Digital twin marine engine health assessment method and system

Abstract

The invention discloses a digital twin marine engine health assessment method, and relates to the technical field of marine engines, and the method comprises the steps: obtaining the operation data of a marine engine, and carrying out the preprocessing of the operation data of the marine engine; constructing a health index function model for reflecting the performance degradation process of the marine engine based on the preprocessed operation data of the marine engine; and collecting real-time operation data of the marine engine, and inputting the real-time operation data into the health index function model to obtain health degree evaluation data of the engine. According to the method, the health indexes capable of representing the degradation characteristics of multiple sensors of the marine engine are constructed based on the time-frequency domain characteristic parameters and kernel principal component analysis, so that the accuracy of health assessment and prediction is improved, and meanwhile, the hyper-parameters in the BiLSTM health prediction model are automatically adjusted and optimized based on the improved HHO algorithm, so that more efficient health state prediction is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of marine engines, and in particular to a digital twin marine engine health assessment method and system. Background Art

[0002] Marine engines are a critical component of ship systems, and their health status is directly related to the safe operation and economic performance of the ship. However, traditional health assessment system research relies primarily on regular inspections and subsequent repairs. This approach not only lacks real-time performance but also fails to effectively predict potential failures, resulting in high maintenance costs and a high risk of sudden failures. Digital twin technology, through real-time interaction between virtual models and physical equipment, enables real-time monitoring, accurate assessment, and prediction of the health status of marine engines, significantly improving maintenance efficiency and equipment reliability. Therefore, designing a digital twin marine engine health assessment system is of great significance for ensuring the safe operation and economic benefits of ships.

[0003] Current research on digital twin marine equipment health assessment systems includes Zhu Mingliang et al. from East China University of Science and Technology, who published an intelligent life prediction method for gearbox bearings based on a digital twin framework (Zhu Mingliang, Lu Wenqing, Liu Yuke et al. A method for intelligent life prediction of gearbox bearings based on a digital twin framework [P]. Chinese Patent: CN119004396A: 2024-11-22). This approach applies digital twins and deep learning technologies to the intelligent operation and maintenance of gearbox bearings through the Unity3D visualization platform, providing a more comprehensive and accurate understanding of the equipment's operating conditions and health status. However, this approach suffers from the following drawbacks: lack of systematic data storage and management; a single Gaussian filtering method easily loses detailed information; the monotonically decreasing health index constructed through linear fitting is inconsistent with the actual nonlinear degradation of marine equipment; manual adjustment of multiple hyperparameter combinations is time-consuming and prone to omissions, resulting in low efficiency; simple 3D model visualization fails to reflect actual operating conditions, failing to provide effective early warnings and maintenance guidance; and the user experience lacks the efficient operation and maintenance provided by terminal interaction. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the shortcomings of the existing technology and provide a digital twin marine engine health assessment method and system.

[0005] In order to solve the above technical problems, the technical solutions of the present invention are as follows: A digital twin marine engine health assessment method, comprising: Acquire marine engine operating data and pre-process the marine engine operating data; Based on the pre-processed marine engine operating data, a health index function model is constructed to reflect the performance degradation process of the marine engine. Real-time operating data of the marine engine is collected and input into the health index function model to obtain engine health assessment data.

[0006] As a preferred embodiment of the digital twin marine engine health assessment method of the present invention, after collecting the real-time operating data of the marine engine and inputting it into the health index function model to obtain the engine health assessment data, the method further includes: Based on the preprocessed marine engine operation data, a health prediction model based on IHHO-BiLSTM is constructed, and the engine health prediction data is obtained.

[0007] As a preferred solution of the digital twin marine engine health assessment method of the present invention, after constructing the IHHO-BiLSTM-based health prediction model based on the preprocessed marine engine operating data and obtaining the engine health prediction data, the method further includes: A marine engine health assessment visualization platform is built based on the Unity3D platform.

[0008] As a preferred embodiment of the digital twin marine engine health assessment method of the present invention, the steps of obtaining marine engine operating data and preprocessing the marine engine operating data include: Acquiring marine engine operation data through sensors deployed at marine engine components; De-noising the marine engine operating data based on an adaptive de-noising autoencoder to obtain pre-processed marine engine operating data; The pre-processed marine engine operating data is stored in the database.

[0009] As a preferred embodiment of the digital twin marine engine health assessment method of the present invention, the health index function model for reflecting the marine engine performance degradation process based on the preprocessed marine engine operating data includes: Characteristic parameters that can reflect the performance degradation process of marine engines are selected and their monotonicity analysis is performed. The monotonicity calculation formula of characteristic parameters is: ,in, is a monotonic function with a value range of 0 to 1. represents the feature sequence extracted by the device, Indicates that the device is The characteristic sequence of moments, Indicates the length of the device feature sequence, represents the differentials of adjacent values ​​in the sequence, and Represents the calculated values ​​of positive and negative differentials respectively; Based on the monotonicity value, the feature parameters are sorted from large to small, and the first five feature parameters are selected to construct a feature set for kernel principal component analysis. The kernel function is: ; The first principal component characteristic parameter after domain projection of the characteristic parameter with the best monotonicity in the time-frequency domain is compared with the first kernel principal component characteristic parameter, and the characteristic parameter with the best monotonicity is selected to construct a health index function model for the marine engine; The sensor engine is divided into health status levels, and the final health index function model is obtained by max-min normalization: , where max and min represent the maximum and minimum values ​​of the characteristic parameters for constructing the health index function, respectively.

[0010] As a preferred solution of the digital twin marine engine health assessment method of the present invention, the steps of constructing an IHHO-BiLSTM-based health prediction model based on preprocessed marine engine operating data and obtaining engine health prediction data include: Divide the dataset into training, validation, and test sets; Initialize the BiLSTM hyperparameters and IHHO parameters; Tent mapping is used to initialize the population. A BiLSTM prediction model is trained for each individual and the RMSE is calculated. The optimal individual is selected as the prey location. The escape energy is updated using a cosine annealing strategy, and a chaotic search strategy is used to enhance population diversity. The flight path of the Harris hawk is adjusted based on the prey location and the optimal solution is updated. Determine whether the maximum number of iterations has been reached. If not, return to the previous step to continue optimization. If reached, output the global optimal solution and the optimal position of the prey. The BiLSTM health status prediction model is updated and trained based on the output optimal result, and the engine health prediction data is obtained based on the trained BiLSTM health status prediction model.

[0011] As a preferred solution of the digital twin marine engine health assessment method of the present invention, the marine engine health assessment visualization platform based on the Unity3D platform includes: Use Python as a mid-end server to connect the Unity3D platform to the database based on the TCP / IP protocol; Use Python as a TCP server and Unity3D as a TCP client. Unity3D sends a data query request. After receiving the data, Python connects and interacts with the database through database commands and sends the data to Unity3D via TCP. The Unity 3D platform parses the received data and displays it on the interface.

[0012] The present invention also provides a digital twin marine engine health assessment system, comprising: A data acquisition module is used to obtain and pre-process the marine engine operating data; A first model building module is used to build a health index function model for reflecting the performance degradation process of the marine engine based on the preprocessed marine engine operation data; The second model building module is used to build an IHHO-BiLSTM-based health prediction model based on the preprocessed marine engine operation data and obtain engine health prediction data; The visualization module is used to build a marine engine health assessment visualization platform based on the Unity3D platform.

[0013] The present invention also provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the method described in any one of the above-mentioned digital twin marine engine health assessment methods is implemented.

[0014] The present invention also provides a computer-readable storage medium storing a computer program, characterized in that when the program is executed by a processor, the method described in any of the above-mentioned digital twin marine engine health assessment methods is implemented.

[0015] The beneficial effects of the present invention are: Compared with the existing technology, the present invention stores, manages, queries and interacts with database data such as real-time sensor data, historical data, digital twin model data and assessment and warning data of marine engines; uses an adaptive denoising autoencoder to effectively eliminate different types of noise while retaining key fault characteristics; constructs health indicators that can characterize the degradation characteristics of multiple sensors of marine engines based on time-frequency domain feature parameters and kernel principal component analysis, thereby improving the accuracy of health assessment prediction; automatically tunes the hyperparameters in the BiLSTM health prediction model based on the improved HHO algorithm, thereby achieving more efficient health status prediction; constructs an intuitive health assessment visualization module based on the digital twin visualization platform, providing effective health warning and maintenance guidance; based on terminal interaction and application layer design, it improves the user's operating efficiency and operation and maintenance efficiency, and realizes intelligent operation and maintenance and decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0017] Figure 1 A schematic flow chart of the digital twin marine engine health assessment method provided by the present invention; Figure 2 This is a model framework diagram of the adaptive denoising autoencoder in the digital twin marine engine health assessment method provided by the present invention; Figure 3 A flowchart for constructing a health indicator function model in the digital twin marine engine health assessment method provided by the present invention; Figure 4 A flowchart for constructing a health prediction model based on IHHO-BiLSTM in the digital twin marine engine health assessment method provided by the present invention; Figure 5 This is a diagram of the TCP / IP protocol cross-platform communication data interaction framework in the digital twin marine engine health assessment method provided by the present invention; Figure 6 Schematic diagram of a marine engine health assessment visualization platform according to an embodiment of the present invention; Figure 7 The overall framework diagram of the marine engine health assessment method based on digital twin; Figure 8 A schematic diagram of the marine engine health assessment system provided by the present invention; Figure 9 A schematic diagram of a computer device provided by the present invention. DETAILED DESCRIPTION

[0018] In order to make the contents of the present invention more clearly understood, the present invention is further described below in detail based on specific implementation methods in conjunction with the accompanying drawings.

[0019] In response to the problems mentioned in the above background technology, such as the lack of systematic storage and management of data in the related technologies, the easy loss of detailed information by a single Gaussian filtering processing method, the error between the monotonically decreasing health index constructed by linear fitting and the actual nonlinear degradation of ship equipment, the time-consuming and easy omission of multiple hyperparameter combinations by manual adjustment, the low efficiency, the lack of reflection of the actual operating status by the visualization of simple three-dimensional models, the inability to provide effective early warning and maintenance guidance, and the lack of efficient operation and maintenance brought to users by terminal interaction, the present invention provides a digital twin marine engine health assessment method and system, which solves the problem that the single Gaussian filtering processing method easily loses detailed information; constructs a health indicator that can characterize the degradation characteristics of multiple sensors of ship equipment, thereby improving the accuracy of health assessment prediction; automatically tunes the hyperparameters in the health prediction model based on deep learning to achieve more efficient health status prediction; constructs an intuitive health assessment visualization module based on the digital twin visualization platform, which provides effective health early warning and maintenance guidance; based on terminal interaction and application layer design, improves the user's operation efficiency and operation efficiency, and realizes intelligent operation and decision-making.

[0020] See also Figures 1 to 6 The embodiment of the present invention provides a digital twin marine engine health assessment method, which mainly studies the fuel system of a typical marine engine system and specifically includes the following steps: Step S101: Acquire marine engine operating data and pre-process the marine engine operating data.

[0021] Specifically, sensors are installed on components in the fuel system, including the fuel tank, fuel pump, fuel filter, fuel injector, fuel pressure regulator, fuel supply line, high-pressure fuel line, fuel injector, sensors and control unit, governor, fuel cooler, oil-water separator, pilot oil pump, pilot oil tank, and pilot oil nozzle, to collect health assessment and prediction data. A signal conversion box converts sensor data via the RS485 interface and MODBUS RTU communication protocol to obtain marine engine operating parameters, and the data is recorded in decimal format.

[0022] It is understandable that when a ship is sailing, there is severe noise interference around the engine, and different types of noise signals will be mixed into the original signal, which will affect and deviate from the signal characteristics of the marine engine fuel system, the reliability of the health indicator construction, the accuracy of the health status prediction, and the authenticity of the digital twin model. Therefore, this embodiment uses an improved adaptive denoising autoencoder model to denoise the original signal.

[0023] Adaptive denoising autoencoder model such as Figure 2As shown in Figure 1, it mainly consists of a Dropout layer, a residual connection layer, an encoder module, a decoder module, and an adaptive shrinkage unit module. The Dropout layer randomly discards some neurons to improve the generalization ability of the model.

[0024] An autoencoder represents a neural network designed to accurately replicate the input data, extracting essential features while discarding redundant information. An autoencoder consists of an input layer, a hidden layer, and an output layer. Its structure is to encode the input data into a lower-dimensional latent space and then decode it back to its original form. Transformed into a latent representation through an encoder , the formula is as follows: ,in, represents the weight matrix of the encoder, represents nonlinear changes, represents the bias vector of the encoder. Afterwards, the decoder network reconstructs the latent representation Return to original input , this process involves another set of nonlinear change formulas as follows: ,in, represents the nonlinear change of the decoder, represents the weight matrix of the decoder, Represents the bias vector of the decoder.

[0025] In the modular design of encoder and decoder, a new Adaptive Shrinkage Unit (ASU) is selected as the replacement. ASU uses a local attention mechanism to train the shrinkage coefficients and uses these coefficients to attenuate the noise in the input signal. This method allows ASU to adaptively determine the optimal shrinkage value for different parts of the signal, thereby effectively eliminating noise while retaining the basic features related to the fault. Specifically, ASU consists of a convolutional layer or deconvolution layer, a The activation function and batch normalization are used to limit the shrinkage to the range [0,1]. The function ensures that the coefficients appropriately shrink the input signal, reducing the noise component while retaining important fault information.

[0026] The obtained shrinkage coefficient is multiplied by the noisy feature signal after two layers of convolution (deconvolution), two layers of batch normalization, and a ReLU layer. This dynamically and effectively attenuates the noise signal and retains the important signal features. At the same time, the residual connection F(x) is added to avoid the problem of gradient vanishing during network model training, thereby promoting smoother training and avoiding information loss during the convolution process, which helps to learn effective feature representations faster and more efficiently. The latent feature representation is output through the fully connected layer in the encoder. , the decoder predicts the continuous value of the target variable through the regression layer to generate the reconstructed signal .

[0027] The pre-processed marine engine operation data is put into the MySQL database for storage and management.

[0028] Step S102: constructing a health index function model for reflecting the performance degradation process of the marine engine based on the preprocessed marine engine operation data.

[0029] Specifically, after preprocessing and storing marine engine data, a health indicator function that reflects the performance degradation process of marine engines is constructed to implement health assessment. First, characteristic parameters that reflect the performance degradation process of marine engines are selected. The time domain includes mean, standard deviation, skewness, kurtosis, peak-to-peak value, root mean square, crest factor, shape factor, impulse factor, edge factor, and energy; the frequency domain includes spectral mean, spectral variance, spectral skewness, and spectral kurtosis.

[0030] In order to better describe the irreversible performance degradation failure process of marine engines, a reasonable health index curve should be monotonically increasing or monotonically decreasing, which is called monotonicity. The monotonicity calculation formula of the equipment characteristic quantity is as follows: ,in, is a monotonic function with a value range of 0 to 1. represents the feature sequence extracted by the device, Indicates that the device is The characteristic sequence of moments, Indicates the length of the device feature sequence, represents the differentials of adjacent values ​​in the sequence, and Indicates the calculated values ​​of the differential as positive and negative, respectively.

[0031] Based on the monotonicity value, the feature parameters are sorted from large to small, and the first five feature parameters are selected to construct a feature set for kernel principal component analysis. The kernel function is: ,in, The eigenvectors of the two eigenvalues ​​with higher cumulative contribution rates are combined into a projection space, and the corresponding core principal component can be obtained after the marine engine feature set is centrally projected in the projection space.

[0032] The first principal component characteristic parameter after domain projection of the characteristic parameter with the best monotonicity in the time-frequency domain is compared with the first kernel principal component characteristic parameter, and the characteristic parameter with the best monotonicity is selected to construct the health index function of the marine engine.

[0033] The health status of each component in a marine engine fuel system is categorized into five quantitative levels: healthy, subhealthy, abnormal, faulty, and scrapped. The health index for healthy status ranges from [0.9, 1], indicating normal operation with no abnormalities. The health index for subhealthy status ranges from [0.8, 0.9], indicating normal operation with minor abnormalities. The health index for abnormal status ranges from [0.7, 0.8], indicating a slight decrease in equipment performance, a minor fault, and a fault warning. The health index for faulty status ranges from [0.4, 0.7], indicating poor performance, signs of component damage, and requiring downtime for repair. The health index for scrapped status ranges from [0, 0.4], indicating the equipment is scrapped.

[0034] In order to eliminate the influence of dimensional variation, the health index results are mapped to [0-1], and the final health index function formula is as follows: , where max and min represent the maximum and minimum values ​​of the characteristic parameters for constructing the health index function, respectively.

[0035] After constructing the health index function model, by collecting the real-time operating data of the marine engine and inputting it into the health index function model, the engine health assessment data can be obtained to realize the health assessment of the marine engine.

[0036] Step S103: constructing an IHHO-BiLSTM-based health prediction model based on the preprocessed marine engine operation data, and obtaining engine health prediction data.

[0037] Health indicator functions can be used to assess the health of marine engines online, but they cannot predict their health. BiLSTM neural networks can solve this prediction problem. However, the prediction accuracy of a single BiLSTM model is low. This embodiment uses an improved IHHO algorithm to automatically optimize the hyperparameters of the BiLSTM neural network, enabling prediction of equipment health.

[0038] The Long Short-Term Memory (LSTM) neural network adds three "gate" structures on the basis of RNN to solve the problems of gradient explosion and gradient disappearance. The forget gate determines whether the memory state of the previous time step will be retained. Function activation, value range [0,1], forget gate calculation formula is as follows: ,in, yes function, and are learnable parameters, and Represent the short-term memory state of the previous time step and the current input data respectively.

[0039] The input gate determines the amount of new information extracted from the current input data and is also determined by Activate, and also The activation function generates new memory candidates, and the input gate calculation formula is as follows: ,in, 、 、 、 are all learnable parameters. Indicates a new memory candidate.

[0040] The output gate determines what information the output at the current moment is based on. The activation function is used to obtain the proportional coefficient of the combination of long-term and short-term memory. This coefficient determines the specific part of the output long-term memory state. The output gate calculation formula is as follows: ,in, 、 are learnable parameters, Represents current long-term memory.

[0041] BiLSTM makes up for the deficiency of LSTM in encoding information from back to front. It is composed of a forward and a reverse LSTM network, and the forward propagation layer and the backward propagation layer are connected to the output layer. t The forward calculation is repeated at each moment to obtain and save the output of the hidden layer at each moment. t The backward calculation is performed to time 0, and the output of the backward hidden layer at each time is obtained and saved. Finally, the final output is obtained by combining the output of the forward propagation layer and the backward propagation layer at the corresponding time.

[0042] The Improved Harris Hawks Optimization (IHHO) algorithm, based on the Harris Hawk Optimization (HHO) algorithm, adopts strategies such as Tent mapping, cosine annealing, and chaos search to automatically optimize hyperparameters such as the number of hidden layer neurons, maximum training cycle, initial learning rate, and L2 regularization coefficient, solving the problem of low prediction accuracy caused by the difficulty in setting hyperparameters of BiLSTM.

[0043] The HHO algorithm is a heuristic optimization algorithm that simulates the process of a Harris hawk hunting prey. It has the advantages of few parameters, strong global search capabilities, and high computational efficiency. The algorithm is mainly divided into an exploration phase, an exploration-to-exploitation transition phase, and an exploitation phase. In the exploration phase, the Harris hawk perches at a certain location and updates its position according to the probability q. The formula is as follows: , where For the next iteration The position vector of the middle eagle; For prey location; is the eagle's current position vector; 、 、 、 and q are random numbers in (0,1), updated at each iteration; and To optimize the number of hidden layer neurons, maximum training cycle, initial learning rate and upper and lower limits of L2 regularization coefficient in BiLSM; is the random position of the eagle in the current population; is the current average position of the eagle group.

[0044] In the exploration to development transition phase, Harris Hawk escapes energy As a basis for judgment, switch between different strategies in global exploration and local development 2. The escape energy formula is as follows: ,in is the current iteration number, is the maximum number of iterations; is the initial state of the escape energy, and changes randomly in the interval (-1,1) at each iteration.

[0045] During the development phase, the Harris Hawk conducts local searches based on the escape energy. and the chance of escape r to hunt using one of four different strategies. The formula is as follows: in, is the energy intensity during escape, is a random number in the interval (-1,1); when The formula is as follows: ; when The formula is as follows: ; when The formula is as follows: ,in, is a random vector of dimension D, is the fitness function, is the Levy flight function.

[0046] In the aforementioned HHO algorithm, the randomly generated initial population has low diversity, which is not conducive to rapid convergence of the algorithm. Tent mapping has strong ergodicity, uniformity, and randomness, which can effectively cover a larger search space and improve population diversity. Therefore, this paper adopts the Tent mapping method for population initialization. The formula is as follows: ,in It has the most typical form when it is equal to 0.5, and the value of this embodiment is 0.5.

[0047] The switch between global search and local development is determined by the escape energy E However, due to its linear reduction characteristics, the later E Small, individuals almost only conduct local search and are prone to fall into local optimality. To solve this problem, this paper adopts the cosine annealing strategy for nonlinear processing. The specific calculation formula is as follows: .

[0048] Since all individuals in the group gather towards the optimal individual position, the group diversity is indeed. To solve this problem, a chaotic search strategy is used to linearly map the current optimal individual position vector to the optimization space of chaotic variables. First, each dimension of the optimal individual position vector is mapped to (0,1) for normalization to generate variables. .in and are the maximum and minimum values ​​of the population respectively, and the formula is as follows: .

[0049] Will Mapped to the original solution space through the formula, a new optimal individual position vector is generated , the formula is as follows: .

[0050] Will Iterate and compare the fitness function values ​​after iteration to retain the better individual positions.

[0051] When the number of iterations reaches the maximum number of chaotic search times, the search is terminated. At the same time, the mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE) and goodness of fit (R 2 ) as the evaluation criteria to compare the model prediction results. The closer MAE, MSE, and RMSE are to 0, the better the R 2 The closer it is to 1, the higher the prediction accuracy of the model and the stronger the model performance.

[0052] The final health prediction model based on IHHO-BiLSTM is as follows: Figure 4 As shown, the specific steps include: Step S103a: Divide the data set into training set, validation set and test set in a ratio of 0.6:0.2:0.2; Step S103b: Initialize the hyperparameters of BiLSTM: the individual representation is to set the number of neurons in the range of [64, 1024], the learning rate in the range of [1e-5, 1e-3], the number of training rounds not exceeding 200, and the L2 regularization coefficient in the range of [0.001, 0.1].

[0053] Step S103c: Initialize IHHO parameters: The population size is set to 50 individuals, each individual represents the dimension of 4 hyperparameters. The maximum number of iterations is set to 20, and the initial escape energy is set to Set to 0.1, the number of chaotic searches K is 10, and the Adam optimizer is used to train the model parameters.

[0054] Step S103d: Initialize the population using Tent mapping, train BiLSTM for each individual and calculate RMSE, select the current optimal individual as the prey position; update the escape energy according to the cosine annealing strategy E , dynamically adjust the ratio of global search to local exploitation; use chaotic search strategies to enhance population diversity and avoid falling into local optimality. Based on the location of prey, adjust the flight path of Harris's hawk and update the optimal solution.

[0055] Step S103f: Iterative determination: After updating the optimal position of the prey, determine whether the algorithm has reached the maximum number of iterations. If so, output the global optimal solution and the optimal position of the prey. Otherwise, return to step S103d to continue optimization.

[0056] Step S103g: Build a BiLSTM health status prediction model based on the output optimal result and train it, predict the health value and calculate the RMSE and R 2 Equal error values ​​are used to compare the health status prediction results of different types of algorithms.

[0057] Step S104: Building a marine engine health assessment visualization platform based on the Unity3D platform.

[0058] Specifically, the MySQL database is an open source, cross-platform database management software based on SQL language queries. In the digital twin-based marine engine health assessment system, the database plays a vital role. It is used to store and manage various engine-related data: real-time marine engine monitoring data, historical monitoring data, health status records, maintenance history, health prediction results, digital twin model data, etc.

[0059] The MySQL database visual management tool MySQL Workbench is used to manage the database. The main table items include: Column Name column name; Datatype data type; PK primary key in the data table, used to index a certain data; NN not empty; UQ value unique; B binary; UN unsigned; ZF fills with 0; AI value automatically increases; G this column is calculated based on other columns; INT integer; VARCHAR string; TEXT character data with unlimited storage size; Default / Expression default / expression.

[0060] Based on the TCP / IP protocol, Python is used as the mid-end server by compiling C# scripts to connect the digital twin visualization status monitoring platform Uniyt3D with the MySQL database, visualizing the various engine-related data stored and managed.

[0061] The Python server interacts with the MySQL database, builds SQL queries and sends them to the MySQL database, and performs health assessment predictions based on historical data or real-time data. Using Python as a TCP server and Unity3D as a TCP client, the Unity 3D client first initializes the port number and sends a connection request. The Python server listens on the port and accepts the request. After receiving the information, the client will reply to the server again to confirm, thus completing the "three-way handshake" and establishing a connection. Unity 3D sends requests such as health assessment predictions and data queries. After receiving the data, the Python server interacts with the MySQL database through database command connections and sends it back to Unity 3D via TCP. Unity 3D parses the received results and displays them on the interface. After the data transmission is completed, the client and server determine whether the connection needs to be closed and perform a disconnection operation to ensure the stability and efficiency of communication. Finally, a cross-platform communication data interaction based on the TCP / IP protocol is constructed, such as Figure 5 shown.

[0062] The 3D model built in Solidworks was imported into Assests in Unity3D. The movement of fuel in the fuel system was simulated by adding a Particle System. The Emission parameters, Shape, Renderer and other parameters in the particle system were adjusted to simulate the flow of fuel, combustion in the engine chamber, and filtering by the fuel filter. Rigidbodies were set for the components and Colliders were configured to ensure that the movement of fuel in the fuel system was more realistic. At the same time, the movement of mechanical components can be simulated by compiling functions and formulas in C# scripts.

[0063] Scripts were used to embed component information from MySQL data into the model. Clicking a component displayed component-related information, such as the part number, physical information, and name. C# script object modules connected to the MySQL database were added to each component monitoring point to act as sensors. When the user clicked the module, monitoring data from the corresponding measurement point appeared, enabling model linkage.

[0064] In Unity3D, a Canvas is added as the display basis for the visualization of marine engine health assessment, and the RawImage element is used as the carrier of real-time operating data, health index, data time-frequency domain curves, and health prediction results.

[0065] Setting the Canvas rendering mode to Screen Space – Overlay maintains the visual perspective of the elements. Text is added to represent data axis labels such as time, sensor type, and health index (HI). Interaction with the MySQL database is achieved through C# scripting. Custom functions are used to generate visualizations of real-time data, health assessment results, and health predictions in the form of pointer gauges, bar charts, pie charts, and line segments.

[0066] The health levels of marine engine components of different health levels are color-coded, with green representing healthy, light green representing sub-healthy, yellow representing abnormal, orange representing faulty, and red representing scrapped. The health index ranges and corresponding health levels are processed as described in step S102. The colors of the indicator instrument panel, bar chart, pie chart, and line segment and point icons and curves are set according to the health index ranges and health levels.

[0067] When the health index value reaches abnormal, faulty, or scrapped status, a C# script is designed to display the warning sign on the warning sign model according to the corresponding color and health status. At the same time as the warning sign is displayed, the alarm sound file is imported into Assets, an Audio Source component is added to the Scene, and the alarm sound file is set as a Clip in the Audio Source. When the user observes or hears the situation, he clicks the mouse and performs maintenance according to the fault condition and maintenance recommendations.

[0068] Design the login and registration interfaces for the marine engine health assessment system software. The login interface primarily includes entering a username and password, displaying the password, entering a verification code, following the verification code, logging in, selecting automatic login, and a button to register a new user. The registration interface primarily includes entering a username, entering a password, re-entering the password, receiving a registration and login feedback prompt, and a button to register. Once the user's account and password match the database, the login is complete. If the user does not have an account, they must first register an account and then log in. This closed-loop process allows access to the marine engine health assessment system.

[0069] After the user registration is completed, different permissions are assigned to different roles and role levels using SQL statements in MySQL. Permissions include: reading, writing, adding, modifying, and deleting. After entering the page, the main module entrances of the system are provided, including the health assessment visualization module, the database module, and the model library module. The database module mainly contains main information such as hierarchical tree information, equipment information, operating data, case analysis, maintenance suggestions, historical maintenance records, and collection device and measurement point configuration. The software database data is displayed by connecting to the MySQL database. The model library module mainly contains different algorithm models, such as automatic optimization algorithm models, health index function models, health status prediction models, denoising models, and other storage models. Model information mainly includes model name, algorithm type, parameter configuration, optimization goals, and model description.

[0070] In the health assessment visualization platform, corresponding maintenance measures are selected based on the corresponding health index and health status output by the assessment and prediction results. If it is healthy, keep running; if it is sub-healthy, there are abnormal signs and attention should be paid to the health status of the equipment; if it is in an abnormal state, there is a minor fault and a fault warning display; if it is in a fault state, there is damage to components and it needs to be shut down for maintenance; if it is in a scrapped state, the equipment has been scrapped and cannot be used anymore.

[0071] After completing the design of each module, the mode switch button is designed in the navigation bar. According to the mode selected by the user, the user jumps to the corresponding module. At the same time, a button for directly exporting the required data to CSV, Excel and other format files can be set above the corresponding data. Finally, a digital twin-based marine engine health assessment visualization platform is built. Figure 6 shown.

[0072] Figure 7 The overall framework diagram of the digital twin marine engine health assessment method provided in this embodiment. Compared with the existing technology, this embodiment stores, manages, queries and interacts with database data such as real-time sensor data, historical data, digital twin model data and assessment and warning data of marine engines; uses an adaptive denoising autoencoder to effectively eliminate different types of noise while retaining key fault characteristics; constructs health indicators that can characterize the degradation characteristics of multiple sensors of marine engines based on time-frequency domain feature parameters and kernel principal component analysis, thereby improving the accuracy of health assessment prediction; automatically tunes the hyperparameters in the BiLSTM health prediction model based on the improved HHO algorithm, thereby achieving more efficient health status prediction; constructs an intuitive health assessment visualization module based on the digital twin visualization platform, providing effective health warnings and maintenance guidance; and based on terminal interaction and application layer design, improves user operation and maintenance efficiency and realizes intelligent operation and decision-making.

[0073] Figure 8 A schematic diagram of a digital twin marine engine health assessment system provided in this embodiment. The system includes a data acquisition module, a first model building module, a second model building module, and a visualization module.

[0074] Among them, the data acquisition module is used to obtain the marine engine operating data and pre-process the marine engine operating data.

[0075] The first model building module is used to build a health index function model for reflecting the performance degradation process of the marine engine based on the preprocessed marine engine operation data.

[0076] The second model building module is used to build a health prediction model based on IHHO-BiLSTM based on the preprocessed marine engine operation data and obtain engine health prediction data.

[0077] The visualization module is used to build a marine engine health assessment visualization platform based on the Unity3D platform.

[0078] See also Figure 9This embodiment also provides a computer device, the components of which may include but are not limited to: one or more processors or processing units, a system memory, and a bus connecting different system components (including the system memory and the processing unit).

[0079] The term "bus" refers to one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures. Examples of these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0080] The computer system / server typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the computer system / server, including volatile and non-volatile media, removable and non-removable media.

[0081] The system memory may include computer system readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory. The computer device may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system may be used to read and write non-removable, non-volatile magnetic media. A disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk") may be provided, as well as an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media). In these cases, each drive may be connected to the bus via one or more data medium interfaces. The memory may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of various embodiments of the present invention.

[0082] A program / utility having a set (at least one) of program modules, which may be stored, for example, in a memory, includes, but is not limited to, an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may include an implementation of a network environment. The program modules generally perform the functions and / or methods of the embodiments described herein.

[0083] A computer device may also communicate with one or more external devices, such as a keyboard, pointing device, display, etc. Such communication may be performed via an input / output (I / O) interface. Furthermore, a computer device may also communicate with one or more networks, such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet, via a network adapter.

[0084] The processing unit executes the functions and / or methods described in the embodiments of the present invention by running the programs stored in the system memory.

[0085] The above-mentioned computer program can be set in a computer storage medium, that is, the computer storage medium is encoded with a computer program, and when the program is executed by one or more computers, it enables one or more computers to perform the method flow and / or device operation shown in the above-mentioned embodiments of the present invention.

[0086] As time goes by and technology develops, the meaning of medium becomes more and more extensive. The dissemination path of computer programs is no longer limited to tangible media, and can also be downloaded directly from the Internet. Any combination of one or more computer-readable media can be used. Computer-readable media can be computer-readable signal media or computer-readable storage media. Computer-readable storage media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or components, or any combination of the above. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, device or device.

[0087] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take a variety of forms, including, but not limited to, electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0088] Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0089] Computer program code for performing the operations of the present invention may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0090] In addition to the above embodiments, the present invention may also have other implementation methods; any technical solutions formed by equivalent replacement or equivalent transformation fall within the protection scope required by the present invention.

Claims

1. A digital twin marine engine health assessment method, characterized by: include: Acquire marine engine operating data and pre-process the marine engine operating data; Based on the pre-processed marine engine operating data, a health index function model is constructed to reflect the performance degradation process of the marine engine. Real-time operating data of the marine engine is collected and input into the health index function model to obtain engine health assessment data.

2. The digital twin marine engine health assessment method according to claim 1, characterized in that: After collecting the real-time operating data of the marine engine and inputting it into the health index function model to obtain the engine health assessment data, the method further includes: Based on the preprocessed marine engine operation data, a health prediction model based on IHHO-BiLSTM is constructed, and the engine health prediction data is obtained.

3. The digital twin marine engine health assessment method according to claim 2, characterized in that: After constructing the IHHO-BiLSTM-based health prediction model based on the preprocessed marine engine operating data and obtaining engine health prediction data, the method further includes: A marine engine health assessment visualization platform is built based on the Unity3D platform.

4. The digital twin marine engine health assessment method according to claim 3 is characterized by: The obtaining of the marine engine operating data and preprocessing the marine engine operating data includes: Acquiring marine engine operation data through sensors deployed at marine engine components; De-noising the marine engine operating data based on an adaptive de-noising autoencoder to obtain pre-processed marine engine operating data; The pre-processed marine engine operating data is stored in the database.

5. The digital twin marine engine health assessment method according to claim 1, characterized in that: The health index function model for reflecting the performance degradation process of the marine engine is constructed based on the pre-processed marine engine operation data, including: Characteristic parameters that can reflect the performance degradation process of marine engines are selected and their monotonicity analysis is performed. The monotonicity calculation formula of characteristic parameters is: ,in, is a monotonic function with a value range of 0 to 1. represents the feature sequence extracted by the device, Indicates that the device is The characteristic sequence of moments, Indicates the length of the device feature sequence, represents the differentials of adjacent values ​​in the sequence, and Represents the calculated values ​​of positive and negative differentials respectively; Based on the monotonicity value, the feature parameters are sorted from large to small, and the first five feature parameters are selected to construct a feature set for kernel principal component analysis. The kernel function is: ; The first principal component characteristic parameter after domain projection of the characteristic parameter with the best monotonicity in the time-frequency domain is compared with the first kernel principal component characteristic parameter, and the characteristic parameter with the best monotonicity is selected to construct a health index function model for the marine engine; The sensor engine is divided into health status levels, and the final health index function model is obtained by max-min normalization: , where max and min represent the maximum and minimum values ​​of the characteristic parameters for constructing the health index function, respectively.

6. The digital twin marine engine health assessment method according to claim 2, characterized in that: The process of constructing a health prediction model based on IHHO-BiLSTM based on the preprocessed marine engine operation data and obtaining engine health prediction data includes: Divide the dataset into training, validation, and test sets; Initialize the BiLSTM hyperparameters and IHHO parameters; Tent mapping is used to initialize the population. A BiLSTM prediction model is trained for each individual and the RMSE is calculated. The optimal individual is selected as the prey location. The escape energy is updated using a cosine annealing strategy, and a chaotic search strategy is used to enhance population diversity. The flight path of the Harris hawk is adjusted based on the prey location and the optimal solution is updated. Determine whether the maximum number of iterations has been reached. If not, return to the previous step to continue optimization. If reached, output the global optimal solution and the optimal position of the prey. The BiLSTM health status prediction model is updated and trained based on the output optimal result, and the engine health prediction data is obtained based on the trained BiLSTM health status prediction model.

7. The digital twin marine engine health assessment method according to claim 4, characterized in that: The construction of a marine engine health assessment visualization platform based on the Unity3D platform includes: Use Python as a mid-end server to connect the Unity3D platform to the database based on the TCP / IP protocol; Use Python as a TCP server and Unity3D as a TCP client. Unity3D sends a data query request. After receiving the data, Python connects and interacts with the database through database commands and sends the data to Unity3D via TCP. The Unity 3D platform parses the received data and displays it on the interface.

8. A digital twin marine engine health assessment system, characterized by: include: A data acquisition module is used to obtain and pre-process the marine engine operating data; A first model building module is used to build a health index function model for reflecting the performance degradation process of the marine engine based on the preprocessed marine engine operation data; The second model building module is used to build an IHHO-BiLSTM-based health prediction model based on the preprocessed marine engine operation data and obtain engine health prediction data; The visualization module is used to build a marine engine health assessment visualization platform based on the Unity3D platform.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images