Hybrid prediction method and system for dynamic response of ocean dynamic cable

Abstract

The invention provides a hybrid prediction method and system for the dynamic response of a marine dynamic cable, and relates to the technical field of computer data processing, and the system comprises a data collection and processing module which is used for collecting sea condition multi-source data, carrying out the time synchronization, coordinate unification and linear interpolation complementation of the sea condition multi-source data, and outputting preprocessed data; the sea condition multi-source data comprises sea condition observation data, platform attitude, operation state data and dynamic cable data, and according to the hybrid prediction method and system for the dynamic response of the ocean dynamic cable, time synchronization, coordinate unification and linear interpolation supplementation are carried out on the sea condition multi-source data, and the reliability evaluation and evidence fusion technology is utilized to predict the dynamic response of the ocean dynamic cable. A multi-source feature set is mapped into prediction evidence representation, reduction calculation and conflict suppression are carried out, information of different data sources is fused, and accuracy and reliability of a prediction result are ensured.

Description

Technical Field

[0001] This invention relates to the field of computer data processing technology, specifically to a hybrid prediction method and system for the dynamic response of marine dynamic cables. Background Technology

[0002] Currently, response prediction technologies for marine dynamic cables are mainly divided into two categories: numerical simulation methods and data-driven methods. Numerical simulation methods describe the motion and response of dynamic cables by establishing complex dynamic models; commonly used techniques include the finite element method and multibody dynamics. By considering the coupling effects of factors such as ocean waves, currents, wind loads, and platform motion, relatively accurate analysis of the dynamic cable's response can be achieved. Data-driven methods rely more on historical monitoring data and use algorithms such as machine learning and neural networks for dynamic response prediction, and have already been applied in some marine engineering projects.

[0003] However, existing models struggle to accurately reflect changing sea conditions and platform operational status, leading to reduced prediction accuracy. Furthermore, current technologies cannot effectively integrate multi-source heterogeneous data, such as sea state observations, platform attitude data, and sensor data, making it difficult to achieve stable and reliable predictions in practical applications. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a hybrid prediction method and system for the dynamic response of marine dynamic cables. The technical problem this invention aims to solve is: how to address the insufficient prediction accuracy of existing technologies when facing changes in sea state and platform operation status by introducing a credibility assessment and evidence fusion model.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a hybrid prediction system for the dynamic response of marine dynamic cables, comprising: Data acquisition and processing module: used to acquire multi-source sea state data, and to perform time synchronization, coordinate unification and linear interpolation on the multi-source sea state data, and output preprocessed data. The multi-source sea state data includes sea state observation data, platform attitude, operation status data and dynamic cable data. Feature extraction module: used to extract sea state features, platform status features and dynamic cable response features from the preprocessed data to form a multi-source feature set; The credibility assessment and evidence fusion module includes a credibility parameter generation unit, an evidence mapping unit, a conflict degree calculation unit, and an evidence combination unit. The credibility parameter generation unit calculates the credibility of the preprocessed data and generates credibility parameters. The evidence mapping unit maps the multi-source feature set into an evidence representation for prediction and converts the evidence representation into a basic probability allocation function under a unified predictive evidence framework. The evidence mapping unit performs a reduction calculation on the basic probability allocation function based on the credibility parameters to obtain a reduced evidence set. The conflict degree calculation unit calculates the evidence conflict degree of the evidence set and performs conflict suppression processing on the evidence set that causes conflict when the evidence conflict degree exceeds a preset threshold to form a stable evidence set. The conflict suppression processing includes conflict reallocation and weight reduction. The evidence combination unit performs evidence combination on the stable evidence set and outputs a fused evidence representation. Hybrid prediction module: used to perform prediction calculations on the fused evidence representation and output the dynamic response prediction results of the dynamic cable.

[0006] Preferably, the credibility parameter generation unit is used to perform residual calculation based on the preprocessed data within a preset sliding time window to obtain data quality indicators, which include data integrity, time delay, noise level, drift index and cross-source consistency test results.

[0007] Preferably, the credibility parameter generation unit normalizes the data quality indicators respectively, and obtains the data quality score by weighted summation according to preset weights. The credibility parameter is obtained by the data quality score through a monotonic mapping function. The credibility parameter is limited to the range of [0,1] so that the credibility parameter decreases when the data quality score decreases.

[0008] Preferably, the unified predictive evidence framework is used to predetermine a set of predictive propositions for at least one predictive target of the dynamic response of the dynamic cable. The set of predictive propositions consists of several mutually exclusive predictive propositions obtained by discretizing the value range of the predictive target. The set of predictive propositions is used as the domain of the basic probability allocation function. The evidence representation is converted into the basic probability allocation function according to the domain.

[0009] Preferably, the reduction calculation uses a reduction coefficient, which is a real number that is monotonically increasing with the confidence parameter and has a value range of [0, 1].

[0010] Preferably, the calculation of the degree of conflict of evidence includes the following steps: S61. Using the basic probability allocation function corresponding to each sea state multi-source data as input, combine the basic probability allocation functions of any two sea state multi-source data on the same prediction proposition set in pairs. S62. Calculate the product of the assigned values ​​of the basic probability assignment functions for the mutually exclusive predictive propositions in the predicted proposition set and sum them to obtain the evidence conflict degree. The larger the value of the evidence conflict degree, the higher the degree of incompatibility between the reduced evidence sets.

[0011] Preferably, the conflict suppression process includes, when the degree of evidence conflict exceeds a preset threshold, performing a weight reduction process on the assignment of the basic probability allocation function corresponding to the mutually exclusive prediction proposition in the evidence set that caused the conflict, and allocating the weighted part of the assignment to prediction propositions compatible with the evidence set to obtain the stable evidence set.

[0012] Preferably, the evidence combination is based on the basic probability allocation function, and according to the preset evidence combination rules, the values ​​of the basic probability allocation function of the same prediction proposition on the same prediction proposition set are weighted and summed, and the weighted sum is normalized to ensure that the sum of the values ​​of the basic probability allocation function is 1, so as to obtain the fused evidence representation.

[0013] Preferably, the fused evidence representation includes sea state feature representation, platform state feature representation, and dynamic cable response feature representation. The prediction calculation inputs the fused evidence representation into a preset prediction function, evaluates the prediction function, and obtains the dynamic cable dynamic response prediction result. The preset function is a nonlinear function, and the dynamic cable dynamic response prediction result includes time-series prediction results of tension, displacement, velocity, and curvature.

[0014] A hybrid prediction method for the dynamic response of marine dynamic cables includes: S1. Collect multi-source sea state data, and perform time synchronization, coordinate unification and linear interpolation to complete the multi-source sea state data to obtain preprocessed data. The multi-source sea state data includes sea state observation data, platform attitude data, operation status data and dynamic cable data. S2. Extract sea state features, platform status features, and dynamic cable response features from the preprocessed data to form a multi-source feature set; S3. Calculate the confidence parameters of the multi-source data for each sea state based on the preprocessed data. Under a unified predictive evidence framework, map the multi-source feature set into an evidence representation for prediction and convert it into a basic probability allocation function. Perform a reduction calculation on the basic probability allocation function according to the confidence parameters to obtain a reduced evidence set. Calculate the evidence conflict degree based on the reduced evidence set. When the evidence conflict degree exceeds a preset threshold, perform conflict suppression processing on the evidence set that causes conflict to form a stable evidence set. Perform evidence combination on the stable evidence set to obtain a fused evidence representation. S4. Input the fused evidence representation into a preset prediction function for prediction calculation, and output the prediction result of the dynamic response of the marine dynamic cable. The prediction function adopts a nonlinear regression function, and the prediction result of the dynamic response of the marine dynamic cable includes the time-series prediction results of tension, displacement, velocity and curvature.

[0015] This invention provides a hybrid prediction method and system for the dynamic response of marine dynamic cables. It has the following beneficial effects:

[0016] This hybrid prediction method and system for the dynamic response of marine dynamic cables synchronizes time, unifies coordinates, and completes linear interpolation of multi-source sea state data. It uses credibility assessment and evidence fusion techniques to map multi-source feature sets into predictive evidence representations, performs reduction calculations and conflict suppression, and integrates information from different data sources to ensure the accuracy and reliability of prediction results.

[0017] The evidence fusion and conflict suppression mechanism adopted reduces the impact of conflict on the prediction results by weighting and redistribution when the evidence conflict is too high. Finally, the fused evidence representation is input into the nonlinear prediction function to realize the time series prediction of dynamic cables, including accurate prediction of key indicators such as tension, displacement, velocity and curvature, thus improving the stability and accuracy of the system. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the system structure of the present invention; Figure 2 This is a flowchart of the data acquisition and processing process of the present invention; Figure 3 This is a flowchart illustrating the generation process of the credibility parameters for this invention. Figure 4 This is a flowchart of the evidence fusion and conflict resolution process of the present invention; Figure 5 This is a flowchart of the prediction method of the present invention. Detailed Implementation

[0019] 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, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] like Figure 1-5 As shown, this embodiment of the invention provides a hybrid prediction system for the dynamic response of marine dynamic cables, comprising: Data acquisition and processing module: used to acquire multi-source sea state data, and to perform time synchronization, coordinate unification and linear interpolation to complete the multi-source sea state data, and output preprocessed data. The multi-source sea state data includes sea state observation data, platform attitude, operation status data and dynamic cable data.

[0021] Feature extraction module: used to extract sea state features, platform status features and dynamic cable response features from preprocessed data to form a multi-source feature set.

[0022] Sea state observation data refers to data acquired related to the marine environment, including wave height, seawater temperature, ocean current speed, and wind speed. Platform attitude data refers to changes in the platform's attitude during offshore operations, including pitch, roll, and yaw angles. Operational status data includes operational status information for the platform or dynamic cable. Dynamic cable data involves data related to the marine dynamic cable itself, including cable tension, displacement, velocity, and curvature.

[0023] The credibility assessment and evidence fusion module includes a credibility parameter generation unit, an evidence mapping unit, a conflict degree calculation unit, and an evidence combination unit. The credibility parameter generation unit calculates the credibility of preprocessed data and generates credibility parameters. The evidence mapping unit maps multi-source feature sets to evidence representations for prediction and converts these representations into basic probability allocation functions within a unified predictive evidence framework. The evidence mapping unit performs reduction calculations on the basic probability allocation functions based on the credibility parameters to obtain a reduced evidence set. The conflict degree calculation unit calculates the evidence conflict degree of the evidence set and performs conflict suppression processing on the conflict-inducing evidence set when the conflict degree exceeds a preset threshold, forming a stable evidence set. Conflict suppression processing includes conflict reallocation and weight reduction. The evidence combination unit performs evidence combination on the stable evidence set and outputs a fused evidence representation. The credibility parameter generation unit performs residual calculations based on preprocessed data within a preset sliding time window to obtain data quality indicators, including data integrity, time delay, noise level, drift index, and cross-source consistency test results. The credibility parameter generation unit normalizes the data quality indicators and sums them according to preset weights to obtain the data quality score. The credibility parameter is obtained from the data quality score through a monotonic mapping function. The credibility parameter is limited to the range [0,1] so that the credibility parameter decreases when the data quality score decreases. A unified predictive evidence framework is used to predetermine a set of predictive propositions for at least one predictive target of the dynamic response of the dynamic cable. The set of predictive propositions consists of several mutually exclusive predictive propositions obtained by discretizing the value range of the predictive target. The set of predictive propositions is used as the domain of the basic probability allocation function. The evidence representation is converted into the basic probability allocation function according to the domain. The reduction calculation uses a reduction coefficient, which is a real number that is monotonically increasing with the credibility parameter and has a value range of [0,1]. The calculation of the degree of evidence conflict includes the following steps:

[0024] S61. Using the basic probability allocation function corresponding to each sea state multi-source data as input, perform pairwise combinations of the basic probability allocation functions of any two sea state multi-source data on the same prediction proposition set.

[0025] S62. Calculate the product of the assigned values ​​of the basic probability assignment functions for mutually exclusive predictive propositions in the predictive proposition set and sum them to obtain the degree of evidence conflict. The larger the value of the degree of evidence conflict, the higher the degree of incompatibility between the reduced evidence sets.

[0026] Conflict suppression processing involves reducing the weighting of the basic probability allocation functions corresponding to mutually exclusive prediction propositions in the evidence set that cause conflict when the degree of evidence conflict exceeds a preset threshold. The reduced-weighted values ​​are then allocated to prediction propositions compatible with the evidence set, resulting in a stable evidence set. Evidence combination, based on the basic probability allocation function, performs a weighted summation of the basic probability allocation functions for the same prediction proposition on the same prediction proposition set according to preset evidence combination rules. The weighted summation result is then normalized to ensure that the sum of the basic probability allocation function values ​​is 1, resulting in a fused evidence representation. The fused evidence representation includes sea state characteristic representation, platform state characteristic representation, and dynamic cable response characteristic representation. Prediction calculation inputs the fused evidence representation to a preset prediction function, evaluates the prediction function, and obtains the dynamic cable dynamic response prediction result. The preset function is a nonlinear function, and the dynamic cable dynamic response prediction result includes time-series prediction results for tension, displacement, velocity, and curvature.

[0027] Hybrid prediction module: used to perform prediction calculations on the fused evidence representation and output the prediction results of the dynamic response of the dynamic cable.

[0028] A hybrid prediction method for the dynamic response of marine dynamic cables includes: S1. Collect multi-source sea state data, and perform time synchronization, coordinate unification and linear interpolation to complete the multi-source sea state data to obtain preprocessed data. The multi-source sea state data includes sea state observation data, platform attitude data, operational status data and dynamic cable data.

[0029] S2. Extract sea state features, platform status features, and dynamic cable response features from the preprocessed data to form a multi-source feature set.

[0030] S3. Calculate the credibility parameters of multi-source data for each sea state based on preprocessed data. Under a unified predictive evidence framework, map the multi-source feature set to the evidence representation for prediction and convert it into a basic probability allocation function. Perform a reduction calculation on the basic probability allocation function according to the credibility parameters to obtain the reduced evidence set. Calculate the evidence conflict degree based on the reduced evidence set. When the evidence conflict degree exceeds a preset threshold, perform conflict suppression processing on the evidence set that causes conflict to form a stable evidence set. Perform evidence combination on the stable evidence set to obtain a fused evidence representation.

[0031] S4. Input the fused evidence representation into the preset prediction function for prediction calculation, and output the dynamic response prediction results of the marine dynamic cable. The dynamic response prediction results of the marine dynamic cable include the time-series prediction results of tension, displacement, velocity and curvature.

[0032] The process of transforming evidence representations into basic probability assignment functions based on the domain involves discretizing the value range of the prediction target to construct a domain containing mutually exclusive prediction propositions. Features from different data sources are mapped to evidence representations, and these evidence representations are mapped to the set of prediction propositions, thus transforming them into basic probability assignment functions. Based on this, the basic probability assignment functions are reduced according to a confidence parameter to adjust the influence of the evidence. Through evidence combination and conflict suppression, a stable fused evidence representation is generated, which is then input into a nonlinear regression function for calculation to obtain the final dynamic response prediction result for the dynamic cable.

[0033] The reduction calculation refers to the weighted attenuation of the basic probability allocation function based on the credibility parameters of each data source before multi-source evidence fusion. This reduces the contribution of low-credibility evidence to the overall result, resulting in a set of evidence with credibility correction, which is then used for subsequent conflict detection and evidence combination.

[0034] By calculating the residuals between the preprocessed data and the corresponding reference values ​​over a sliding time scale, the bias in the data is reflected. Sliding time window selection: Choose an appropriate time window for calculation, typically 10 to 30 seconds, with the specific length chosen based on the data sampling frequency and dynamic response characteristics. Within each sliding time window, the data deviation is quantified by calculating the residuals.

[0035] Selection of reference values: Historical mean: The mean within a sliding window, used as a reference when data changes are relatively small.

[0036] Predicted value: The value is obtained by using the model prediction or the prediction result of the previous time point as a reference value. It is suitable for situations where the data changes significantly.

[0037] Fitted curve: The expected value obtained by fitting a curve using a polynomial or smoothing filter, suitable when the data exhibits a specific trend.

[0038] The residual calculation formula is as follows: The preprocessed data is collected in real time, while the reference value can be obtained through the selection method mentioned above. The residual reflects the deviation between the actual value and the expected value.

[0039] Statistical analysis of the residuals quantifies the deviation characteristics of the preprocessed data relative to the reference state. Mean residual: Calculate the mean of the residuals within the sliding time window to reflect the systematic bias of the data. If the mean of the residuals is close to zero, it indicates that there is no significant systematic bias in the data.

[0040] Standard deviation or variance: Calculate the standard deviation or variance of the residuals to reflect the volatility of the data. A larger standard deviation indicates greater data volatility, which may indicate stronger noise or instability.

[0041] Trend Analysis: By analyzing the trend of residual changes within a sliding window, we can determine whether there is long-term drift or offset. Calculating the linear trend of the residuals, if the trend shows a monotonically increasing or decreasing trend, indicates that the system may have systematic drift.

[0042] The correlation between deviation characteristics and subsequent statistical characteristics is strong, specifically: Mean residuals: reflect the systematic bias of the data; a mean close to zero indicates no systematic bias.

[0043] Standard deviation and residuals: These reflect the volatility of the data. A larger standard deviation indicates that the data is more volatile and may contain noise or be unstable.

[0044] Trend analysis: Reflects the long-term trend of data. If a monotonic trend exists, it indicates that the system may have a bias or drift.

[0045] The aforementioned statistical characteristics collectively determine the stability, reliability, and consistency of the data.

[0046] After quantifying the bias characteristics, the statistical features of the residuals are transformed into computable data quality indicators that characterize the stability, reliability, and consistency of the data: Normalization: Statistical characteristics are normalized to ensure they have the same units of measurement for consistent comparison. The following standardization formula is used:

[0047] Weighted summation: This involves summing multiple standardized features with weights to obtain a comprehensive data quality score. The weights of each feature are set according to their impact on data quality. The weighting formula is as follows:

[0048] .

[0049] in, , , The weighting coefficients for the corresponding features are set based on the reliability of the data source, the characteristics of data changes, and the degree of impact on the overall data quality.

[0050] Mapping to a confidence parameter: The data quality score is converted into a confidence parameter using a monotonic mapping function, ensuring that the confidence level varies within the range [0, 1]. The monotonic mapping function guarantees that a higher data quality score corresponds to a higher confidence parameter, and vice versa. The mapping function is shown below:

[0051] The mapping function is chosen based on actual needs and can be either a linear function or a sigmoid function.

[0052] Data quality scores and credibility parameters provide an objective basis for subsequent credibility modeling. Through these quality indicators, the system can comprehensively evaluate the reliability of different data sources and data streams, providing support for subsequent decision-making.

[0053] Data quality indicators such as data integrity, time delay, noise level, drift index, and cross-source consistency are normalized with unified dimensions. Preset weights are set according to the degree of influence of each indicator on the reliability of prediction. The normalized indicators are then weighted and synthesized to form a data quality score that represents the overall reliability level of the data source.

[0054] The degree of evidence conflict is calculated by accumulating the product of the basic probability assignment functions corresponding to different data sources on mutually exclusive prediction propositions. It is used to quantitatively characterize the degree of compatibility between the reduced evidence sets and serves as a trigger criterion for conflict suppression processing.

[0055] Weight reduction processing refers to the process of reducing the weight of the basic probability allocation functions corresponding to mutually exclusive predicted propositions in the evidence set that cause conflict when the degree of conflict exceeds a preset threshold. This reduces the influence of mutually exclusive predicted propositions in the evidence combination process. The reduced values ​​due to weight reduction are then redistributed to the basic probability allocation functions corresponding to predicted propositions compatible with the evidence set. This ensures that the redistributed basic probability allocation functions maintain consistency constraints on the predicted proposition set, forming a stable evidence set after conflict suppression.

[0056] Example 2 This embodiment presents a hybrid prediction method and system based on the dynamic response of marine dynamic cables. By collecting and processing multi-source data from the marine environment, a hybrid prediction method is employed to accurately predict the dynamic response of marine dynamic cables. The specific implementation is as follows:

[0057] 1. Multi-source sea state data acquisition On December 15, 2025, sensors were used to monitor the marine environment in real time, with data collected hourly. Data acquisition included sea state observations, platform attitude data, operational status data, and dynamic cable data, as detailed below:

[0058] Sea state observation data: Wave height, wind speed, and ocean current velocity data acquired through marine meteorological platforms. Data was collected in the sea area located at 32°N, 120°E. According to the data collected by the ocean buoy at 03:00, the wave height was 10.5m, the ocean current speed was 1.8m / s, and the wind speed monitored by the meteorological satellite system was 12m / s.

[0059] Platform posture data: The platform's attitude data is collected in real time by an inertial measurement unit installed on the platform: the data is collected on an offshore platform with a tilt angle of 2.5° and a rotation angle of 5°.

[0060] Job status data: Job status information provided by sensors on the job platform. The marine platform control system collected data showing an operating depth of 100m and an equipment load of 75%.

[0061] Dynamic cable data: Data provided by the dynamic cable sensor system: The dynamic cable monitoring system detected the following data: tension 3.5kN, displacement 0.25m, velocity 0.5m / s, and curvature 1.2m.

[0062] 2. Data Preprocessing The collected data undergoes preprocessing to ensure its accuracy, consistency, and completeness. The data preprocessing steps are as follows:

[0063] Time synchronization: All data needs to be synchronized across different devices to ensure that timestamps from different data sources are aligned.

[0064] Because different sensors may have slight time discrepancies, all data is adjusted uniformly according to the device acquisition time. The system's built-in time synchronization algorithm is used to adjust the data collected by each device to a unified timestamp.

[0065] Coordinate unification: Different data sources may use different coordinate systems, so it is necessary to unify all data to the same coordinate system.

[0066] The coordinate systems for sea state observation data and platform attitude data are both geographic coordinate systems, while dynamic cable data is converted according to the platform's spatial coordinate system to ensure data compatibility.

[0067] Linear interpolation completion: In the actual data collection process, some data is missing.

[0068] When the sea state data at 01:00 is missing, the wave height collected at 00:00 is 9.8m, and the wave height collected at 02:00 is 10.9m, linear interpolation is used to complete the data. This is achieved by calculating the linear interpolation between the preceding and following data.

[0069] The interpolation formula is: in, , , and t represents the timestamps of the preceding and following data, respectively, and t represents the timestamp of the missing data.

[0070] The calculated missing wave height is 10.35m.

[0071] Noise Removal: Noise removal is performed on the data. A low-pass filter is used to smooth the acquired wave height, wind speed, tension, and other data to reduce the impact of measurement noise.

[0072] Filtering algorithms improve data quality by eliminating high-frequency noise signals and retaining only valid signals.

[0073] Data standardization: To avoid the impact of different units of measurement between different data sources, all data is standardized.

[0074] 3. Feature extraction After data preprocessing, meaningful features are extracted from sea state observation data, platform attitude data, operational status data, and dynamic cable data to provide a basis for subsequent prediction calculations. The feature extraction process is crucial to the accuracy of the model.

[0075] Sea state feature extraction: The following features are extracted from the preprocessed sea state observation data: The maximum wave height was 10.5 m, the average wind speed was 12 m / s, and the average ocean current speed was 1.8 m / s. Platform state feature extraction: Features extracted from platform posture data: maximum tilt angle is 2.5°, maximum rotation angle is 5°.

[0076] Dynamic cable response feature extraction: The following features are extracted from the dynamic cable data, directly taken from the preprocessed data: The maximum tension is 3.5 kN, the maximum displacement is 0.25 m, the maximum speed is 0.5 m / s, and the maximum curvature is 1.2 m.

[0077] 4. Form a multi-source feature set All extracted features are combined to form a complete multi-source feature set. Features from each data source provide key inputs for subsequent evidence fusion, credibility assessment, and dynamic response prediction. Based on these features, the system can make more accurate dynamic response predictions based on features from different data sources.

[0078] Through the above steps, accurate time-series predictions of the dynamic responses of marine dynamic cables, such as tension, displacement, velocity, and curvature, are made, improving the accuracy of dynamic cable monitoring in marine operations, providing reliable support for operational decisions, and ensuring the safety and stability of operations.

[0079] Example 3 This embodiment presents a hybrid prediction method and system based on the dynamic response of marine dynamic cables. Through a credibility assessment and evidence fusion module, it utilizes multi-source data from marine operation platforms to accurately predict the tension, displacement, velocity, and curvature responses of the dynamic cables, providing reliable decision support for marine operations. The specific implementation method is as follows:

[0080] 1. Data Collection This embodiment applies to a dynamic cable dynamic response prediction system in an offshore platform, using a credibility assessment and evidence fusion module. The following is the actual collected data:

[0081] Sea state data acquisition: Wave height: Real-time data obtained through the ocean buoy system; the wave height was 2.5 meters. Data from sea state observations on March 12, 2024.

[0082] Wave period: The collected wave period is 8 seconds, the data source is the same buoy system, and the timestamp is March 12, 2024.

[0083] Platform posture data collection: Platform tilt angle: Data collected via accelerometer and gyroscope sensors mounted on the platform. The platform tilt angle changed from 0° to 5° over 30 minutes.

[0084] Angular velocity: The angular velocity measured by the gyroscope sensor is 0.1° / second, and the time period is consistent with the change of the platform tilt angle.

[0085] Dynamic cable data acquisition: Tension: The dynamic cable tension sensor monitors the tension in real time via an optical fiber sensor. Between 12:00 and 12:30 on March 12, 2024, the tension increased from 1000N to 1200N.

[0086] Displacement: Fiber optic sensors are used to measure the displacement of the dynamic cable. The initial displacement is 10 meters, which increases to 12 meters over time.

[0087] 2. Credibility Parameter Generation Based on the above data, a credibility assessment was conducted, and a credibility parameter for each data source was calculated. The detailed calculation process and basis are as follows.

[0088] Data integrity: Data integrity is used to represent the validity of a data source, and the calculation formula is as follows: In this embodiment, the sea state data comes from an ocean buoy system. No data loss or anomalies were observed during the system's observation process. The expected data volume and the actual effective data volume accounted for 90%. Therefore, the sea state data integrity is 0.9.

[0089] Time delay: Time delay represents the time difference between data acquisition and processing. The calculation of time delay involves the system's sampling frequency and the data transmission and processing delays. The normalized formula for time delay is:

[0090] System testing and log analysis revealed a latency of 2 seconds. There is a delay when the platform's sensors collect data and send it to the central processing system; the maximum tolerable latency is set at 5 seconds. Therefore, the normalized value for the time delay is 0.4.

[0091] Noise level: The noise level measures random disturbances or uncertainties in the data.

[0092] The noise value of the dynamic cable tension sensor was obtained through actual measurement. The noise of the dynamic cable tension data was 10N, and the noise threshold was set to 50N, therefore the noise level was 0.2. This indicates that the noise level in the data is low because it is within the noise threshold range.

[0093] Drift Metric: The drift metric indicates the trend of data changes over time. The calculation formula is: Analysis of the platform's attitude data shows that the platform's tilt angle drift is 0.5° / second. The drift threshold, set based on the maximum rate of change in the platform's operating environment, is 1° / second. Therefore, the drift index is 0.5, indicating that the platform's drift level is relatively stable and below the maximum permissible drift threshold.

[0094] Cross-source consistency test: Cross-source consistency is used to assess the similarity between different data sources. Consistency is determined by calculating the correlation between two data sources.

[0095] The correlation between sea state data and dynamic cable data was obtained through statistical analysis or experiments, with a correlation of 0.85 and a cross-source consistency of 0.85. This indicates a strong correlation between the two data sources and a high degree of similarity between them.

[0096] Data quality score calculation: The data quality score is obtained by weighted summation based on the above quality indicators.

[0097] Weight: Since data integrity is directly related to the reliability and effectiveness of the model, it is given a high weight, indicating that it has a greater impact on the final credibility assessment.

[0098] Time delay has some impact on prediction accuracy, but in some application scenarios, the impact of time delay is relatively small.

[0099] Noise levels have a significant impact on data accuracy, but since systems typically perform noise filtering and data cleaning, the degree of noise impact is usually controllable.

[0100] Because drift has a relatively small impact on short-term forecasts, its weight is set low. Drift is typically mitigated through calibration and regular maintenance, therefore it has a small weight in the overall assessment.

[0101] Cross-source consistency reflects the relative stability between data sources, and its impact is usually smaller than that of other factors in the short term.

[0102] After analyzing the relative importance of each quality indicator, the following weights were determined: data integrity weight is 0.3, time delay weight is 0.2, noise level weight is 0.2, drift index weight is 0.15, and cross-source consistency weight is 0.15.

[0103] The formula for calculating the data quality score is: .

[0104] Substitute into the calculation: .

[0105] With a data quality score of 0.68, the confidence parameter is obtained using a linear mapping function: Confidence parameter = 0.68.

[0106] 3. Evidence Mapping and Reduction Based on a confidence parameter of 0.68, features from each data source are mapped and reduced. The initial base probability allocation is as follows:

[0107] Wave height: basic probability distribution = [0.6, 0.4], tension: basic probability distribution = [0.3, 0.7], reduction factor is 0.68, the reduced basic probability distribution is as follows: .

[0108] .

[0109] 4. Degree of conflict of evidence Calculate the degree of conflict in the basic probability distribution of wave height and tension.

[0110] Substitute the data into the calculation: .

[0111] Because the conflict level of 0.212704 is greater than the set threshold of 0.15, conflict suppression processing is triggered.

[0112] 5. Conflict Suppression and Evidence Combination Conflict suppression: According to the conflict suppression rule, the basic probability distribution of wave height is reduced to [0.25, 0.15], and the weighted part is reassigned to tension, resulting in a new basic probability distribution of tension as [0.45, 0.55].

[0113] Evidence combination: The weighted summation of the basic probability assignments of the reduced wave height and tension yields the final fused evidence representation: .

[0114] 6. Predictive Calculation The fused evidence representation [0.35, 0.65] is input into a nonlinear regression function to calculate the response prediction of the dynamic cable. The nonlinear regression function adopts a quadratic polynomial model, and the formula is:

[0115] in, It is a dynamic response prediction value. It is a representation of fused evidence. These are the coefficients of the regression model.

[0116] Determination of fitting coefficients: Using historical data, the least squares method is used to fit the regression model to determine... .

[0117] for : for : The predicted value of the dynamic cable response obtained by calculating the nonlinear regression function is... The prediction results provide a quantitative estimate of the dynamic cable status of marine operating platforms, which helps to assess the tension, displacement and other responses of dynamic cables under specific sea conditions and platform attitude conditions.

[0118] In summary, this embodiment verifies the effectiveness of credibility assessment and evidence fusion technology in complex marine environments and its reliability in practical applications.

[0119] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A hybrid prediction system for the dynamic response of a marine dynamic cable, characterized in that, include: Data acquisition and processing module: used to acquire multi-source sea state data, and to perform time synchronization, coordinate unification and linear interpolation on the multi-source sea state data, and output preprocessed data. The multi-source sea state data includes sea state observation data, platform attitude, operation status data and dynamic cable data. Feature extraction module: used to extract sea state features, platform status features and dynamic cable response features from the preprocessed data to form a multi-source feature set; The credibility assessment and evidence fusion module includes a credibility parameter generation unit, an evidence mapping unit, a conflict degree calculation unit, and an evidence combination unit. The credibility parameter generation unit calculates the credibility of the preprocessed data and generates credibility parameters. The evidence mapping unit maps the multi-source feature set into an evidence representation for prediction and converts the evidence representation into a basic probability allocation function under a unified predictive evidence framework. The evidence mapping unit performs a reduction calculation on the basic probability allocation function based on the credibility parameters to obtain a reduced evidence set. The conflict degree calculation unit calculates the evidence conflict degree of the evidence set and performs conflict suppression processing on the evidence set that causes conflict when the evidence conflict degree exceeds a preset threshold to form a stable evidence set. The conflict suppression processing includes conflict reallocation and weight reduction. The evidence combination unit performs evidence combination on the stable evidence set and outputs a fused evidence representation. Hybrid prediction module: used to perform prediction calculations on the fused evidence representation and output the dynamic response prediction results of the dynamic cable.

2. The hybrid prediction system for dynamic response of marine dynamic cables according to claim 1, characterized in that: The credibility parameter generation unit is used to perform residual calculation based on the preprocessed data within a preset sliding time window to obtain data quality indicators. The data quality indicators include data integrity, time delay, noise level, drift index, and cross-source consistency test results.

3. The hybrid prediction system for dynamic response of marine dynamic cables according to claim 2, characterized in that: The credibility parameter generation unit normalizes the data quality indicators respectively, and obtains the data quality score by weighted summation according to preset weights. The credibility parameter is obtained by the data quality score through a monotonic mapping function, and the credibility parameter is limited to the range of [0,1].

4. The hybrid prediction system for dynamic response of marine dynamic cables according to claim 1, characterized in that: The unified predictive evidence framework is used to predetermine a set of predictive propositions for at least one predictive target of the dynamic response of a dynamic cable. The set of predictive propositions consists of several mutually exclusive predictive propositions obtained by discretizing the value range of the predictive target. The set of predictive propositions is used as the domain of the basic probability allocation function. The evidence representation is converted into the basic probability allocation function according to the domain.

5. The hybrid prediction system for dynamic response of marine dynamic cables according to claim 1, characterized in that: The reduction calculation uses a reduction coefficient, which is a real number that is monotonically increasing with the confidence parameter and has a value range of [0, 1].

6. The hybrid prediction system for dynamic response of a marine dynamic cable according to claim 1, characterized in that: The calculation of the degree of conflict of evidence includes the following steps: S61. Using the basic probability allocation function corresponding to each sea state multi-source data as input, combine the basic probability allocation functions of any two sea state multi-source data on the same prediction proposition set in pairs. S62. Calculate the product of the values ​​of the basic probability assignment functions for the mutually exclusive predictive propositions in the set of predicted propositions and sum them to obtain the degree of evidence conflict.

7. The hybrid prediction system for dynamic response of a marine dynamic cable according to claim 6, characterized in that: The conflict suppression process includes, when the degree of conflict of the evidence exceeds a preset threshold, performing a weight reduction process on the assignment of the basic probability allocation function corresponding to the mutually exclusive prediction proposition in the evidence set that caused the conflict, and allocating the weighted part of the assignment to the prediction proposition that is compatible with the evidence set, so as to obtain the stable evidence set.

8. The hybrid prediction system for dynamic response of marine dynamic cables according to claim 1, characterized in that: The evidence combination is based on the basic probability allocation function, and according to the preset evidence combination rules, the values ​​of the basic probability allocation functions of the same prediction propositions on the same set of prediction propositions are weighted and summed, and the result of the weighted sum is normalized to obtain the fused evidence representation.

9. The hybrid prediction system for dynamic response of marine dynamic cables according to claim 1, characterized in that: The fused evidence representation includes sea state characteristic representation, platform state characteristic representation, and dynamic cable response characteristic representation. The prediction calculation inputs the fused evidence representation into a preset prediction function, evaluates the prediction function, and obtains the dynamic cable dynamic response prediction result. The preset function is a nonlinear function, and the dynamic cable dynamic response prediction result includes time-series prediction results of tension, displacement, velocity, and curvature.

10. A hybrid prediction method for the dynamic response of marine dynamic cables. A hybrid prediction system for the dynamic response of a marine dynamic cable according to any one of claims 1-9 is characterized in that, include: S1. Collect multi-source sea state data, and perform time synchronization, coordinate unification and linear interpolation to complete the multi-source sea state data to obtain preprocessed data. The multi-source sea state data includes sea state observation data, platform attitude data, operation status data and dynamic cable data. S2. Extract sea state features, platform status features, and dynamic cable response features from the preprocessed data to form a multi-source feature set; S3. Calculate the confidence parameters of the multi-source data for each sea state based on the preprocessed data. Under a unified predictive evidence framework, map the multi-source feature set into an evidence representation for prediction and convert it into a basic probability allocation function. Perform a reduction calculation on the basic probability allocation function according to the confidence parameters to obtain a reduced evidence set. Calculate the evidence conflict degree based on the reduced evidence set. When the evidence conflict degree exceeds a preset threshold, perform conflict suppression processing on the evidence set that causes conflict to form a stable evidence set. Perform evidence combination on the stable evidence set to obtain a fused evidence representation. S4. Input the fused evidence representation into a preset prediction function for prediction calculation, and output the prediction result of the dynamic response of the marine dynamic cable. The prediction function adopts a nonlinear regression function, and the prediction result of the dynamic response of the marine dynamic cable includes the time-series prediction results of tension, displacement, velocity and curvature.

Images