the constraints, and the optimized 3-axis 6-pose attitude parameters are obtained.

6. The method for determining a ship's motion status using 3-axis 6-attitude information according to claim 1, characterized in that, The prediction model comprises a three-layer LSTM network and a Softmax classification output layer connected in sequence. The three-layer LSTM network receives the optimized 3-axis 6-pose attitude parameters, dynamic coupling coefficients, and dynamic confidence scores, performs time-series processing on the input data, and obtains the predicted attitude data. The Softmax classification output layer is used to receive the predicted attitude data, and maps the predicted attitude data to the corresponding four neurons through the weight matrix and bias. The Softmax activation function outputs a four-dimensional risk probability vector, which corresponds to four ship roll risk levels: safe, caution, danger, and emergency.

7. The method for determining a ship's motion status using 3-axis 6-attitude information according to claim 1, characterized in that, The adaptive LSTM swing risk short-term prediction model is updated based on the feedback data, including: If the feedback data indicates that the prediction error of the adaptive LSTM swing risk short-term prediction model exceeds the preset threshold or the confidence level is lower than the preset threshold for n consecutive times, the first two layers of the three-layer LSTM network are frozen, and the third layer LSTM network and the output layer are updated using an incremental learning mechanism.

8. A system for determining a ship's motion status using 3-axis 6-attitude information, used to implement the method for determining a ship's motion status using 3-axis 6-attitude information as described in claim 1, characterized in that, include: Multi-source redundant data acquisition and synchronization module, dynamic noise preprocessing module, dynamic coupling collaborative optimization module, adaptive risk prediction module, and scenario-based decision-making and feedback closed-loop module; The multi-source redundant data acquisition and synchronization module is used to acquire core data of the ship in three directions and six attitudes, environmental interference data and ship operation data at a dynamic sampling frequency, and to perform redundancy verification and time synchronization on the acquired data to obtain synchronized raw time-series data. The dynamic noise preprocessing module is used to perform dynamic noise preprocessing on the synchronized original time-series data to obtain a clean dataset. The dynamic coupling collaborative optimization module is used to obtain the dynamic coupling coefficients and dynamic confidence of 3-axis 6-pose based on the clean dataset, construct the 3-axis 6-pose weighted optimization objective function that minimizes the weighted error based on the clean dataset and dynamic confidence, construct the constraint conditions based on the 3-axis 6-pose dynamic coupling coefficients, solve the 3-axis 6-pose weighted optimization objective function under the constraint conditions, and obtain the optimized 3-axis 6-pose attitude parameters. The adaptive risk prediction module is used to input the optimized 3-axis 6-pose attitude parameters, dynamic coupling coefficient and dynamic confidence into the updated adaptive LSTM short-term sway risk prediction model for prediction, so as to obtain the sway risk level and corresponding attitude data for a preset duration in the future. The scenario-based decision-making and feedback closed-loop module is used to query the scenario-based decision-making rule base according to the sway risk level, ship type and current sea state, generate structured decision suggestions, and generate feedback data based on the decision execution effect to update the adaptive risk prediction module.

Images