Intelligent Control Method and System for Shipborne Six-Degree-of-Freedom Platform Based on Multi-Source Data

Abstract

The present invention provides an intelligent control method and system for a shipborne six-degree-of-freedom platform based on multi-source data. The method includes the following steps: collecting multiple time-series data of wave characteristics around a target ship in the current time period; performing inversion calculation on the multiple time-series data of wave characteristics to obtain a wave fusion frequency spectrum; acquiring the current attitude data of all supporting hydraulic rods; calculating the real-time attitude data of the shipborne six-degree-of-freedom platform; performing spectral estimation analysis on the real-time attitude data by using fast Fourier transform to obtain a real-time attitude power spectrum; evenly fusing the real-time attitude power spectrum and the wave fusion frequency spectrum into a real-time control signal for the next time period of the hydraulic rod controller in the current time period; and controlling the shipborne six-degree-of-freedom platform to be in a balanced attitude based on the real-time control signal through the hydraulic rod controller. The present invention has the effect of quickly and stably maintaining the stability of the six-degree-of-freedom platform according to the real-time characteristics of the ocean.

Description

Technical Field

[0001] The present invention belongs to the technical field of six-degree-of-freedom platform control, and particularly relates to an intelligent control method and system for a shipborne six-degree-of-freedom platform based on multi-source data. Background Art

[0002] With the development of marine resources and the increase in maritime activities, the use of various ships and marine six-degree-of-freedom platforms has become increasingly common. When performing tasks, these six-degree-of-freedom platforms often need to maintain stable and precise operations in harsh marine environments. For example, in marine scientific research, measurement equipment installed on a six-degree-of-freedom platform needs to remain stable on a turbulent sea surface to obtain accurate data. However, natural factors such as waves, wind speed, and tides in the ocean are constantly changing, which pose severe challenges to the stability and control difficulty of shipborne platforms.

[0003] Traditional control methods for shipborne six-degree-of-freedom platforms usually rely on a single data source or a simple feedback control mechanism. When facing constantly changing and unpredictable marine conditions, due to the lack of effective integration and analysis of real-time environmental data, traditional systems often respond slowly or even fail when dynamically adjusting the platform attitude, and traditional control methods often seem inadequate. The dynamic characteristics of ocean waves and changing environmental factors make the platform vulnerable to interference, thus affecting its performance and functions. Summary of the Invention

[0004] The present invention provides an intelligent control method and system for a shipborne six-degree-of-freedom platform based on multi-source data to solve the problem of difficultly and quickly stabilizing the control of a shipborne six-degree-of-freedom platform to maintain stability under complex marine conditions.

[0005] In a first aspect, the present invention provides an intelligent control method for a shipborne six-degree-of-freedom platform based on multi-source data, which is applied to a shipborne six-degree-of-freedom platform arranged on a target ship. The shipborne six-degree-of-freedom platform includes a support base, a support platform, and six support hydraulic rods. Attitude sensors are arranged on all the support hydraulic rods. One end of each of the support hydraulic rods is connected to the top of the support base through a ball joint, and the other end of each of the support hydraulic rods is connected to the bottom of the support platform through a ball joint. The method includes the following steps:

[0006] Collect a plurality of time-series data of wave characteristics around the target ship in the current time period through a plurality of detection sensors pre-deployed around the hull of the target ship;

[0007] Perform inversion calculation on the plurality of time-series data of wave characteristics to obtain a wave fusion frequency spectrum around the target ship;

[0008] Obtain the real-time telescopic lengths of all the support hydraulic rods through the hydraulic rod controller of the support hydraulic rods, and obtain the current attitude data of all the support hydraulic rods through the attitude sensor;

[0009] Calculate the real-time attitude data of the shipborne six-degree-of-freedom platform by combining the real-time telescopic lengths and the current attitude data of all the support hydraulic rods;

[0010] Perform spectral estimation analysis on the real-time attitude data using the fast Fourier transform to obtain the real-time attitude power spectrum;

[0011] Equalize and fuse the real-time attitude power spectrum and the wave fusion frequency domain spectrum into the real-time control signal for the next time period of the hydraulic rod controller in the current time period;

[0012] Based on the real-time control signal and through the hydraulic rod controller, control the shipborne six-degree-of-freedom platform to be in a balanced attitude.

[0013] The step of performing inversion calculation on the multiple pieces of the wave feature time series data to obtain the wave fusion frequency domain spectrum around the target ship includes the following steps:

[0014] Optionally, for the wave feature time series data collected by each of the detection sensors, retrieve the historical feature time series data of an adjacent historical time period based on the data timestamp of the wave feature time series data;

[0015] If the time interval between the adjacent historical time period and the current time period does not exceed a preset interval time threshold, use the historical feature time series data as the inversion prior knowledge;

[0016] If all the historical feature time series data are used as the inversion prior knowledge, construct the Akaike Bayesian information criterion function by combining all the inversion prior knowledge and preset hyperparameters;

[0017] Obtain the optimal hyperparameters by calculating the minimum value of the Akaike Bayesian information criterion function;

[0018] Based on the optimal hyperparameters and according to the Gaussian distribution, determine the error parameters of each of the wave feature time series data;

[0019] Calculate the fusion cross-spectrum matrix by combining all the wave feature time series data;

[0020] Calculate the wave fusion frequency domain spectrum around the target ship by combining the fusion cross-spectrum matrix and the error parameters and using Bayesian model inversion;

[0021] Optionally, the step of calculating the fusion cross-spectrum matrix by combining all the wave feature time series data includes the following steps:

[0022] Arbitrarily select two adjacent ones of the said detection sensors as target detection sensors;

[0023] Fuse the two pieces of the sea wave feature time series data corresponding to the two said target detection sensors into a first sea wave feature matrix;

[0024] Perform a fast Fourier transform on the first sea wave feature matrix to obtain a first fused cross-spectrum matrix;

[0025] According to the deployment positions of the detection sensors and in sequence according to a preset selection direction, select adjacent detection sensors adjacent to the target detection sensors;

[0026] When any one of the adjacent detection sensors is selected, fuse the sea wave feature time series data corresponding to the adjacent detection sensor into the first fused cross-spectrum matrix through the fast Fourier transform;

[0027] Repeat the above data fusion steps until the sea wave feature time series data corresponding to all the adjacent detection sensors are fused into the first fused cross-spectrum matrix to obtain a fused cross-spectrum matrix.

[0028] Optionally, the method further includes the following steps:

[0029] If any one or more of the historical feature time series data are not used as the inversion prior knowledge, calculate the comprehensive signal-to-noise ratio of all the sea wave feature time series data;

[0030] If the comprehensive signal-to-noise ratio exceeds a preset signal-to-noise ratio threshold, use all pairs of adjacent detection sensors as detection sensor groups, and each detection sensor exists in two different detection sensor groups at the same time;

[0031] For each detection sensor group, fuse the two pieces of the sea wave feature time series data corresponding to the two detection sensors in the detection sensor group into a second sea wave feature matrix;

[0032] Convert all the second sea wave feature matrices into second fused cross-spectrum matrices through fast Fourier transform;

[0033] Based on a preset proportionality coefficient, linearly superimpose all the second fused cross-spectrum matrices to obtain a sea wave fusion frequency domain estimation spectrum, and obtain the sea wave fusion frequency domain spectrum around the target ship by minimizing the sea wave fusion frequency domain estimation spectrum.

[0034] Optionally, the calculation formula of the sea wave fusion frequency domain estimation spectrum is as follows:

[0035]

[0036] In the formula: D represents the estimated spectrum of the fused wave in the frequency domain, and δ represents the proportionality coefficient. represents the inverse matrix of the second fused cross-spectrum matrix between the detection sensor A n and the detection sensor B m in the same detection sensor group, represents the transfer function of the wave characteristics at the deployment position of the detection sensor A n ; represents the transfer function of the wave characteristics at the deployment position of the detection sensor B m , k represents the wave vector, ω represents the circular frequency, exp(·) represents the exponential function, and i represents the beam weight. represents the number of abscissas at the deployment position of the detection sensor A n ; represents the number of abscissas at the deployment position of the detection sensor B m .

[0037] Optionally, the method further includes the following steps:

[0038] If the comprehensive signal-to-noise ratio does not exceed the signal-to-noise ratio threshold, then all pairwise adjacent detection sensors are used as a detection sensor group, and each detection sensor exists in two different detection sensor groups at the same time;

[0039] For each detection sensor group, the two pieces of wave feature time series data corresponding to the two detection sensors in the detection sensor group are fused into a third wave feature matrix;

[0040] All the third wave feature matrices are converted into third fused cross-spectrum matrices through fast Fourier transform;

[0041] All the third fused cross-spectrum matrices are split into a third fused cross-spectrum signal matrix and a third fused cross-spectrum noise matrix;

[0042] All the third fused cross-spectrum signal matrices and all the third fused cross-spectrum noise matrices are linearly superimposed respectively to obtain a fused cross-spectrum signal matrix and a fused cross-spectrum noise matrix;

[0043] The fused cross-spectrum signal matrix and the fused cross-spectrum noise matrix are fused and added, and the fused cross-spectrum signal matrix part is minimized to obtain the fused wave frequency domain spectrum around the target ship.

[0044] Optionally, the calculation of the real-time attitude data of the shipborne six-degree-of-freedom platform by combining the real-time telescopic lengths of all the support hydraulic rods and the current attitude data includes the following steps:

[0045] Construct the platform kinematic model of the shipborne six-degree-of-freedom platform by combining the lengths of all the supporting hydraulic rods and the installation positions of all the spherical hinges.

[0046] Associate the real-time telescopic lengths and the current attitude data of all the supporting hydraulic rods with the platform kinematic model by means of inverse kinematics.

[0047] Use the Kalman filtering method and combine the real-time telescopic lengths and the current attitude data of the supporting hydraulic rods to calculate the real-time hydraulic rod attitude data of all the supporting hydraulic rods.

[0048] Input the real-time hydraulic rod attitude data into the platform kinematic model to calculate the real-time attitude data of the shipborne six-degree-of-freedom platform.

[0049] Optionally, the step of evenly fusing the real-time attitude power spectrum and the wave fusion frequency domain spectrum into the real-time control signal of the hydraulic rod controller in the next time period of the current time period includes the following steps:

[0050] Evenly fuse the real-time attitude power spectrum and the wave fusion frequency domain spectrum into a fusion power spectrum.

[0051] Count the amount of power spectrum data of the fusion power spectrum.

[0052] If the amount of power spectrum data exceeds the preset data amount threshold, use the overlapping window method to process the fusion power spectrum into a target fusion power spectrum with continuous signals.

[0053] If the amount of power spectrum does not exceed the data amount threshold, process the fusion power spectrum into the target fusion power spectrum through a pre-configured time-domain signal smoothing model.

[0054] Perform an inverse fast Fourier transform on the target fusion power spectrum to obtain the real-time control signal of the hydraulic rod controller in the next time period of the current time period.

[0055] In a second aspect, the present invention also provides an intelligent control system for a shipborne six-degree-of-freedom platform based on multi-source data, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the intelligent control method for a shipborne six-degree-of-freedom platform based on multi-source data as described in the first aspect.

[0056] In a third aspect, the present invention further provides a computer-readable storage medium, on which instructions are stored, characterized in that when the instructions are executed by a processor, the processor is configured to execute the intelligent control method for a shipborne six-degree-of-freedom platform based on multi-source data described in the first aspect.

[0057] The beneficial effects of the present invention are as follows:

[0058] Through the intelligent integration and real-time analysis of multi-source data, the stability and response ability of the platform in the dynamic marine environment are significantly improved. Especially when facing complex and changeable natural factors such as waves and wind speeds, it can more accurately control the platform attitude to ensure high precision and high reliability when performing tasks. By fusing multi-source data, analysis can be carried out in a wider frequency domain range, thereby providing more accurate control signals, enabling the platform to make adjustments in the shortest time and maintain balance and stability. In addition, through advanced data processing technologies such as fast Fourier transform, detailed spectral analysis can be performed on real-time attitude data, so as to better understand and predict the dynamic behavior of the platform, providing solid data support for the optimization of control strategies. By evenly fusing the real-time attitude power spectrum and the wave fusion frequency spectrum, the dynamic relationship between the platform and the external environment can be better coordinated, enabling the platform to maintain the best working state when performing complex tasks and avoiding performance degradation caused by external interference. In addition, through the precise control of the hydraulic rod controller, fine adjustment of the platform attitude can be achieved without affecting the overall structure and function of the platform, thereby improving its flexibility and adaptability while ensuring the stability of the platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a schematic flowchart of the intelligent control method for a shipborne six-degree-of-freedom platform based on multi-source data in one embodiment of the present application.

[0060] Figure 2 It is a schematic flowchart of the process of inversely calculating the wave fusion frequency spectrum in one embodiment of the present application.

[0061] Figure 3 It is a schematic flowchart of the process of calculating the real-time attitude data of a shipborne six-degree-of-freedom platform in one embodiment of the present application.

[0062] Figure 4 It is a schematic flowchart of the process of generating the real-time control signal of the hydraulic rod controller in one embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0063] Next, the technical solutions in the embodiments of the present application will be clearly described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art belong to the scope of protection of the present application.

[0064] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, rather than to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are usually of the same category, and do not limit the number of objects. For example, the first object can be one or multiple. In addition, "and / or" in the specification and claims means at least one of the connected objects, and the character " / " generally indicates an "or" relationship between the associated objects before and after.

[0065] Figure 1 It is a schematic flow chart of an intelligent control method for a shipborne six-degree-of-freedom platform based on multi-source data in an embodiment. It should be understood that although Figure 1 the steps in the flow chart are shown in sequence according to the indication of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, Figure 1 at least a part of the steps in

[0066] An intelligent control method for a shipborne six-degree-of-freedom platform based on multi-source data disclosed by the present invention is applied to a shipborne six-degree-of-freedom platform provided on a target ship. The shipborne six-degree-of-freedom platform includes a support base, a support platform, and six support hydraulic rods. Attitude sensors are provided on all the support hydraulic rods. One end of all the support hydraulic rods is connected to the top of the support base through a spherical hinge, and the other end of all the support hydraulic rods is connected to the bottom of the support platform through a spherical hinge. As Figure 1 shown, the intelligent control method for a shipborne six-degree-of-freedom platform based on multi-source data specifically includes the following steps:

[0067] S101. Collect multiple sets of time-series data of wave characteristics around the target ship in the current time period through multiple detection sensors pre-deployed around the hull of the target ship.

[0068] Among them, the types of sensors include wave sensors, acceleration sensors, wind speed and direction sensors, etc. These sensors are designed to be corrosion-resistant, waterproof and shock-resistant to adapt to the harsh conditions of the marine environment. The arrangement of sensors needs to be precisely calculated to ensure that they can cover all key areas around the target ship. Each sensor transmits data to the central processing unit in real time through a wireless or wired network. The data collected by the sensors includes the height, period, direction of ocean waves, as well as the speed and direction of sea winds, etc. These data are recorded as time-series data, that is, a continuous data set measured at different time points. By analyzing these time-series data, characteristic patterns of ocean waves can be identified, such as changes in wave frequency and amplitude. The significance of real-time acquisition is that it can promptly reflect changes in the marine environment and provide accurate basic data for subsequent control strategies. For example, when the wind speed increases, significant changes may occur in the height and frequency of ocean waves, and these changes will be captured and recorded by the sensors.

[0069] S102. Perform inversion calculations on multiple time-series data of ocean wave characteristics to obtain the fused frequency-domain spectrum of ocean waves around the target ship.

[0070] Among them, first, the time-series data of ocean wave characteristics collected by each sensor need to be analyzed. By comparing with the data in adjacent historical time periods, the changing trend of the current ocean wave characteristics can be judged. If the time interval between the adjacent historical time period and the current time period does not exceed the preset interval time threshold, these historical data can be used as prior knowledge for inversion calculations. Combining this prior knowledge and the preset hyperparameters, the Akaike Bayesian information criterion function can be constructed. Calculate the minimum value of this function to determine the optimal hyperparameters, and then based on these hyperparameters, determine the error parameters of each time-series data of ocean wave characteristics through a Gaussian distribution. Next, combining all the time-series data of ocean wave characteristics, calculate the fused cross-spectrum matrix. Finally, using the inversion calculation of the Bayesian model, combining the fused cross-spectrum matrix and the error parameters, obtain the fused frequency-domain spectrum of ocean waves around the target ship. This spectrogram provides important information about the frequency and energy distribution of ocean waves and provides a basis for the attitude adjustment of the platform.

[0071] S103. Obtain the real-time telescopic lengths of all supporting hydraulic rods through the hydraulic rod controller of the supporting hydraulic rods, and obtain the current attitude data of all supporting hydraulic rods through the attitude sensors.

[0072] Among them, the shipborne six-degree-of-freedom platform consists of a support base and a support platform, which are connected by six supporting hydraulic rods. One end of the hydraulic rod is connected to the support base through a ball joint, and the other end is connected to the support platform through a ball joint. The design of the ball joint allows the hydraulic rod to rotate freely in multiple directions, thereby providing flexible movement capabilities. The attitude sensor installed on each hydraulic rod is able to monitor the angle and direction changes of the hydraulic rod in real time. The hydraulic rod controller is responsible for monitoring and recording the telescopic length of the hydraulic rod, which is the key parameter for the hydraulic rod to perform attitude adjustment. Through the collaborative work of the attitude sensor and the hydraulic rod controller, the attitude information of the platform can be obtained in real time.

[0073] S104. Calculate the real-time attitude data of the shipborne six-degree-of-freedom platform by combining the real-time telescopic lengths of all supporting hydraulic rods and the current attitude data.

[0074] First, it is necessary to build a kinematic model of the platform, which takes into account the length of all supporting hydraulic rods and the installation position of the ball joint. Through the inverse kinematics method, the real-time telescopic length and current posture data of the hydraulic rod are associated with the platform kinematic model. Inverse kinematics is a mathematical method used to calculate the position and posture of a mechanical system in space. On this basis, the Kalman filter method is used to combine the real-time telescopic length and posture data to calculate the real-time hydraulic rod posture data of all supporting hydraulic rods. Kalman filtering is a recursive algorithm that can estimate the system state through a series of incomplete and noisy measurement data in a dynamic system. The calculated real-time hydraulic rod posture data is input into the platform kinematic model, and finally the real-time posture data of the six-degree-of-freedom platform is obtained.

[0075] S105. Perform spectrum estimation analysis on the real-time attitude data using fast Fourier transform to obtain a real-time attitude power spectrum.

[0076] Among them, the fast Fourier transform is an efficient algorithm for converting time domain signals into frequency domain information. By converting real-time attitude data into frequency domain information, the dynamic response characteristics of the platform can be analyzed more clearly. The power spectrum is the energy distribution of the signal in the frequency domain, which can reveal the main frequency components and energy concentration areas of attitude changes. By analyzing the power spectrum, the response characteristics of the platform at different frequencies can be identified, helping to identify vibration modes that may affect the stability of the platform.

[0077] S106. The real-time attitude power spectrum and the wave fusion frequency domain spectrum are equally fused into a real-time control signal of the hydraulic rod controller in the next time period of the current time period.

[0078] Among them, the real-time attitude power spectrum and the fused frequency spectrum of ocean waves are combined into a fused power spectrum. The amount of power spectrum data of the fused power spectrum is statistically analyzed. If the amount of data exceeds a preset threshold, the overlapping window method is used to process the fused power spectrum into a target fused power spectrum with continuous signals. The overlapping window method can smooth the spectrum data and reduce the spectrum leakage phenomenon. If the amount of power spectrum data does not exceed the threshold, the fused power spectrum is processed through a pre-configured time-domain signal smoothing model. The time-domain signal smoothing model is used to eliminate high-frequency noise in the signal and ensure the smoothness and continuity of the signal. Finally, the inverse fast Fourier transform is performed on the target fused power spectrum to obtain the real-time control signal of the hydraulic rod controller in the next time period.

[0079] S107. Based on the real-time control signal and through the hydraulic rod controller, control the shipborne six-degree-of-freedom platform to be in a balanced attitude.

[0080] Among them, the hydraulic rod controller receives the real-time control signal from the fused power spectrum and instructs the hydraulic rod to perform corresponding telescopic adjustments. By precisely controlling the length changes of each hydraulic rod, the platform can perform attitude adjustments in six degrees of freedom, including pitching, rolling, heaving, yawing, swaying, and rotating. The control process needs to consider the dynamic response characteristics of the platform and the changes in the ocean environment to ensure the stability and safety of the platform under various conditions.

[0081] In one implementation, referring to Figure 2 , the inversion calculation of multiple ocean wave feature time series data to obtain the fused frequency spectrum of ocean waves around the target ship includes the following steps:

[0082] S201. For the ocean wave feature time series data collected by each detection sensor, based on the data timestamp of the ocean wave feature time series data, retrieve the historical feature time series data of the adjacent historical time period.

[0083] S202. If the interval time between the adjacent historical time period and the current time period does not exceed the preset interval time threshold, the historical feature time series data is used as the inversion prior knowledge.

[0084] S203. If all the historical feature time series data are used as the inversion prior knowledge, combine all the inversion prior knowledge and the preset hyperparameters to construct the Akaike Bayesian information criterion function.

[0085] S204. Obtain the optimal hyperparameters by calculating the minimum value of the Akaike Bayesian information criterion function.

[0086] S205. Based on the optimal hyperparameters and according to the Gaussian distribution, determine the error parameter of each ocean wave feature time series data.

[0087] S206. Calculate the fused cross-spectrum matrix by combining all the ocean wave feature time series data.

[0088] S207. Combine the fused cross - spectral matrix and the error parameter, and use the Bayesian model to perform inversion calculation to obtain the fused frequency - domain spectrum of the sea waves around the target ship.

[0089] In this embodiment, the data acquisition process of the sea - wave characteristic time - series data is continuous. The sensor continuously records the dynamic characteristics of the sea waves, such as wave height, period, wavelength, etc. within a predetermined time interval. These data are marked with timestamps, which are the specific times of data acquisition, usually in seconds. This precise time marking allows for chronological sorting and matching of the data in subsequent analysis. In order to retrieve the data of adjacent historical time periods, an efficient data storage and retrieval system must be established first. This system needs to be able to quickly access a large amount of historical data and find the corresponding historical data according to the current timestamp. Usually, a database management system is used to store these data, and the database system needs to support fast retrieval and efficient storage to ensure that the data can be obtained in a timely manner when needed.

[0090] After obtaining the data of adjacent historical time periods, the next step is to determine whether the time interval between these historical data and the current time period exceeds a preset interval time threshold. The setting of this threshold is based on the understanding of the marine environment and the changes in sea - wave characteristics. The changes in sea - wave characteristics usually have periodicity and regularity. Therefore, it is crucial to select a suitable time - interval threshold. This threshold is usually determined through statistical analysis of a large amount of historical data, aiming to ensure the correlation and effectiveness between real - time data and historical data. If the time interval between the adjacent historical time period and the current time period does not exceed the preset threshold, it can be considered that these historical data have high correlation and reference value and can be used as prior knowledge for inversion calculation. Prior knowledge plays a key role in inversion calculation. It provides the initial parameter estimation and reference framework for the model. By using this prior knowledge, the accuracy and stability of the inversion calculation can be improved, and the uncertainty and error in the calculation process can be reduced. In actual operation, statistical analysis tools are usually used to evaluate the correlation and effectiveness of historical data.

[0091] After confirming that all historical feature time-series data can be used as inversion prior knowledge, the next step is to construct the Akaike Bayesian Information Criterion (ABIC) function by combining this prior knowledge with preset hyperparameters. The Akaike Bayesian Information Criterion (ABIC) is a statistical tool for model selection and hyperparameter estimation. ABIC combines the advantages of the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC), enabling effective parameter selection and optimization in complex models. When constructing the ABIC function, all inversion prior knowledge needs to be used as input and combined with preset hyperparameters, which typically include the complexity of the model, the noise level of the data, and other control parameters. By constructing the ABIC function, the fitting effect and complexity of the model under different hyperparameter combinations can be evaluated, and the best hyperparameter combination can be found to improve the prediction performance and stability of the model. In practical operations, numerical optimization algorithms are usually used to calculate the value of the ABIC function, and these algorithms can efficiently search the parameter space to find the parameter combination that minimizes the ABIC value.

[0092] The minimum value of the ABIC function corresponds to the point where the model achieves the best balance between complexity and fitting accuracy. In actual calculations, different hyperparameter combinations need to be traversed and evaluated, and the ABIC value for each combination is calculated. Numerical optimization algorithms such as the gradient descent method, Newton's method, or genetic algorithms can be used to efficiently search for the optimal hyperparameter combination. During the search process, the efficiency and accuracy of the calculation need to be considered to ensure accurate results within a reasonable calculation time. The process of determining the optimal hyperparameters is not only a mathematical optimization problem but also an engineering problem that combines practical application requirements. The selection of optimal hyperparameters directly affects the results of the inversion calculation and the performance of the model. Therefore, when selecting, the characteristics of the data, the complexity of the model, and the feasibility of the calculation need to be comprehensively considered. After obtaining the optimal hyperparameters, based on these parameters and according to the Gaussian distribution, the error parameters of each ocean wave feature time-series data are determined. The Gaussian distribution, i.e., the normal distribution, is a common probability distribution widely used in statistical analysis and error estimation. In this step, a Gaussian distribution model needs to be constructed based on the optimal hyperparameters to describe the error characteristics of the ocean wave feature time-series data. The error parameters usually include the mean and the standard deviation, where the mean represents the central tendency of the data and the standard deviation represents the degree of dispersion of the data. By estimating the errors of each ocean wave feature time-series data, outliers and noise in the data can be identified, thereby improving the quality and reliability of the data. This process is crucial for subsequent inversion calculations because accurate estimation of the error parameters can effectively reduce error accumulation during the calculation process and improve the accuracy and stability of the calculation results.

[0093] Then, by combining all the time-series data of the sea-wave characteristics, a fused cross-spectrum matrix is calculated. The cross-spectrum matrix is a mathematical tool used to describe the mutual relationship between multiple time-series signals. In this step, all the time-series data of the sea-wave characteristics need to be transformed into the frequency domain. By methods such as the fast Fourier transform (FFT), the time-domain signals are transformed into frequency-domain signals. Then, the cross-spectra between different signals are calculated, and these cross-spectra reflect the phase and amplitude relationships of different signals in the frequency domain. By combining all the cross-spectra together, a fused cross-spectrum matrix can be constructed. This matrix not only contains the spectral information of individual signals but also the interaction information between signals. By analyzing the fused cross-spectrum matrix, the correlations and mutual influences between different sea-wave characteristics can be identified, providing important frequency-domain information support for subsequent inversion calculations.

[0094] Finally, by combining the fused cross-spectrum matrix and the error parameters and using Bayesian model inversion calculation, the fused frequency-domain spectrum of the sea waves around the target ship is obtained. The Bayesian model is a statistical model based on probability theory that can perform parameter estimation and prediction in the presence of uncertainty and noise. In this step, the fused cross-spectrum matrix and the error parameters need to be used as the input of the Bayesian model. Through inversion calculation, the frequency-domain characteristics of the sea waves around the target ship are estimated. Inversion calculation is a process of inferring unknown parameters from known observation data. In the Bayesian model, prior knowledge and observation data are combined and updated through Bayes' theorem to obtain the posterior probability distribution. By maximizing the posterior probability, the most likely frequency-domain characteristics of the sea waves can be estimated. The finally obtained fused frequency-domain spectrum of the sea waves not only reflects the frequency and amplitude characteristics of the sea waves but also takes into account the interaction and error effects between different sea-wave characteristics, providing important basic data for the dynamic control and stability analysis of the ship.

[0095] In one implementation, calculating the fused cross-spectrum matrix by combining all the time-series data of the sea-wave characteristics includes the following steps:

[0096] Arbitrarily select two adjacent detection sensors as the target detection sensors;

[0097] Fuse the two time-series data of the sea-wave characteristics corresponding to the two target detection sensors into the first sea-wave characteristic matrix;

[0098] Perform a fast Fourier transform on the first sea-wave characteristic matrix to obtain the first fused cross-spectrum matrix;

[0099] According to the deployment positions of the detection sensors and in accordance with a preset selection direction, sequentially select the adjacent detection sensors adjacent to the target detection sensors;

[0100] When any adjacent detection sensor is selected, the time-series data of the ocean wave characteristics corresponding to the adjacent detection sensor is fused into the first fused cross-spectrum matrix through fast Fourier transform;

[0101] Repeat the above data fusion steps until the time-series data of the ocean wave characteristics corresponding to all adjacent detection sensors are fused into the first fused cross-spectrum matrix, obtaining the fused cross-spectrum matrix.

[0102] In this embodiment, since the ocean wave characteristic data collected by two adjacent detection sensors have the characteristic of mutual influence, two adjacent detection sensors are first selected as the target detection sensors. Specifically, in implementation, it can be determined which sensors are adjacent through the sensor numbers or coordinate information. For example, in a linearly arranged sensor array, detection sensor 1 is adjacent to detection sensor 2, detection sensor 2 is adjacent to detection sensor 3, and so on. The two selected sensors will be used to extract their respective time-series data of the ocean wave characteristics, which are usually obtained by real-time monitoring of characteristics such as the height, frequency, and period of ocean waves.

[0103] In the two selected target detection sensors, the obtained time-series data of the ocean wave characteristics will be fused into the first ocean wave characteristic matrix. The fusion process can be achieved by splicing or weighted averaging the time-series data of the two sensors. Next, perform a fast Fourier transform (FFT) on the first ocean wave characteristic matrix to obtain the first fused cross-spectrum matrix. FFT is an efficient algorithm for converting a time-domain signal into a frequency-domain signal to analyze the frequency components of the signal. Specifically, in implementation, the FFT algorithm can be used to process the first ocean wave characteristic matrix to obtain its frequency-domain representation. Through FFT calculation, the amplitude and phase information of each frequency component can be obtained, and then the first fused cross-spectrum matrix is formed, which can be specifically represented by the following formula:

[0104]

[0105] where C(f 1 , f 2 ) represents the fused cross-spectrum matrix, N is the total number of signal sampling points, is the complex conjugate of X f (n).

[0106] According to the deployment positions of the detection sensors, and in accordance with the preset selection direction, other detection sensors adjacent to the target detection sensor are sequentially selected. During specific implementation, a loop structure can be set up to check the adjacent relationships of each sensor one by one, ensuring that new adjacent sensors can be found in each iteration. For example, if the currently selected target detection sensor is detection sensor 2, then it can be checked whether detection sensor 1 and detection sensor 3 are its adjacent detection sensors. After any adjacent detection sensor is selected, the time series data of the ocean wave characteristics corresponding to this adjacent detection sensor is fused into the first fusion cross-spectrum matrix through fast Fourier transform. During specific implementation, first, the time series data of the ocean wave characteristics of this adjacent sensor is obtained, and then it is fused with the existing first fusion cross-spectrum matrix. The time series data of the ocean wave characteristics of the adjacent sensor can be used as a one-dimensional array, and then the information matrix fusion algorithm is used to achieve data fusion and form a new fusion matrix. At this time, the new fusion cross-spectrum matrix will contain more frequency components and interrelationships, thereby enhancing the complexity and information content of the data. By continuously repeating this data fusion step until the time series data of the ocean wave characteristics corresponding to all adjacent detection sensors are fused into the first fusion cross-spectrum matrix, the finally obtained fusion cross-spectrum matrix will contain comprehensive information from multiple sensors, which can effectively improve the accuracy and reliability of ocean wave characteristic analysis.

[0107] In one implementation, the method further includes the following steps:

[0108] If any one or more historical characteristic time series data are not used as inversion prior knowledge, then calculate the comprehensive signal-to-noise ratio of all the ocean wave characteristic time series data;

[0109] If the comprehensive signal-to-noise ratio exceeds the preset signal-to-noise ratio threshold, then all pairs of adjacent detection sensors are used as detection sensor groups, and each detection sensor exists in two different detection sensor groups at the same time;

[0110] For each detection sensor group, fuse the two time series data of the ocean wave characteristics corresponding to the two detection sensors in the detection sensor group into a second ocean wave characteristic matrix;

[0111] Convert all the second ocean wave characteristic matrices into a second fusion cross-spectrum matrix through fast Fourier transform;

[0112] Based on the preset proportionality coefficient, linearly superimpose all the second fusion cross-spectrum matrices to obtain an ocean wave fusion frequency domain estimation spectrum, and obtain the ocean wave fusion frequency domain spectrum around the target ship by minimizing the ocean wave fusion frequency domain estimation spectrum.

[0113] In this embodiment, when one or more historical feature time series data are not used as inversion prior knowledge, it will be difficult to obtain an accurate fused frequency spectrum of ocean waves if the Bayesian method is used for inversion calculation at this time. Therefore, first, the comprehensive signal-to-noise ratio of all ocean wave feature time series data can be calculated. The signal-to-noise ratio (SNR) is a key indicator used to measure the ratio of useful information to noise in a signal. A high SNR means better signal quality and less interference from noise to the signal. Conversely, a low SNR indicates that the noise has a greater impact on the signal, which may lead to signal distortion. When calculating the comprehensive signal-to-noise ratio, the ocean wave feature time series data collected by all detection sensors need to be considered. First, the data of each sensor needs to be preprocessed to eliminate obvious incorrect data and outliers. Then, the signal-to-noise ratio of each sensor's data and the comprehensive signal-to-noise ratio are calculated. The signal-to-noise ratio of each data is achieved by comparing the average power of the signal with the average power of the noise. The comprehensive signal-to-noise ratio is an indicator measuring the quality of all sensor data, which comprehensively considers the ratio of the signal intensity to the background noise collected by all sensors. The method for calculating the comprehensive signal-to-noise ratio is usually to perform a weighted average of the signal-to-noise ratios of each sensor, and the weights can be determined according to the importance or reliability of the sensors.

[0114] For example, assume there are n sensors, the signal-to-noise ratio of the i-th sensor is SNRi, and the weight is wi. Then the comprehensive signal-to-noise ratio SNRRC can be expressed as: SNRRC = (w1SNR1 + w2SNR2 +... + wn*SNRn) / (w1 + w2 +... + wn). The setting of the signal-to-noise ratio threshold is usually based on empirical values or obtained through the analysis of a large amount of experimental data, and it represents the lowest signal quality level that the system can accept.

[0115] If the comprehensive signal-to-noise ratio exceeds the preset signal-to-noise ratio threshold, it means that the data quality collected by all detection sensors at the current moment is high and the noise interference signal is less. Therefore, the maximum likelihood method with a lower computational complexity can be directly selected as the inversion calculation method. The specific process is as follows:

[0116] All pairs of adjacent detection sensors are used as detection sensor groups, and each detection sensor exists in two different detection sensor groups at the same time. For example, assume there are 5 sensors A, B, C, D, and E arranged evenly in a straight line. Then the formed detection sensor groups will be: (A, B), (B, C), (C, D), (D, E). Note that B exists in both the (A, B) and (B, C) groups, C exists in both the (B, C) and (C, D) groups, and so on.

[0117] For each detection sensor group, the two pieces of sea wave feature time series data corresponding to the two detection sensors in the detection sensor group are fused into a second sea wave feature matrix. Data fusion refers to integrating data from different sources to obtain more comprehensive and accurate information. In this step, it is necessary to fuse the sea wave feature time series data collected by the two sensors within each sensor group. First, standardize the data of each sensor to eliminate the scale differences between different sensors. Then, use the linear weighting method or other data fusion techniques to combine the data of the two sensors to form a new data matrix, which is called the second sea wave feature matrix. The linear weighting method is one of the most commonly used data fusion methods. By assigning different weights to different data sources, weighted averaging of the data is achieved. The selection of weights can be based on the accuracy of the sensors, signal-to-noise ratio, or other performance indicators.

[0118] Then, all the second sea wave feature matrices are converted into second fusion cross-spectrum matrices through fast Fourier transform (FFT). In this step, it is necessary to perform fast Fourier transform on each second sea wave feature matrix to calculate its spectral characteristics. Specifically, first perform FFT calculations on the rows or columns of each matrix to obtain the spectral representation of each signal. Then, calculate the cross-spectrum between different signals, and these cross-spectra reflect the phase and amplitude relationships of the signals in the frequency domain. By combining all the cross-spectra together, a second fusion cross-spectrum matrix can be constructed. This matrix not only contains the spectral information of individual signals but also contains the interaction information between signals.

[0119] Next, based on a preset proportionality coefficient, all the second fusion cross-spectrum matrices are linearly superimposed to obtain a sea wave fusion frequency domain estimation spectrum, and the sea wave fusion frequency domain spectrum around the target ship is obtained by minimizing the sea wave fusion frequency domain estimation spectrum. Linear superposition refers to performing weighted averaging on multiple matrices or signals to obtain a comprehensive result.

[0120] In this embodiment, the calculation formula of the sea wave fusion frequency domain estimation spectrum is as follows:

[0121]

[0122] In the formula: D represents the sea wave fusion frequency domain estimation spectrum, δ represents the proportionality coefficient, represents the inverse matrix of the second fusion cross-spectrum matrix between detection sensor A n and detection sensor B m within the same detection sensor group, represents the transfer function of the sea wave characteristics at the deployment position of detection sensor A n and represents the transfer function of the sea wave characteristics at the deployment position of detection sensor B m ​The transfer function of the sea wave characteristics at the deployment location, where k represents the wave vector, ω represents the circular frequency, exp(·) represents the exponential function, and i represents the beam weight, represents the detection sensor A n The number of abscissas at the deployment location, represents the detection sensor B m The number of abscissas at the deployment location.

[0123] By linearly superimposing all the cross-spectrum matrices, a comprehensive fused frequency-domain estimation spectrum of sea waves can be obtained. Next, by minimizing the frequency-domain estimation spectrum, the accuracy and stability of the frequency-domain spectrum are further optimized. The minimization process usually involves the application of optimization algorithms such as the gradient descent method or the genetic algorithm to find the combination of parameters that minimizes the error of the frequency-domain spectrum. The finally obtained fused frequency-domain spectrum of sea waves not only reflects the frequency and amplitude characteristics of the sea waves in the target area but also takes into account the interactions and error effects between different sea wave characteristics, providing important basic data for the dynamic control and stability analysis of the shipborne six-degree-of-freedom platform.

[0124] In one implementation, the method further includes the following steps:

[0125] If the comprehensive signal-to-noise ratio does not exceed the signal-to-noise ratio threshold, then all pairs of adjacent detection sensors are used as detection sensor groups, and each detection sensor exists in two different detection sensor groups simultaneously;

[0126] For each detection sensor group, the two sea wave characteristic time series data corresponding to the two detection sensors in the detection sensor group are fused into a third sea wave characteristic matrix;

[0127] All the third sea wave characteristic matrices are converted into third fused cross-spectrum matrices through fast Fourier transform;

[0128] All the third fused cross-spectrum matrices are split into a third fused cross-spectrum signal matrix and a third fused cross-spectrum noise matrix;

[0129] The third fused cross-spectrum signal matrices and the third fused cross-spectrum noise matrices are linearly superimposed respectively to obtain a fused cross-spectrum signal matrix and a fused cross-spectrum noise matrix;

[0130] The fused cross-spectrum signal matrix and the fused cross-spectrum noise matrix are fused and added, and the part of the fused cross-spectrum signal matrix is minimized to obtain the fused frequency-domain spectrum of the sea waves around the target ship.

[0131] In this embodiment, if the comprehensive signal-to-noise ratio exceeds the preset signal-to-noise ratio threshold, it indicates that the data quality collected by all detection sensors at the current moment is poor and there are many noise interference signals. Therefore, a method for high-complexity or high-noise interference data needs to be selected as the inversion calculation method. The specific process of the inversion calculation method in this embodiment is as follows:

[0132] Fuse the time-series data of ocean wave characteristics collected by two adjacent sensors in each detection sensor group. This process involves multiple links such as data preprocessing, time synchronization, feature extraction, and data fusion. First, it is necessary to preprocess the original time-series data collected by the two sensors. Preprocessing usually includes operations such as denoising, detrending, and normalization. Denoising can use methods such as low-pass filters or wavelet transforms to remove high-frequency noise; detrending is to eliminate the long-term change trend in the data and highlight the short-term fluctuation characteristics; normalization is to adjust the data of different sensors to the same scale range for subsequent processing. Next, it is necessary to ensure that the data of the two sensors are synchronized in time. Although adjacent sensors usually use the same sampling frequency, there may be slight time deviations. Time synchronization can be achieved through interpolation or resampling methods to ensure that the timestamps of the two sets of data exactly correspond. Then, extract the ocean wave characteristics from the preprocessed and synchronized data. Common ocean wave characteristics include wave height, wave period, wave direction, etc. These characteristics can be extracted through time-domain analysis (such as the zero-crossing method) or frequency-domain analysis (such as power spectral density analysis). Finally, fuse the characteristic time series extracted by the two sensors to form a third ocean wave characteristic matrix. The fusion method can use simple weighted averaging or more complex algorithms such as Kalman filtering. Repeating the fusion operation for all characteristics and time points can obtain the third ocean wave characteristic matrix.

[0133] In one embodiment, a specific implementation of the fusion operation can be: applying a Butterworth low-pass filter to the original data of the two sensors for denoising; using polynomial fitting to remove the long-term trend; mapping the data to the [0,1] interval through max-min normalization; using linear interpolation to ensure the time alignment of the two sets of data; using the zero-crossing method to extract wave height and wave period characteristics; using the fast Fourier transform (FFT) to calculate the power spectral density and extract the main wave direction; assigning weights according to the positions and reliabilities of the sensors and using the weighted averaging method to fuse the characteristics.

[0134] Next, all third sea wave feature matrices are converted into third fusion cross-spectrum matrices through fast Fourier transform. The core of this step is to convert the sea wave feature information in the time domain to the frequency domain to obtain richer frequency characteristics and phase information. Specifically, first, the third sea wave feature matrices need to be preprocessed, including mean removal and windowing. Mean removal can eliminate the influence of the DC component, and windowing (such as using a Hanning window) can reduce spectral leakage. Then, the FFT library function (such as FFTW) is used to perform FFT operations on each column of data. The FFT results are normalized to facilitate the comparison of signals of different lengths. The cross-spectrum between all feature pairs is calculated. The cross-spectrum is smoothed, such as using the Welch method, to reduce the variance of the estimation. Finally, the process of converting the third sea wave feature matrices into third fusion cross-spectrum matrices can be achieved.

[0135] Next, all third fusion cross-spectrum matrices are split into third fusion cross-spectrum signal matrices and third fusion cross-spectrum noise matrices. The core of this step is to separate the third fusion cross-spectrum matrices containing mixed signal and noise information into pure signal components and pure noise components. The matrix separation method adopted in this embodiment is the subspace decomposition technology based on singular value decomposition (SVD). The basic idea of this method is that signals usually have strong structural and correlation properties, while noise is more random and uncorrelated. Through SVD, the cross-spectrum matrix can be decomposed into different components, and signals and noise can be distinguished according to the magnitudes of the singular values. The specific implementation steps are as follows:

[0136] Perform SVD decomposition on each third fusion cross-spectrum matrix S:

[0137] S = U * Σ * V^H

[0138] where U and V are unitary matrices, and Σ is a diagonal matrix whose diagonal elements are singular values σ_i (arranged in descending order).

[0139] Next, determine the dimension of the signal subspace. This can be achieved by setting a threshold τ. For example, select the first k largest singular values such that:

[0140] (σ_1^2 +... + σ_k^2) / (σ_1^2 +... + σ_N^2) ≥ τ

[0141] where N is the total number of singular values, and τ is usually set to 0.9 or 0.95.

[0142] Next, reconstruct the signal matrix:

[0143] S_signal = U_k * Σ_k * V_k^H

[0144] Among them, $U_k$, $\Sigma_k$, and $V_k$ are the first $k$ columns of $U$, $\Sigma$, and $V$ respectively.

[0145] Finally, calculate the noise matrix: $S_{noise}=S - S_{signal}$.

[0146] For example, assume there is a $3\times3\times1024$ third fusion cross-spectrum matrix (3 features, 1024 frequency points). For each frequency point $k$, we have a $3\times3$ matrix $S_k$. After performing SVD decomposition on $S_k$, we may obtain three singular values $\sigma_1 = 10$, $\sigma_2 = 3$, $\sigma_3 = 0.5$. If $\tau = 0.95$ is set, then the first two singular values are selected because $(10^2 + 3^2) / (10^2 + 3^2 + 0.5^2)\approx0.9975>0.95$. Then, use the first two singular values and their corresponding singular vectors to reconstruct the signal matrix, and the remaining part is used as the noise matrix.

[0147] Next, linearly superimpose all the third fusion cross-spectrum signal matrices and all the third fusion cross-spectrum noise matrices respectively to obtain the fused cross-spectrum signal matrix and the fused cross-spectrum noise matrix. The core of this step is to linearly superimpose the multiple third fusion cross-spectrum signal matrices and noise matrices obtained in the previous step respectively, so as to obtain a comprehensive fused cross-spectrum signal matrix and a comprehensive fused cross-spectrum noise matrix. The purpose of this superimposing operation is to integrate the information from different sensor groups to obtain a more comprehensive and stable representation of the sea wave characteristics. Linear superposition is a simple but effective method. It assumes that the information provided by different sensor groups is independent of each other and can be combined by simple weighted summation. The advantages of this method are simple calculation, easy implementation, and wide application in signal processing. The specific implementation steps are as follows:

[0148] For each frequency point $k$, collect all the third fusion cross-spectrum signal matrices $S_{signal\_i}(k)$ and noise matrices $S_{noise\_i}(k)$, where $i$ represents different sensor groups. Assign a weight $w_i$ to each sensor group. The selection of the weight can be based on various factors, such as the reliability of the sensor, the signal-to-noise ratio, or the distance to the target, etc.

[0149] Perform weighted summation on the signal matrix:

[0150] $S_{signal\_fused}(k)=\sum_{i}w_i*S_{signal\_i}(k)$

[0151] Perform weighted summation on the noise matrix:

[0152] $S_{noise\_fused}(k)=\sum_{i}w_i*S_{noise\_i}(k)$

[0153] Normalize the results to ensure that the superimposed matrix has an appropriate scale:

[0154] S_signal_fused_norm(k) = S_signal_fused(k) / Σ(i)w_i

[0155] S_noise_fused_norm(k) = S_noise_fused(k) / Σ(i)w_i

[0156] Another specific embodiment of the above steps can be:

[0157] First, preprocess all the third fused cross-spectrum signal matrices and noise matrices, such as checking dimension consistency and numerical range.

[0158] Assign weights based on the position and reliability of the sensors.

[0159] Use library functions such as numpy to implement the weighted summation operation of the matrices.

[0160] Normalize and verify the superimposed results to ensure the rationality of the results.

[0161] Finally, fuse and add the fused cross-spectrum signal matrix and the fused cross-spectrum noise matrix and minimize the fused cross-spectrum signal matrix part to obtain the fused frequency-domain spectrum of the sea waves around the target ship. The core objective of this step is to maximize the suppression of noise while retaining the necessary information, so as to obtain a fused frequency-domain spectrum that can accurately reflect the characteristics of the sea waves around the target ship. This process involves the balance between signals and noise, the application of optimization algorithms, and the interpretation of the final results. The specific implementation steps are as follows:

[0162] First, add the fused cross-spectrum signal matrix S_signal and the fused cross-spectrum noise matrix S_noise to obtain the initial fused frequency-domain spectrum S_initial:

[0163] S_initial = S_signal + S_noise

[0164] Define an objective function J, which needs to minimize the signal part while maintaining sufficient information. A possible form is:

[0165] J = ||S_final - S_initial||_F^2 + λ * ||S_final||_1

[0166] Where ||·||_F represents the Frobenius norm, ||·||_1 represents the L1 norm, and λ is a trade-off parameter.

[0167] Use an optimization algorithm (such as proximal gradient descent) to minimize the objective function J and obtain the final fused frequency-domain spectrum S_final.

[0168] Post-process S_final, such as smoothing and normalization, to obtain the final fused frequency-domain spectrum of the sea waves around the target ship.

[0169] The key to this step lies in choosing the appropriate objective function and optimization algorithm. The objective function needs to strike a balance between information retention and noise suppression, while the optimization algorithm needs to be able to efficiently process large-scale data. In practical applications, methods such as cross-validation may be required to determine the optimal λ value.

[0170] In one implementation, referring to Figure 3 , calculating the real-time attitude data of the shipborne six-degree-of-freedom platform by combining the real-time telescopic lengths and current attitude data of all supporting hydraulic rods includes the following steps:

[0171] S301. Construct a platform kinematic model of the shipborne six-degree-of-freedom platform by combining the lengths of all supporting hydraulic rods and the installation positions of all ball joints.

[0172] S302. Associate the real-time telescopic lengths and current attitude data of all supporting hydraulic rods with the platform kinematic model through inverse kinematics methods.

[0173] S303. Use the Kalman filtering method and combine the real-time telescopic lengths and current attitude data of the supporting hydraulic rods to calculate the real-time hydraulic rod attitude data of all supporting hydraulic rods.

[0174] S304. Input the real-time hydraulic rod attitude data into the platform kinematic model to calculate the real-time attitude data of the shipborne six-degree-of-freedom platform.

[0175] In this embodiment, first, a fixed reference coordinate system is defined, usually a coordinate system fixedly connected to the hull. Then, a reference point is selected on the platform to establish the platform coordinate system. Next, the coordinates of the connection points of each support hydraulic rod in the reference coordinate system and the platform coordinate system are determined. These connection points are ball joints, and the ball joints allow the hydraulic rods to rotate freely in multiple directions. For each hydraulic rod, it can be represented by two vectors: one is the position vector in the reference coordinate system, and the other is the position vector in the platform coordinate system. The position and attitude of the platform can be described by six parameters: three translation parameters (x, y, z) and three rotation parameters (θ, φ, ψ), corresponding to roll, pitch, and yaw respectively. Using the rotation matrix R and the translation vector T, the transformation relationship between the reference coordinate system and the platform coordinate system can be established. For each hydraulic rod, its length can be expressed as the distance between the two connection points in the reference coordinate system. In this way, a set of non-linear equations can be established to relate the lengths of the hydraulic rods to the position and attitude parameters of the platform. This set of equations constitutes the kinematic model of the platform. In practical applications, factors such as the movement limitations of the hydraulic rods and the working space of the platform also need to be considered to further constrain and optimize the model. For example, the D-H parameter method can be used to describe the movement of each hydraulic rod, or the spherical geometry method can be adopted to simplify the calculation.

[0176] For each hydraulic rod, calculate the coordinates of its two connection points (ball joints) in the reference coordinate system. This can be achieved by multiplying the connection point coordinates in the platform coordinate system by the rotation matrix and then adding the translation vector. Next, calculate the length of each hydraulic rod, that is, the Euclidean distance between the two connection points. This process can be completed using vector subtraction and norm calculation. Since the kinematic equations are non-linear, numerical methods are usually required to solve them, such as the Newton-Raphson method or the Jacobi iterative method. In practical applications, the movement limitations of the hydraulic rods may also need to be considered to ensure that the calculated lengths are within the feasible range. In addition, to improve the calculation efficiency, parallel computing technology can be used to process the calculations of multiple hydraulic rods simultaneously. The result of the inverse kinematic calculation is a set of hydraulic rod lengths that correspond to the current platform attitude. By comparing the calculated lengths with the actually measured hydraulic rod lengths, the accuracy of the kinematic model can be verified and corrected if necessary.

[0177] Next, the Kalman filtering method is used to calculate the real-time attitude data of the supporting hydraulic rods. This usually involves non-linear transformations, so variants such as the Extended Kalman Filter or Unscented Kalman Filter may be used. Specifically, first, the Kalman filter is initialized by setting the initial state estimate and the error covariance matrix. At each time step, a prediction step is performed, using the previous state estimate and control inputs to predict the current state. Then, an update step is carried out, using the actually measured hydraulic rod length and attitude data to correct the prediction result. This process involves calculating the Kalman gain and updating the state estimate and the error covariance matrix. A key advantage of the Kalman filter is its ability to fuse multi-source data, such as the length sensors and attitude sensors of the hydraulic rods, and to account for various measurement noises and system uncertainties. In practical applications, parameter tuning may be required, such as adjusting the process noise covariance and the measurement noise covariance, to balance the filter's response speed and stability. Additionally, to handle possible non-Gaussian noise or strong non-linearity, more complex filtering techniques such as particle filtering may be considered. Finally, the Kalman filter outputs the optimal state estimate for each hydraulic rod, including position, velocity, and acceleration. These data can be used to infer the real-time attitude of the hydraulic rod, such as the tilt angle and azimuth angle. For example, the direction vector of the hydraulic rod can be calculated from the positions at both ends of the rod, and then the tilt angle and azimuth angle can be obtained.

[0178] Finally, the real-time hydraulic rod attitude data is input into the platform kinematic model to calculate the real-time attitude data of the shipborne six-degree-of-freedom platform. Specifically, first, the attitude data of each hydraulic rod needs to be converted into a form suitable for input to the kinematic model. This usually involves converting the tilt angle, azimuth angle, etc. of the hydraulic rod into its unit direction vector in the reference coordinate system. Then, combining the length information of the hydraulic rod, the exact positions of the connection points at both ends of each hydraulic rod in the reference coordinate system can be determined. Next, using this position information, a set of non-linear equations is constructed. This set of equations describes the relationship between the connection points of the hydraulic rods in the platform coordinate system and the reference coordinate system, which includes the six degrees-of-freedom parameters of the platform (three translations and three rotations) as unknowns. Solving this set of non-linear equations usually requires the use of numerical optimization methods, such as the Newton-Raphson method or the Levenberg-Marquardt algorithm. During the solution process, multiple iterations may be required, and the estimated value of the platform attitude is updated at each iteration until a preset convergence condition is reached. To improve the computational efficiency and stability, some techniques can be adopted, such as using the previous attitude estimate as the initial guess, or introducing constraint conditions to limit the search space of the solution. Additionally, the physical feasibility of the calculation results needs to be considered to ensure that the obtained attitude parameters are within the working range of the platform. The finally obtained platform real-time attitude data includes three translation parameters (x, y, z) and three rotation angles (roll, pitch, yaw).

[0179] In one implementation, referring toFigure 4 Fusing the real-time attitude power spectrum and the wave fusion frequency domain spectrum into the real-time control signal for the next time period of the hydraulic rod controller in the current time period includes the following steps:

[0180] S401. Fuse the real-time attitude power spectrum and the wave fusion frequency domain spectrum into a fusion power spectrum;

[0181] S402. Count the amount of power spectrum data of the fusion power spectrum;

[0182] S403. If the amount of power spectrum data exceeds the preset data volume threshold, use the overlapping window method to process the fusion power spectrum into a target fusion power spectrum with continuous signals;

[0183] S404. If the number of power spectra does not exceed the data volume threshold, process the fusion power spectrum into a target fusion power spectrum through a pre-configured time-domain signal smoothing model;

[0184] S405. Perform an inverse fast Fourier transform on the target fusion power spectrum to obtain the real-time control signal for the next time period of the hydraulic rod controller in the current time period.

[0185] In this embodiment, the real-time attitude power spectrum and the wave fusion frequency domain spectrum are evenly fused into a fusion power spectrum. This process needs to consider the relative importance of the two spectra in different frequency ranges. Generally, the low-frequency part is more affected by the waves, while the high-frequency part mainly reflects the dynamic characteristics of the platform itself. The weighted average method can be used for fusion, and the weight function can be a function of frequency. In practical applications, it may also be necessary to consider the normalization of the spectrum to ensure that the fused spectrum satisfies energy conservation. In addition, to handle possible spectrum leakage problems, a window function, such as the Hanning window or Hamming window, can be applied before the FFT. The finally obtained fusion power spectrum comprehensively reflects the characteristics of the platform attitude and the ocean environment, providing an important basis for subsequent control strategies.

[0186] Next, it is necessary to count the amount of power spectrum data of the fusion power spectrum, which directly affects the method selection and calculation efficiency of subsequent processing. In the frequency domain, each frequency point and its corresponding power value are usually regarded as a data point. Therefore, the data volume is actually a reflection of the frequency resolution. The specific counting process is as follows: First, determine the frequency range of the fusion power spectrum. Next, calculate the number of frequency points N. This can be obtained through the following formula: N = (f_max - f_min) / Δf + 1. Where f_max and f_min are the maximum and minimum frequencies respectively. Each frequency point has a corresponding power value, so the amount of power spectrum data is equal to the number of frequency points N. In practical applications, it may also be necessary to consider the difference between the bilateral spectrum and the unilateral spectrum. For real-valued signals, the unilateral spectrum is usually used, and in this case, the data volume is (N / 2 + 1).

[0187] The amount of data obtained by statistics not only reflects the complexity of the spectrum but also directly affects the computational burden of subsequent processing. When the amount of power spectrum data exceeds the preset data volume threshold, it indicates that the fusion power spectrum at this time has a high data complexity and is not suitable for data smoothing through a convolutional model. Therefore, the overlapping window method can be used to process the fusion power spectrum. The overlapping window method is an effective data dimensionality reduction and smoothing technology for low-complexity data. The core idea of this method is to divide a long time series into multiple overlapping short time segments, apply a window function to each segment, and then calculate the average power spectrum.

[0188] The specific implementation steps are as follows:

[0189] Determine the window size: The choice of window size needs to balance between time resolution and frequency resolution. A larger window provides better frequency resolution but reduces time resolution. Usually, the window size is selected as a power of 2, such as 256, 512, or 1024 points.

[0190] Select the overlap rate: The overlap rate is usually selected as 50% or 75%. A higher overlap rate can reduce data loss but increase the computational amount. For example, for a 50% overlap rate, each new window will overlap with the previous window by half.

[0191] Select the window function: Commonly used window functions include the Hanning window, Hamming window, and Blackman window, etc. The choice of window function affects spectral leakage and frequency resolution. For example, the Hanning window performs well in suppressing spectral leakage.

[0192] Apply the window function: Apply the selected window function to each short time segment. This can be achieved through element-wise multiplication.

[0193] Calculate the power spectrum of each window: Perform FFT on each windowed segment and then calculate the power spectrum.

[0194] Average processing: Average the power spectra of all windows to obtain the final power spectrum estimate.

[0195] The mathematical expression of this method can be written as:

[0196] P_welch(f) = (1 / K) * Σ(k = 1 to K) |FFT(x_k[n] * w[n])|^2

[0197] where K is the total number of windows, and x_k[n] is the data of the k-th window.

[0198] The overlapping window method reduces the variance of the power spectrum estimate by averaging multiple estimates. Moreover, compared to directly performing an FFT on the entire signal, this method can improve the frequency resolution while maintaining a certain time resolution. At the same time, the application of window functions helps reduce the spectral leakage effect. In practical applications, parameter tuning may be required, such as adjusting the window size, overlap rate, and window function type, to balance computational efficiency and spectral estimation quality.

[0199] When the number of power spectra does not exceed the data volume threshold, it indicates that the fused power spectrum at this time has a low data complexity. Therefore, the fused power spectrum can be processed into the target fused power spectrum through a pre-configured time-domain signal smoothing model. The core idea of this method is to apply smoothing techniques in the time domain and then transform back to the frequency domain to obtain a smoother and more continuous power spectrum.

[0200] The specific implementation steps are as follows:

[0201] Inverse Fourier transform: First, transform the fused power spectrum back to the time-domain signal through the inverse fast Fourier transform (IFFT). This step can be expressed as:

[0202] x(t) = IFFT(P(f))

[0203] where P(f) is the fused power spectrum and x(t) is the corresponding time-domain signal.

[0204] Select a smoothing model: Select an appropriate smoothing model according to the characteristics of the signal. Commonly used models include moving average, exponential smoothing, Savitzky-Golay filtering, signal convolution smoothing model, etc. The selection criteria are usually based on the signal bandwidth, noise characteristics, and desired smoothing degree.

[0205] Apply the smoothing model: Taking the moving average as an example, its mathematical expression is:

[0206] y(t) = (1 / N) * Σ(i=-(N - 1) / 2 to (N - 1) / 2) x(t + i)

[0207] where N is the window size (usually an odd number) and y(t) is the smoothed signal.

[0208] Parameter adjustment: Adjust the parameters of the smoothing model according to actual needs. For example, for the moving average, the window size can be adjusted; for exponential smoothing, the smoothing factor α can be adjusted. The selection of these parameters needs to be balanced between retaining the main features of the signal and removing noise.

[0209] Boundary handling: At the start and end parts of the signal, special processing is required to avoid edge effects. Commonly used methods include signal extension, mirror reflection, etc.

[0210] Fourier transform: The smoothed time-domain signal is converted back to the frequency domain through the Fast Fourier Transform (FFT) to obtain the target fusion power spectrum.

[0211] The advantage of this method is that it can significantly improve the quality and interpretability of the spectrum while keeping the data volume unchanged. The smoothed spectrum usually has better continuity and lower noise level, which is beneficial to subsequent analysis and the application of control algorithms.

[0212] Finally, perform the Inverse Fast Fourier Transform (IFFT) on the target fusion power spectrum to obtain the real-time control signal for the next time period of the hydraulic rod controller. The specific implementation steps are as follows:

[0213] Ensure that the target fusion power spectrum P(f) is in a format suitable for IFFT. Usually, IFFT requires the input to be a complex-valued spectrum. If the power spectrum only contains amplitude information, the phase information needs to be reconstructed. A simple method is to assume that the phase of all frequency components is zero, but this may lead to signal distortion. More complex methods include minimum-phase reconstruction or using the phase information of the original signal.

[0214] Perform IFFT: Apply the IFFT algorithm to convert the spectrum to a time-domain signal. Mathematically, this process can be expressed as:

[0215] x(t) = IFFT(sqrt(P(f)) * exp(j * φ(f)))

[0216] where φ(f) is the phase information and x(t) is the resulting time-domain signal.

[0217] Process the IFFT output: The output of IFFT is usually in complex form. For a real-valued signal, take the real part as the final time-domain signal. In addition, scaling may be required to ensure that the signal amplitude is within an appropriate range.

[0218] Time alignment: Ensure that the generated signal is correctly aligned with the current time period.

[0219] Signal smoothing: Since IFFT may introduce high-frequency noise, the resulting time-domain signal usually needs to be smoothed. Methods such as low-pass filters or moving averages can be used. For example, using an n-point moving average:

[0220] x_smooth(t) = (1 / n) * Σ(i = 0 to n - 1) x(t - i)

[0221] Range limitation: Limit the signal within the effective working range of the hydraulic rod controller. This usually involves signal normalization and truncation operations. For example:

[0222] x_limited(t) = max(min(x_smooth(t), upper_limit), lower_limit)

[0223] where upper_limit and lower_limit are the upper and lower limits of the controller.

[0224] Sampling rate adjustment: Ensure that the sampling rate of the generated control signal matches the operating frequency of the hydraulic rod controller. Downsampling or interpolation operations may be required.

[0225] Predictive compensation: Considering the response delay of the system, predictive compensation of the control signal may be required. This can be achieved by shifting the signal forward or using a prediction algorithm.

[0226] Smooth transition: To ensure a smooth transition with the control signal in the current time period, a gradual change function can be applied at the junction of the two time periods. For example, using linear interpolation: x_transition(t) = (1 - α) * x_current(t) + α * x_next(t). Where α is the transition coefficient from 0 to 1.

[0227] The present invention also discloses an intelligent control system for a shipborne six - degree - of - freedom platform based on multi - source data, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the intelligent control method for the shipborne six - degree - of - freedom platform based on multi - source data described in any of the above - mentioned embodiments.

[0228] Among them, the processor can adopt a central processing unit (CPU). Of course, according to the actual usage situation, other general - purpose processors, digital signal processors (DSPs), application - specific integrated circuits (ASICs), field - programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. can also be adopted. The general - purpose processor can adopt a microprocessor or any conventional processor, etc. This application does not make any restrictions on this.

[0229] Among them, the memory can be an internal storage unit of the computer device, for example, the hard disk or memory of the computer device, or it can also be an external storage device of the computer device, for example, a plug - in hard disk, a smart media card (SMC), a secure digital card (SD), or a flash card (FC) equipped on the computer device. And the memory can also be a combination of the internal storage unit and the external storage device of the computer device. The memory is used to store the computer program and other programs and data required by the computer device. The memory can also be used to temporarily store the data that has been output or will be output. This application does not make any restrictions on this.

[0230] The present invention also discloses a computer-readable storage medium, on which instructions are stored. It is characterized in that when the instructions are executed by a processor, the processor is configured to execute the intelligent control method of the shipborne six-degree-of-freedom platform based on multi-source data described in any of the above embodiments.

[0231] Among them, the computer program can be stored in a machine-readable medium. The computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some middleware form, etc. The machine-readable medium includes any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the machine-readable medium includes but is not limited to the above components.

[0232] Those of ordinary skill in the art should understand that: the discussion of any of the above embodiments is only exemplary and is not intended to imply that the protection scope of the present application is limited to these examples; under the idea of the present application, the technical features in the above embodiments or different embodiments can also be combined, and the steps can be implemented in any order, and there are many other variations in different aspects of one or more embodiments in the present application as above. For the sake of brevity, they are not provided in detail.

[0233] One or more embodiments in the present application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the present application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments in the present application shall be included in the protection scope of the present application.

Claims

1. A shipborne six-degree-of-freedom platform intelligent control method based on multi-source data, characterized in that: The method is applied to a shipborne six-degree-of-freedom platform provided on a target ship, wherein the shipborne six-degree-of-freedom platform comprises a support base, a support platform and six support hydraulic rods, all of which are provided with attitude sensors, one end of each of which is connected to the top of the support base through a ball joint, and the other end of each of which is connected to the bottom of the support platform through a ball joint, and the method comprises the following steps: Collecting multiple wave feature time series data around the target ship in the current time period through multiple detection sensors pre-deployed around the hull of the target ship; For each of the wave characteristic time series data collected by the detection sensor, retrieve the historical characteristic time series data of the adjacent historical time period based on the data timestamp of the wave characteristic time series data; If the interval between the adjacent historical time periods and the current time period does not exceed a preset interval threshold, the historical characteristic time series data is used as inversion prior knowledge; If all of the historical characteristic time series data are used as the inversion prior knowledge, then the Akaike Bayesian information standard function is constructed by combining all of the inversion prior knowledge and preset hyperparameters; The optimal hyperparameter is obtained by calculating the minimum value of the function of the Akaike Bayesian information standard function; Determine the error parameter of each of the ocean wave characteristic time series data based on the optimal hyperparameter and according to Gaussian distribution; Combining all the wave characteristic time series data, a fused cross-spectrum matrix is ​​calculated; Combining the fused cross-spectrum matrix and the error parameter and using a Bayesian model to invert and calculate, a fused frequency domain spectrum of the waves around the target ship; If any one or more of the historical characteristic time series data are not used as the inversion prior knowledge, calculating the comprehensive signal-to-noise ratio of all the wave characteristic time series data; If the comprehensive signal-to-noise ratio exceeds a preset signal-to-noise ratio threshold, all the detection sensors adjacent to each other are regarded as detection sensor groups, and each of the detection sensors exists in two different detection sensor groups at the same time; For each of the detection sensor groups, two of the ocean wave feature time series data corresponding to two of the detection sensors in the detection sensor group are fused into a second ocean wave feature matrix; Convert all the second wave feature matrices into a second fused cross-spectral matrix by fast Fourier transform; Based on a preset proportionality coefficient, all the second fused cross-spectrum matrices are linearly superimposed to obtain a wave fusion frequency domain estimation spectrum, and the wave fusion frequency domain spectrum around the target ship is obtained by minimizing the wave fusion frequency domain estimation spectrum; The calculation formula of the wave fusion frequency domain estimation spectrum is as follows: ， Where: represents the wave fusion frequency domain estimated spectrum, represents the proportionality coefficient, Indicates the detection sensor in the same detection sensor group and the detection sensor The inverse matrix of the second fusion cross-spectral matrix between Indicates the detection sensor The transfer function of the wave characteristics at the deployment location, Indicates the detection sensor The transfer function of the wave characteristics at the deployment location, represents the wave vector, represents the circular frequency, represents the exponential function, represents the beam weight, Indicates the detection sensor The number of horizontal coordinates of the deployment location, Indicates the detection sensor The number of horizontal coordinates of the deployment position; Acquiring the real-time telescopic lengths of all the supporting hydraulic rods through the hydraulic rod controller of the supporting hydraulic rods, and acquiring the current posture data of all the supporting hydraulic rods through the posture sensor; Calculate the real-time attitude data of the shipborne six-degree-of-freedom platform by combining the real-time telescopic lengths of all the supporting hydraulic rods and the current attitude data; Performing spectrum estimation analysis on the real-time attitude data using fast Fourier transform to obtain a real-time attitude power spectrum; The real-time attitude power spectrum and the wave fusion frequency domain spectrum are balanced and fused into a real-time control signal of the hydraulic rod controller in the next time period of the current time period; Based on the real-time control signal, the shipborne six-degree-of-freedom platform is controlled to be in a balanced posture through the hydraulic rod controller.

2. The shipborne six-degree-of-freedom platform intelligent control method based on multi-source data according to claim 1 is characterized in that: The step of combining all the ocean wave characteristic time series data to calculate the fused cross-spectrum matrix comprises the following steps: Randomly select two adjacent detection sensors as target detection sensors; Merging the two ocean wave feature time series data corresponding to the two target detection sensors into a first ocean wave feature matrix; Performing a fast Fourier transform on the first ocean wave characteristic matrix to obtain a first fused cross-spectrum matrix; Selecting adjacent detection sensors adjacent to the target detection sensor in sequence according to the deployment position of the detection sensor and in accordance with a preset selection direction; When any one of the adjacent detection sensors is selected, the wave feature time series data corresponding to the adjacent detection sensor is fused into the first fused cross-spectrum matrix through the fast Fourier transform; The above data fusion step is repeated until the wave feature time series data corresponding to all the adjacent detection sensors are fused into the first fused cross-spectrum matrix to obtain a fused cross-spectrum matrix.

3. The shipborne six-degree-of-freedom platform intelligent control method based on multi-source data according to claim 1 is characterized in that: The method further comprises the steps of: If the comprehensive signal-to-noise ratio does not exceed the signal-to-noise ratio threshold, all the detection sensors adjacent to each other are regarded as detection sensor groups, and each of the detection sensors exists in two different detection sensor groups at the same time; For each of the detection sensor groups, two of the ocean wave feature time series data corresponding to two of the detection sensors in the detection sensor group are fused into a third ocean wave feature matrix; Convert all the third wave feature matrices into a third fused cross-spectrum matrix by fast Fourier transform; Splitting all the third fused cross-spectrum matrices into a third fused cross-spectrum signal matrix and a third fused cross-spectrum noise matrix; Linearly superimposing all the third fused cross-spectrum signal matrices and all the third fused cross-spectrum noise matrices respectively to obtain a fused cross-spectrum signal matrix and a fused cross-spectrum noise matrix; The fused cross-spectrum signal matrix and the fused cross-spectrum noise matrix are fused and added, and the fused cross-spectrum signal matrix is ​​partially minimized to obtain the wave fusion frequency domain spectrum around the target ship.

4. The shipborne six-degree-of-freedom platform intelligent control method based on multi-source data according to claim 1 is characterized in that: The step of calculating the real-time attitude data of the shipborne six-degree-of-freedom platform by combining the real-time telescopic lengths of all the supporting hydraulic rods and the current attitude data comprises the following steps: Constructing a platform kinematics model of the shipborne six-degree-of-freedom platform in combination with the lengths of all the supporting hydraulic rods and the installation positions of all the ball joints; Associating the real-time telescopic lengths and current posture data of all the supporting hydraulic rods with the platform kinematic model by an inverse kinematics method; Using a Kalman filter method and combining the real-time telescopic length and the current posture data of the supporting hydraulic rod, the real-time hydraulic rod posture data of all the supporting hydraulic rods are calculated; The real-time hydraulic rod attitude data is input into the platform kinematic model to calculate and obtain the real-time attitude data of the shipborne six-degree-of-freedom platform.

5. The shipborne six-degree-of-freedom platform intelligent control method based on multi-source data according to claim 1 is characterized in that: The step of equalizing and fusing the real-time attitude power spectrum and the wave fusion frequency domain spectrum into a real-time control signal of the hydraulic rod controller in the next time period of the current time period comprises the following steps: Equalizing and fusing the real-time attitude power spectrum and the wave fusion frequency domain spectrum into a fusion power spectrum; Counting the amount of power spectrum data of the fused power spectrum; If the amount of power spectrum data exceeds a preset data amount threshold, an overlapping windowing method is used to process the fused power spectrum into a target fused power spectrum with continuous signals; If the number of the power spectra does not exceed the data volume threshold, processing the fused power spectrum into the target fused power spectrum through a pre-configured time domain signal smoothing model; Perform an inverse fast Fourier transform on the target fusion power spectrum to obtain a real-time control signal of the hydraulic rod controller in the next time period of the current time period.

6. A shipborne six-degree-of-freedom platform intelligent control system based on multi-source data, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the intelligent control method for a shipborne six-degree-of-freedom platform based on multi-source data as described in any one of claims 1 to 5 is implemented.

7. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instruction is executed by a processor, the processor is configured to execute the shipborne six-degree-of-freedom platform intelligent control method based on multi-source data according to any one of claims 1 to 5.

Images