A multi-modal collaborative accelerated test system and method based on a digital twin model

Abstract

The application discloses a kind of multi-modal collaborative acceleration test system and method based on digital twin model, is specially used for the acceleration test of pure electric ship power system.The method first designs multiple acceleration test scenarios, simulates the running state of pure electric ship power system under different scenarios using digital twin model.During the test process, the system collects multi-modal data, including sensor data, operating parameters, etc., and processes these data to form a complete data set.Through in-depth analysis of the data set, the system can identify potential problems of the current power system and evaluate its environmental performance.Finally, based on the analysis results, targeted optimization suggestions are generated to improve the performance and environmental level of the power system.This method significantly improves the test efficiency and reduces the test cost by combining virtual simulation and data analysis, while providing a scientific basis for the optimization of pure electric ship power system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ship power test, in particular to a multi-modal collaborative accelerated test system and method based on a digital twin model. BACKGROUND

[0002] The existing ship power system accelerated test has the following defects: first, the traditional ship power system test method highly depends on the direct operation test of the entity system. This method not only needs to invest a large amount of manpower and material resources, but also has a long test cycle and high cost. Due to the complexity and diversity of actual operation conditions, the entity test often cannot cover all possible operation scenarios and boundary conditions, which leads to the possibility of one-sidedness of the test results, and the actual performance of the power system cannot be fully reflected. Second, the power system generates various types of data during operation, including sensor data, image data, sound data, etc. These data contain rich information, which is crucial for the performance evaluation and optimization of the power system. However, the traditional test method can only process a single type of data, and cannot effectively fuse and analyze these multi-dimensional data. This leads to the fact that the test personnel cannot fully and deeply understand the actual operation state of the power system, and cannot dig out potential problems and optimization space from the data, thereby seriously affecting the efficiency and accuracy of the development and test of the power system. Third, the traditional ship power system monitoring mainly relies on manual experience and simple instruments. This method has many limitations, such as limited monitoring range, insufficient accuracy, poor real-time performance, etc. Manual monitoring is often limited by experience and knowledge level, and cannot fully and accurately identify abnormalities and potential faults in the system. Simple instruments may not provide sufficient monitoring data and accuracy, leading to misjudgment or missed judgment of the system state. These problems not only affect the real-time grasp of the operation state of the ship power system, but also may pose a potential threat to the safety and reliability of the system. SUMMARY

[0003] In order to solve the problems of high test cost, insufficient data fusion analysis and limited monitoring means of the existing ship power system accelerated test, the present application provides a multi-modal collaborative accelerated test system and method based on a digital twin model. The specific technical solutions of the present application are as follows:

[0004] The application discloses a multi-modal collaborative accelerated test method based on a digital twin model, which is applied to an accelerated test system for performing an accelerated test on a pure electric ship power system.

[0005] Further, before setting the accelerated test scene, the accelerated test system constructs a digital twin model, including the following steps: collecting key parameters of the pure electric ship power system, wherein the key parameters include but are not limited to structural parameters, operating parameters and environmental parameters; constructing a digital twin model corresponding to the pure electric ship power system according to the collected key parameters; then simulating the physical characteristics of the pure electric ship power system in the digital twin model, wherein the physical characteristics include but are not limited to power transmission, energy conversion and thermal effect; comparing the actual operating data of the pure electric ship power system with the predicted data of the digital twin model, and adjusting the digital twin model according to the comparison result.

[0006] Further, the accelerated test system collects multi-modal data of the pure electric ship power system simulated in different accelerated test scenes, including the following steps: collecting sensor data, image data and sound data of the pure electric ship power system in different accelerated test scenes, wherein the sensor data includes but is not limited to battery system data, propulsion system data, ship operation data and environmental safety data; after collecting the sensor data, automatically updating the parameters of the digital twin model through the collected sensor data, and optimizing the boundary conditions of the accelerated test scene by using a reinforcement learning algorithm.

[0007] Further, the accelerated test system processes the multi-modal data to obtain a data set, including the following steps: preprocessing the obtained multi-modal data, wherein the preprocessing includes data cleaning, denoising and format conversion; performing data fusion algorithm calculation on the preprocessed data to obtain a data set.

[0008] Further, the accelerated test system performs data fusion algorithm calculation on the preprocessed data, including the following steps:

[0009] Suppose N sensors measure the same measured value Y at different positions, and the measured value of each sensor is denoted as X j (j = 1, 2,..., N);

[0010] A set of measurements obtained for each sensor is normalized with the maximum measurement value in the set of data, Y j `(n) = Y j (n) / Max(Y j (n))j = 1, 2, …, N, n = 1, 2, …, I, where Max(Y j (n)) is the maximum measurement value of the jth sensor within the estimation length I, Y j `(n) is the normalized measurement value of the jth sensor;

[0011] Different weights are assigned to each sensor according to the type of each sensor, and the estimated value Yx of the measurement value Y of the N sensors after the weights are assigned is obtained:

[0012]

[0013] Then, the estimated value Yx is de-normalized to obtain the fusion data of each sensor.

[0014] Further, the acceleration test system analyzes the data set, including the following steps: analyzing and calculating the data set by data analysis technology and data mining technology to obtain an analysis result; comparing the work efficiency and stability in the analysis result with a first standard value to determine potential problems of the current pure electric ship power system; dividing the propulsion power in the analysis result by the sum of battery production carbon emissions and operating energy consumption to obtain a navigation carbon efficiency ratio, and determining the environmental protection performance of the current pure electric ship power system according to the navigation carbon efficiency ratio and the IMO Tier III standard; generating a parameter adjustment command according to the optimization suggestion, and sending the parameter adjustment command to the pure electric ship power system to make the pure electric ship power system adjust its hardware parameters.

[0015] Further, the acceleration test system calculates the obtained multi-modal data by an anomaly detection algorithm to determine abnormal signals in the current pure electric ship power system; locates and determines the fault position of the current pure electric ship power system according to the abnormal signals, and sends a fault warning signal; generates an LSTM fault prediction model based on historical fault abnormal signals to realize fault prediction of the pure electric ship power system, and sends a maintenance suggestion.

[0016] Further, the acceleration test system stores, backs up and encrypts the obtained multi-modal data, and constructs a data processing command for data access and connection receiving users for the stored multi-modal data, recovers or deletes the multi-modal data; or, receives input parameters of the user, and updates the digital twin model.

[0017] Further, the accelerated test system comprises a user interaction module; the user interaction module is configured to generate the multi-modal data into visual charts; the user interaction module is configured to display the visual charts, the positioning information of the fault location, the judgment result of the fault location, and the early warning signal to the user; and the user interaction module is configured to receive the control commands and feedback information of the user.

[0018] A multi-modal collaborative accelerated test system based on a digital twin model, comprising a digital twin model module, a test design and execution module, a multi-modal data fusion analysis module, a performance evaluation optimization module, a pre-warning and fault diagnosis module, a data storage management module, and a user interaction module; the digital twin model module comprises an entity data acquisition module, a physical property simulation module, a modeling and simulation module, and a model verification module; the test design and execution module comprises a test scenario setting module, a test flow control module, an accelerated working condition simulation module, and a real-time data monitoring module; the multi-modal data fusion analysis module comprises a sensor data acquisition module, a sound data acquisition module, a data fusion algorithm module, an image data acquisition module, a data processing and cleaning module, and a data analysis and mining module; the performance evaluation optimization module comprises a performance evaluation module, an emission and energy consumption monitoring module, a stability analysis module, and an optimization suggestion module; the pre-warning and fault diagnosis module comprises a real-time state monitoring module, a fault pre-warning module, an abnormality checking and identification module, and a fault diagnosis and positioning module; the data storage management module comprises a data storage module, a security and encryption module, a backup and recovery module, and a data query module; and the user interaction module comprises a user interface module, a data visualization module, a feedback and help module, and a real-time data display module.

[0019] Compared with the prior art, the beneficial effects of the present application are that: the present application constructs a digital twin model corresponding to the entity ship power system through digital twin technology. This model not only accurately replicates the structure and function of the entity system, but also can reflect its running state and performance parameters in real time. The construction of the digital twin model enables researchers to comprehensively simulate and test the power system in a virtual environment, greatly improving the test efficiency and accuracy, and reducing the test cost. In addition, the digital twin model can also be used to predict system performance, providing strong support for optimization design and improvement. Moreover, the design of the multi-modal data fusion and analysis method can simultaneously process multiple types of data, such as sensor data, image data, etc., realizing the comprehensive integration and deep mining of data, and obtaining more comprehensive and rich information, so as to more deeply understand the running state and performance of the power system, providing more comprehensive information support for the research and test of the power system, and laying a foundation for subsequent optimization and improvement. At the same time, by collecting and analyzing a large amount of data in the running process of the ship power system, using data mining and machine learning algorithms, the running state of the system can be comprehensively monitored and warned, and abnormalities and potential faults in the system can be found in time, providing strong support for timely maintenance and repair. At the same time, through in-depth analysis of historical data, strong basis can also be provided for the optimization design and improvement of the system. BRIEF DESCRIPTION OF DRAWINGS

[0020] Fig. 1 A flowchart of an accelerated test in an embodiment of the present application is shown in the figure.

[0021] Fig. 2 A structural schematic diagram of an accelerated test system in an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0022] The embodiments of the present application will be described in detail below, and the embodiments are shown in the accompanying drawings, wherein the same or similar reference signs represent the same or similar elements or elements having the same or similar functions throughout.

[0023] In the description of the present application, it should be noted that for orientation words, such as the terms "center", "transverse", "longitudinal", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", etc. The orientation and positional relationship shown in the drawings is based on the orientation or positional relationship shown in the drawings, which is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and cannot be understood as limiting the specific protection scope of the present application.

[0024] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features. Thus, the use of "first" or "second" to define a feature may explicitly or implicitly include one or more of that feature, and in the description of this application, "at least" means one or more, unless otherwise explicitly specified.

[0025] In this application, unless otherwise expressly specified and limited, the terms "assembly," "connection," and "joining" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can also refer to a mechanical connection; they can refer to a direct connection or a connection through an intermediate medium; or they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0026] In the application, unless otherwise specified and limited, "above" or "below" the second feature can include direct contact between the first and second features, or contact between the first and second features through another feature between them. Furthermore, "above," "below," and "over" the second feature includes the first feature being directly above or diagonally above the second feature, or simply indicating that the first feature is at a higher horizontal level than the second feature. "Above," "below," and "below" the second feature includes the first feature being directly below or diagonally below the second feature, or simply indicating that the first feature is at a lower horizontal level than the second feature.

[0027] The following description, in conjunction with the accompanying drawings, further illustrates specific embodiments of this application, making the technical solution and its beneficial effects clearer and more explicit. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, but should not be construed as limiting it.

[0028] like Figs. 1-2 As shown, a multimodal collaborative accelerated testing method based on a digital twin model is applied to an accelerated testing system for accelerating the testing of pure electric ship propulsion systems. The method includes the following steps:

[0029] Different acceleration test scenarios of pure electric ship power systems are designed, and then the pure electric ship power system is simulated in different acceleration test scenarios through a digital twin model. The acceleration test scenarios are designed according to the actual ship operation process. The multi-modal data of the pure electric ship power system in different acceleration test scenarios are collected, and then the multi-modal data are processed to obtain a data set. The potential problems and environmental performance of the current pure electric ship power system are obtained by analyzing and calculating the data set, and optimization suggestions are generated according to the potential problems and environmental performance. The acceleration test using the digital twin model is an efficient and advanced testing method. It can simulate the long-term operation state of the physical entity in a short time through virtual simulation and real-time data driving, and accelerate performance verification and fault diagnosis. It combines the advantages of virtual simulation and real-time data driving, and can simulate the long-term operation state of the physical entity in a short time, accelerate performance verification and fault diagnosis. By constructing a high-fidelity digital twin model, real-time updating driven by data, acceleration test design, and test result analysis and verification, the test efficiency can be significantly improved, the cost can be reduced, the safety can be enhanced, and the accuracy can be improved. The traditional test takes 30 days, and the system only takes 5 days; the fault recognition rate is improved from 82% to 95%.

[0030] As one of the embodiments, before setting the acceleration test scenario, the acceleration test system first constructs a digital twin model, including the following steps: collecting key parameters of the pure electric ship power system, wherein the key parameters include but are not limited to structural parameters, operating parameters and environmental parameters. According to the collected key parameters, a digital twin model corresponding to the pure electric ship power system is constructed. Then the physical properties of the pure electric ship power system are simulated in the digital twin model, wherein the physical properties include but are not limited to power transmission, energy conversion and thermal effect. The actual operation data of the pure electric ship power system and the prediction data of the digital twin model are compared, and the digital twin model is tested and adjusted according to the comparison result. The acceleration test system collects the key parameters of the pure electric ship power system through the entity data collection module, including the structural parameters, the operating parameters and the environmental parameters, and then constructs the digital twin model corresponding to the entity system by using the collected data through the modeling and simulation module. The physical property simulation module is used to simulate the physical properties of the entity system in the digital twin model, such as power transmission, energy conversion and thermal effect, to ensure the accuracy of the model. Finally, the model verification module compares the actual operation data of the entity system with the prediction data of the digital twin model to verify and adjust the model, so as to improve the accuracy and reliability of the model.

[0031] As one of the embodiments, the acceleration test system collects multi-modal data of the pure electric ship power system in different acceleration test scenarios, including the following steps: collecting sensor data, image data and sound data of the pure electric ship power system in different acceleration test scenarios, wherein the sensor data includes but is not limited to battery system data, propulsion system data, ship operation data and environmental safety data. After collecting the sensor data, the parameters of the digital twin model are automatically updated through the collected sensor data (such as battery temperature, battery current, propulsion torque, etc.), and the boundary conditions of the acceleration test scenario are optimized by using the reinforcement learning algorithm.

[0032] The sensor data of the pure electric ship power system is the key to realize system monitoring, fault diagnosis and performance optimization. The following is the common sensor data type in the pure electric ship power system and its specific application:

[0033] Battery system related sensor data, battery voltage: real-time monitoring of the voltage of each battery unit or battery pack through voltage sensor. Voltage data is used to evaluate the state of charge (SOC) and the state of health (SOH) of the battery, and to ensure that the battery operates within a safe voltage range. Battery current: current sensor is used to measure the charge and discharge current of the battery. These data help to calculate the power output and input of the battery, and to monitor the charge and discharge rate of the battery. Battery temperature: temperature sensor is installed inside or around the battery unit to monitor the battery temperature in real time. Temperature data is crucial for preventing battery overheating and ensuring the safety of the battery system. Battery pack insulation resistance: insulation resistance sensor is used to detect the insulation performance of the battery pack to prevent electric leakage and short circuit.

[0034] Propulsion system related sensor data, motor speed: the speed of the propulsion motor is measured by speed sensor (such as photoelectric sensor or Hall effect sensor). These data are used to control the running speed and power output of the motor. Motor torque: torque sensor is used to measure the torque output of the motor. Torque data is very important for evaluating the performance and efficiency of the motor. Propulsion shaft power: the actual power output of the propulsion shaft is determined by power sensor or by calculating the voltage, current and speed of the motor.

[0035] Ship operation state related sensor data, ship position and heading: real-time position and heading information of the ship is obtained by GPS and compass sensor. These data are used for navigation and automatic driving system. Ship speed: speed sensor (such as Doppler log) is used to measure the ship's speed against the ground and against the water. Ship attitude: the roll, pitch and yaw angles of the ship are measured by attitude sensor (such as IMU). These data are used for ship motion control and stability evaluation.

[0036] Environmental and safety-related sensor data, ambient temperature and humidity: Environmental sensors are used to monitor the temperature and humidity inside the ship's engine room and battery compartment, ensuring that equipment operates under suitable environmental conditions. Water level and leakage detection: Water level sensors and leakage detection sensors are used to monitor the water level and leakage in the ship's engine room, preventing water immersion and equipment damage. Fire and smoke detection: Fire and smoke sensors are used to detect fires early, ensuring the safety of the ship and personnel.

[0037] System monitoring and fault diagnosis-related sensor data, system voltage and current: Voltage and current sensors are used to monitor the overall operation of the ship's power system, ensuring the stability and safety of power supply. Fault alarm signals: Various sensors and controllers can output fault alarm signals for timely detection and handling of system faults.

[0038] As one of the embodiments, the accelerated test system processes multi-modal data to obtain a data set, including the following steps: preprocessing the acquired multi-modal data, wherein the preprocessing includes data cleaning, denoising and format conversion to ensure the accuracy and availability of the data. Perform data fusion algorithm calculation on the preprocessed data to obtain a data set, and fuse multiple types of data to form a comprehensive data set to better understand the operation status of the power system.

[0039] Data cleaning, denoising and format conversion are important steps in data preprocessing, and their purpose is to improve the quality and availability of data, providing a reliable data foundation for subsequent data analysis and mining.

[0040] Data cleaning refers to the process of detecting and correcting (or deleting) errors, duplicates, inconsistencies or missing parts in the data. The purpose of data cleaning is to ensure the integrity, accuracy and consistency of the data. Data cleaning methods include removing duplicate data, removing duplicate data methods: through data deduplication algorithms, identify and delete duplicate records. For example, use the drop_duplicates() method in the Pandas library.

[0041] Denoising refers to the process of removing noise from data. Noise may come from errors in data collection, sensor failure or external interference. The purpose of denoising is to improve the accuracy and reliability of the data. Denoising methods, filter denoising, low-pass filter: remove high-frequency noise and retain low-frequency signals. For example, use Fourier transform or wavelet transform. High-pass filter: remove low-frequency noise and retain high-frequency signals. 2. Smoothing techniques, moving average: smooth the data by calculating the average value within the window; median filter: remove spike noise by calculating the median value within the window. 3. Wavelet denoising, wavelet transform: decompose the signal into different frequency components through wavelet transform, then remove high-frequency noise.

[0042] Format conversion refers to transforming data from one format to another to meet the needs of subsequent processing or analysis. The purpose of format conversion is to ensure data consistency and compatibility. Methods of format conversion include: 1. Data type conversion: Numeric types: converting data to integers or floating-point numbers; Date types: converting data to date format. 2. Data encoding: Categorical data encoding: converting categorical data to numeric encoding; One-hot encoding: converting categorical data to one-hot encoding. 3. Data normalization: Min-Max normalization: normalizing data to the [0,1] interval; Z-score normalization: normalizing data to a distribution with a mean of 0 and a standard deviation of 1. 4. Data discretization: Equal-width discretization: dividing data into intervals of equal width; Equal-frequency discretization: dividing data into intervals of equal frequency.

[0043] As one embodiment, the accelerated testing system performs data fusion algorithm calculations on the preprocessed data, including the following steps:

[0044] Suppose N sensors measure the same value Y at different locations, and each sensor's measurement value is denoted as X. j (j = 1, 2, ... N);

[0045] For each set of measurements obtained from the sensor, the maximum measurement value in this set of data is normalized to calculate Y. j `(n)=Y j (n) / Max(Y j (n))j=1,2,...,N,n=1,2,...,I,whereMax(Y) j (n) represents the maximum measurement value of the j-th sensor within the estimated length I, Y j `(n) represents the normalized measurement value of the j-th sensor;

[0046] Each sensor is assigned a different weight based on its type, resulting in an estimated value Yx for the measurements Y from the N sensors after weighting.

[0047]

[0048] Then, the estimated value Yx is denormalized to obtain the fused data from each sensor.

[0049] As one embodiment, the accelerated testing system analyzes the dataset, including the following steps: Analyzing and calculating the dataset using data analysis and data mining techniques to obtain analysis results; comparing the efficiency and stability in the analysis results with a set first standard value to identify potential problems with the current pure electric ship propulsion system; dividing the propulsion power in the analysis results by the sum of battery production carbon emissions and operating energy consumption to obtain the navigation carbon efficiency ratio; judging the environmental performance of the current pure electric ship propulsion system based on the navigation carbon efficiency ratio and the IMO Tier III standard (the IMO Tier III standard is a strict limit standard for ship nitrogen oxide (NOx) emissions established by the International Maritime Organization (IMO) to reduce the negative impact of ships on the environment and human health); generating parameter adjustment commands based on optimization suggestions and sending the parameter adjustment commands to the pure electric ship propulsion system to adjust its hardware parameters; generating compliance optimization suggestions based on the IMO Tier III standard, such as adjusting the propulsion curve to reduce peak power consumption.

[0050] Data analytics refers to the process of processing and analyzing data using statistical, mathematical, and other methods to extract useful information, discover patterns and trends, and support decision-making. The goal of data analytics is to extract meaningful conclusions from large amounts of data to help businesses and organizations better understand their business situation, optimize operations, and improve efficiency and competitiveness.

[0051] Data analysis methods include: Descriptive statistics: calculating the mean, median, variance, and standard deviation of data. Inferential statistics: inferring population characteristics from sample data, such as hypothesis testing and confidence interval estimation. Charts: bar charts, line charts, pie charts, scatter plots, box plots, etc. Dashboards and reports: creating dynamic dashboards and reports using visualization tools (such as Tableau and Power BI). Supervised learning: training models using labeled data for classification and regression tasks. Unsupervised learning: discovering patterns in unlabeled data, such as clustering and dimensionality reduction. Reinforcement learning: learning optimal decision-making strategies through interaction with the environment.

[0052] Data mining is the process of extracting hidden, valuable patterns, associations, and knowledge from large amounts of data. It is typically used to discover unknown regularities and patterns in data that may be significant for decision-making and business optimization. Data mining is an advanced stage of data analysis, focusing more on automated pattern discovery and knowledge extraction.

[0053] Data mining methods include: Hypothesis testing: testing whether a population hypothesis is true using sample data. Regression analysis: establishing a model of the relationship between independent and dependent variables. Supervised learning: training a model using labeled data for classification and regression tasks. Unsupervised learning: discovering patterns in unlabeled data, such as clustering and dimensionality reduction. Deep learning: learning complex patterns in data using neural networks. Data warehousing: building data warehouses to store and manage large amounts of data. Data cubes: discovering patterns in data through multidimensional data analysis.

[0054] Data analysis and data mining are complementary. Data analysis is the foundation of data mining, while data mining is an advanced stage of data analysis. Here's their relationship: Data analysis provides the foundation for data mining. Through descriptive statistics and visualization methods, data analysis helps users understand the basic characteristics and distribution of data. This information provides the basis for data mining, helping users choose appropriate data mining algorithms and models. Data mining is a deepening of data analysis. Through machine learning and statistical methods, data mining discovers hidden patterns and regularities in data. These patterns and regularities can be further used for data analysis, helping users better understand the data and make decisions. In practice, data analysis and data mining are often used together. For example, after understanding the basic characteristics of data through data analysis, data mining algorithms are used to discover patterns and regularities in the data, and then data analysis is used to verify the validity of these patterns and regularities.

[0055] Data analysis and data mining are two important methods for processing and utilizing data. Data analysis focuses on describing and interpreting data, helping users understand its basic characteristics and distribution; data mining, on the other hand, focuses on discovering hidden patterns and regularities within data to support decision-making. They are widely used in fields such as business intelligence, finance, healthcare, and industrial manufacturing. By combining data analysis and data mining, data can be utilized more effectively, improving the scientific rigor and accuracy of decision-making.

[0056] In one embodiment, the accelerated testing system uses anomaly detection algorithms to calculate the acquired multimodal data to identify abnormal signals in the current pure electric ship propulsion system. Based on these abnormal signals, the system locates and determines the fault position in the current pure electric ship propulsion system and issues a fault warning signal. For example, the acoustic signature of battery thermal runaway (20kHz high-frequency whistling) combined with infrared image hotspots (>150℃) improves fault location accuracy by 35%. An LSTM fault prediction model is generated based on historical fault anomaly signals to predict faults in the pure electric ship propulsion system and issue maintenance recommendations. For example, battery pack 3 is expected to have a SOH of 80% after 48 hours.

[0057] As one embodiment, the accelerated testing system stores, backs up, and encrypts the acquired multimodal data, and constructs a data access and connectivity mechanism for the stored multimodal data. It receives user data processing commands to recover or delete deleted multimodal data; or, it receives user input parameters to update the digital twin model.

[0058] As one embodiment, the accelerated testing system includes a user interaction module. The user interaction module generates visualizations from multimodal data. It displays these visualizations, fault location information, fault location assessment results, and warning signals to the user. The user interaction module also receives control commands and feedback from the user.

[0059] A multimodal collaborative accelerated testing system based on a digital twin model includes: a digital twin model construction module, which comprises an entity data acquisition module, a modeling and simulation module, a physical characteristic simulation module, and a model verification module; an experiment design and execution module is connected to the digital twin model construction module; and a multimodal data fusion and analysis module is connected to the experiment design and execution module. The experiment design and execution module includes an experiment scenario setting module, an accelerated operating condition simulation module, an experiment process control module, and a real-time data monitoring module. The experiment scenario setting module sets different accelerated test scenarios to simulate actual operating conditions; the accelerated operating condition simulation module simulates the acceleration process of a ship under different operating conditions; the experiment process control module controls the overall experiment process to ensure the smooth progress of the experiment; and the real-time data monitoring module monitors and records data during the experiment in real time for subsequent analysis. The multimodal data fusion analysis module includes a sensor data acquisition module, an image data acquisition module, an audio data acquisition module, a data processing and cleaning module, a data fusion algorithm module, and a data analysis and mining module. The sensor data acquisition module, image data acquisition module, and audio data acquisition module collect various types of data during the experiment. Then, the data processing and cleaning module preprocesses the collected data, including data cleaning, noise reduction, and format conversion, to ensure the accuracy and usability of the data. The data fusion algorithm module uses data fusion algorithms to fuse multiple types of data to form a comprehensive dataset, so as to gain a deeper understanding of the operating status of the power system. The data analysis and mining module uses data analysis algorithms and mining techniques to conduct in-depth analysis of the fused data and extract valuable information and patterns. The multimodal data fusion and analysis module is connected to a performance evaluation and optimization module. This module includes a performance evaluation module, a stability analysis module, an emissions and energy consumption monitoring module, and an optimization suggestion module. The performance evaluation module comprehensively evaluates the performance of the power system, including efficiency and stability. The stability analysis module analyzes the efficiency and stability of the power system in depth to identify potential problems. The emissions and energy consumption monitoring module monitors the emissions and energy consumption of the power system to evaluate its environmental performance. The optimization suggestion module generates optimization suggestions based on the analysis results, providing a basis for improving the power system.The multimodal data fusion and analysis module is connected to an early warning and fault diagnosis module and a data storage and management module. The early warning and fault diagnosis module includes a real-time status monitoring module, an anomaly detection and identification module, a fault early warning module, and a fault diagnosis and location module. The data storage and management module includes a data storage module, a backup and recovery module, a security and encryption module, and a data query module. The real-time status monitoring module monitors the operating status of the power system in real time to ensure timely detection of anomalies. The anomaly detection and identification module uses algorithms to detect and identify abnormal signals in the system. The fault early warning module issues early warning or alarm signals in a timely manner when a fault is detected. The fault diagnosis and location module diagnoses and locates the fault, facilitating subsequent maintenance and handling. During the test, the data storage and management module stores, backs up, recovers, encrypts, and queries the generated data. Specifically, the data storage module stores relevant data, the backup and recovery module backs up data regularly and restores data when needed, the security and encryption module uses encryption technology to ensure data security and prevent data leakage, and the data query module provides convenient data access and query functions to meet users' data needs. The early warning and fault diagnosis module has a data connection to a user interaction module, which is also connected to the digital twin model construction module, performance evaluation and optimization module, and data storage management module. The user interaction module includes a user interface module, a feedback and help module, a data visualization module, and a real-time data display module. The user interface module provides the operation interface, the feedback and help module collects user feedback and provides necessary help and support, the data visualization module provides intuitive data visualization charts to facilitate understanding of data analysis results, and the real-time data display module displays and updates data in real time to ensure that the latest data information can be obtained.

[0060] Compared with existing technologies, the advantages of this application are as follows: This invention constructs a digital twin model that corresponds one-to-one with the physical ship's power system using digital twin technology. This model not only accurately replicates the structure and function of the physical system but also reflects its operating status and performance parameters in real time. The construction of the digital twin model enables researchers to conduct comprehensive simulation tests of the power system in a virtual environment, greatly improving testing efficiency and accuracy while reducing testing costs. Furthermore, the digital twin model can be used to predict system performance, providing strong support for optimized design and improvement. Moreover, a multimodal data fusion and analysis method is designed, which can simultaneously process multiple types of data, such as sensor data and image data, achieving comprehensive data integration and in-depth mining. This yields more comprehensive and richer information, providing a deeper understanding of the power system's operating status and performance, offering more comprehensive information support for power system research and development and testing, and laying the foundation for subsequent optimization and improvement. Meanwhile, by collecting and analyzing a large amount of data during the operation of the ship's power system, and using data mining and machine learning algorithms, we can achieve comprehensive monitoring and early warning of the system's operating status. This enables us to promptly detect anomalies and potential faults in the system, providing strong support for timely maintenance and repair. Furthermore, in-depth analysis of historical data can also provide a strong basis for the system's optimized design and improvement.

[0061] In the description of this specification, the terms "in one embodiment," "preferred," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. The illustrative expressions of the above terms in this specification do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described can be combined in any suitable manner in one or more embodiments or examples. The connection methods linked in the description of this specification have significant effects and practical utility.

[0062] Based on the above description of the structure and principles, those skilled in the art should understand that this application is not limited to the specific embodiments described above. Improvements and substitutions made using techniques known in the art based on this application all fall within the protection scope of this application and should be defined by the claims.

Claims

1. A multimodal collaborative accelerated testing method based on a digital twin model, applied to an accelerated testing system for accelerating the testing of pure electric ship propulsion systems, characterized in that, The method includes the following steps: Different acceleration test scenarios for pure electric ship propulsion systems were designed, and then the pure electric ship propulsion systems were simulated in different acceleration test scenarios using digital twin models. Multimodal data of pure electric ship propulsion systems were collected and simulated in different acceleration test scenarios. The multimodal data was then processed to obtain a dataset. The dataset is analyzed and calculated to obtain the potential problems and environmental performance of the current pure electric ship power system, and optimization suggestions are generated based on the potential problems and environmental performance.

2. The multimodal collaborative accelerated testing method based on a digital twin model according to claim 1, characterized in that, Before setting up accelerated testing scenarios, the accelerated testing system first constructs a digital twin model, including the following steps: Collect key parameters of the pure electric ship power system, including but not limited to structural parameters, operating parameters and environmental parameters; A digital twin model corresponding to the pure electric ship power system is constructed based on the collected key parameters; Then, the physical characteristics of the pure electric ship propulsion system are simulated in a digital twin model, wherein the physical characteristics include, but are not limited to, power transmission, energy conversion and thermal effects; The actual operating data of the pure electric ship power system is compared with the predicted data of the digital twin model, and the digital twin model is tested and adjusted based on the comparison results.

3. The multimodal collaborative accelerated experimental method based on a digital twin model according to claim 2, characterized in that, The accelerated testing system collects multimodal data from simulations of pure electric ship propulsion systems under different accelerated testing scenarios, including the following steps: The sensor data, image data, and sound data of the pure electric ship power system are collected in different acceleration test scenarios. The sensor data includes, but is not limited to, battery system data, propulsion system data, ship operation data, and environmental safety data. After collecting sensor data, the parameters of the digital twin model are automatically updated using the collected sensor data, and reinforcement learning algorithms are used to optimize and accelerate the boundary conditions of the experimental scenario.

4. The multimodal collaborative accelerated testing method based on a digital twin model according to claim 3, characterized in that, The accelerated testing system processes multimodal data to obtain a dataset, including the following steps: The acquired multimodal data is preprocessed, including data cleaning, noise reduction, and format conversion. The preprocessed data is processed using a data fusion algorithm to obtain the dataset.

5. The multimodal collaborative accelerated testing method based on a digital twin model according to claim 4, characterized in that, The accelerated testing system performs data fusion algorithm calculations on the preprocessed data, including the following steps: Suppose N sensors measure the same value Y at different locations, and each sensor's measurement value is denoted as X. j (j = 1, 2, ... N); For each set of measurements obtained from the sensor, the maximum measurement value in this set of data is normalized to calculate Y. j `(n)=Y j (n) / Max(Y j (n))j=1,2,...,N,n=1,2,...,I,whereMax(Y) j (n) represents the maximum measurement value of the j-th sensor within the estimated length I, Y j `(n) represents the normalized measurement value of the j-th sensor; Each sensor is assigned a different weight based on its type, resulting in an estimated value Yx for the measurements Y from the N sensors after weighting. Then, the estimated value Yx is denormalized to obtain the fused data from each sensor.

6. The multimodal collaborative accelerated testing method based on a digital twin model according to claim 5, characterized in that, The accelerated testing system analyzes the dataset, including the following steps: Data datasets are analyzed and calculated using data analysis and data mining techniques to obtain analytical results; The efficiency and stability in the analysis results are compared with the set first standard values ​​to determine the potential problems of the current pure electric ship power system; Divide the propulsion power in the analysis results by the sum of carbon emissions from battery production and energy consumption during operation to obtain the carbon efficiency ratio of the voyage. The environmental performance of the current pure electric ship propulsion system is judged based on the carbon efficiency ratio of the voyage and the IMO Tier III standard. Based on the optimization suggestions, parameter adjustment commands are generated and sent to the pure electric ship propulsion system, enabling the pure electric ship propulsion system to adjust its own hardware parameters.

7. The multimodal collaborative accelerated testing method based on a digital twin model according to claim 6, characterized in that, The accelerated testing system uses anomaly detection algorithms to calculate the acquired multimodal data in order to identify abnormal signals in the current pure electric ship power system. Based on the abnormal signals, the location and judgment of the fault position of the current pure electric ship power system are determined, and a fault warning signal is issued; An LSTM fault prediction model is generated based on historical fault anomaly signals to predict faults in the power system of pure electric ships and issue maintenance recommendations.

8. The multimodal collaborative accelerated testing method based on a digital twin model according to claim 7, characterized in that, The accelerated testing system stores, backs up, and encrypts the acquired multimodal data, and builds a data access and connectivity mechanism for the stored multimodal data. Receive user data processing commands to recover or delete deleted multimodal data; Alternatively, it can receive user input parameters and update the digital twin model.

9. The multimodal collaborative accelerated testing method based on a digital twin model according to claim 8, characterized in that, The accelerated testing system includes a user interaction module; The user interaction module will create visual charts from multimodal data; The user interaction module will display visual charts, fault location information, fault location judgment results, and warning signals to the user; The user interaction module receives control commands and feedback information from the user.

10. A multimodal collaborative accelerated experimental system based on a digital twin model, characterized in that, include: The module includes a digital twin model module, an experiment design and execution module, a multimodal data fusion and analysis module, a performance evaluation and optimization module, an early warning and fault diagnosis module, a data storage and management module, and a user interaction module. The digital twin model module includes an entity data acquisition module, a physical property simulation module, a modeling and simulation module, and a model verification module; The test design and execution module includes a test scenario setting module, a test process control module, an accelerated operating condition simulation module, and a real-time data monitoring module. The multimodal data fusion and analysis module includes a sensor data acquisition module, a sound data acquisition module, a data fusion algorithm module, an image data acquisition module, a data processing and cleaning module, and a data analysis and mining module. The performance evaluation and optimization module includes a performance evaluation module, an emission and energy consumption monitoring module, a stability analysis module, and an optimization suggestion module. The early warning fault diagnosis module includes a real-time status monitoring module, a fault early warning module, an anomaly detection and identification module, and a fault diagnosis and location module. The data storage management module includes a data storage module, a security and encryption module, a backup and recovery module, and a data query module; The user interaction module includes a user interface module, a data visualization module, a feedback and help module, and a real-time data display module.

Images