Marine engine lubricating oil system fault diagnosis method and system based on transfer learning from simulation domain to real domain

Abstract

The invention provides a marine engine lubricating oil system fault diagnosis method and system based on transfer learning from a simulation domain to a real domain, and the method comprises the steps: carrying out the fault simulation of different working conditions based on a built marine engine lubricating oil system simulation model, and obtaining a simulation fault signal response; acquiring real domain fault data of different working conditions of an engine lubricating oil system based on an engine experiment platform; a feature fusion structure based on an encoder-decoder is adopted to obtain simulation domain fault data; an engine lubricating oil system fault diagnosis model based on data migration and algorithm migration dual migration learning is obtained; and based on the engine lubricating oil fault diagnosis model, performing real-time fault diagnosis on the marine engine lubricating oil system to obtain a diagnosis result. According to the method, transfer learning from simulation domain fault features to real domain fault features is realized, so that the problems of high construction cost and insufficient generalization performance of a diagnosis model are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of engine fault diagnosis, and in particular relates to a method and system for diagnosing faults in a marine engine lubricating oil system based on transfer learning from a simulation domain to a real domain. Background Art

[0002] Engines are widely used in transportation, industrial production, power generation and other fields due to their efficient energy utilization and good economy. However, as the engine runs for a long time, its internal lubricating oil system is prone to various faults, such as oil filter blockage, insufficient lubricating oil, etc. These faults will not only cause a significant decline in engine performance, but also may cause serious safety hazards, and even cause equipment damage or personal injury. Therefore, how to efficiently and accurately diagnose the faults of the engine lubricating oil system has become one of the key technologies to ensure the safe operation of the engine, extend the service life of the equipment and improve maintenance efficiency. In recent years, with the rapid development of data-driven methods and artificial intelligence technologies, intelligent algorithms based on machine learning and deep learning have been widely used in the field of engine fault diagnosis. Traditional data-driven fault diagnosis models mainly rely on feature extraction of historical engine fault data, and recognize different fault states by building classification models. However, this type of method has the following shortcomings: excessive reliance on measured data. The acquisition of measured engine fault data usually requires a large number of actual fault experiments, which are not only costly but also may cause irreversible damage to the equipment. In summary, although the traditional data-driven fault diagnosis method has improved the fault identification ability to a certain extent, its high dependence on measured data significantly limits the feasibility of practical application. It not only leads to high experimental costs, but also may cause damage to the equipment. There is an urgent need for an efficient and economical fault diagnosis method that can reduce the dependence on measured data.

[0003] Reference 1 "Research on Fault Diagnosis Method of Diesel Engine Injection System Based on 1DCNN-GWO-SVM" (Automotive Engine, 2024) proposed a diesel engine injection system fault diagnosis method combining one-dimensional convolutional neural network (1DCNN), gray wolf optimization algorithm (GWO) and support vector machine (SVM). This method uses 1DCNN to extract self-learning features of diesel engine vibration acceleration signals, and trains the support vector machine model based on the extracted feature vectors. At the same time, GWO is used to optimize the hyperparameters of SVM to achieve accurate diagnosis of diesel engine faults. Reference 2 "Intelligent Fault Diagnosis of Marine Diesel Engines Based on Deep Belief Network" (China Ship Research, 2020) proposed an engine intelligent fault diagnosis method based on multi-layer restricted Boltzmann machine. This method uses multi-layer restricted Boltzmann machine to realize feature extraction of simulated fault sample data based on AVLBOOST, achieving a high diagnostic accuracy. The above results have the following two limitations: (1) Although Reference 1 can achieve high accuracy by using actual fault signals collected in experiments to construct diagnostic models, the cost and difficulty of obtaining large-scale measured data are high, which limits its practical promotion and application. (2) The model in Reference 2 is trained and tested using simulation data generated by the AVL BOOST simulation platform. However, its adaptability and diagnostic capabilities to actual diesel engine data have not been fully verified, and it may face the problem of insufficient generalization in practical applications. Summary of the invention

[0004] In order to solve the problems existing in the prior art, the present invention provides a method and system for fault diagnosis of marine engine lubricating oil system based on transfer learning from simulation domain to real domain, which uses a large amount of twin simulation data and a small amount of actual data for model training, replacing the traditional training mode that relies entirely on actual data, and alleviating the problem of excessive dependence on actual data in model training. At the same time, a fault diagnosis method combining a densely connected neural network model and a comprehensive transfer learning strategy is proposed. Through the efficient feature extraction capability of the densely connected neural network and the domain alignment mechanism of the transfer learning strategy, transfer learning from the fault features of the simulation domain to the fault features of the real domain is realized, thereby solving the problems of high cost of diagnostic model construction and insufficient generalization performance.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for fault diagnosis of a marine engine lubricating oil system based on transfer learning from a simulation domain to a real domain, the method comprising:

[0007] Based on the constructed marine engine lubricating oil system simulation model, fault simulations under different working conditions are performed to obtain simulated fault signal responses;

[0008] Based on the engine test platform, obtain real domain fault data of engine lubricating oil system under different working conditions;

[0009] Based on the real domain fault data and the simulated fault response signal, a feature fusion structure based on an encoder-decoder is adopted to obtain simulated domain fault data;

[0010] Based on the transfer learning strategy, the cross-domain feature distribution alignment from the simulation domain to the real domain is performed, and the neural network is trained based on the simulation domain fault data and the real domain fault data to obtain the engine lubricating oil system fault diagnosis model based on dual transfer learning of data migration and algorithm migration;

[0011] Based on the engine lubricating oil fault diagnosis model, real-time fault diagnosis is performed on the lubricating oil system of the marine engine to obtain a diagnosis result.

[0012] Preferably, the marine engine lubricating oil system simulation model includes a pressure model and a temperature model; and the method for constructing the marine engine lubricating oil system simulation model includes:

[0013] Collecting engine lubricating oil system data; wherein the engine lubricating oil system data includes pipeline data, fluid density and lubricating oil pump characteristic curve;

[0014] The node-based modeling strategy is adopted to divide the engine oil system into five computing nodes: oil pan, oil pump, heat exchanger, filter, and engine body.

[0015] Based on the engine lubricating oil system data, the pressure model and the temperature model of each calculation node are calculated respectively, and all the calculation nodes are connected according to a logical relationship to obtain the marine engine lubricating oil system simulation model.

[0016] Preferably, based on the constructed marine engine lubricating oil system simulation model, the fault types simulated include five fault modes: lubricating oil filter blockage F1, lubricating oil cooler fouling F2, lubricating oil shortage F3, lubricating oil leakage F4 and bypass valve leakage F5.

[0017] Preferably, the simulation domain fault data includes the temperature after the lubricating oil pump, the temperature before the lubricating oil filter, the temperature after the lubricating oil filter, the temperature of the lubricating oil main oil channel, the pressure after the lubricating oil pump, the pressure before the lubricating oil filter, the pressure after the lubricating oil filter and the pressure of the lubricating oil main oil channel.

[0018] Preferably, the method for obtaining simulation domain fault data includes:

[0019] Performing signal decomposition on the real domain fault data based on an encoder to obtain high frequency noise;

[0020] Based on the high-frequency noise, a noise distribution characteristic is obtained; wherein the noise distribution characteristic includes a noise mean and a noise standard deviation;

[0021] Reconstructing the noise distribution characteristics based on the decoder, and fusing the reconstructed noise distribution characteristics with the simulated fault signal response to obtain twin simulation data;

[0022] Based on the twin simulation data, the simulation domain fault data is obtained.

[0023] Preferably, the method for aligning cross-domain feature distribution from the simulation domain to the real domain includes:

[0024] Using the simulation domain fault data and the real domain fault data to train a feature extractor and a label classifier to obtain a fault type prediction probability, and obtaining a comprehensive classification loss based on the fault type prediction probability;

[0025] Using the trained feature extractor to extract features from source domain samples and target domain samples, obtain feature representation, and calculate CORAL loss; wherein the source domain samples include the simulation domain fault data; the target domain samples include the real domain fault data;

[0026] Based on the discriminator, it is predicted whether the feature representation extracted by the feature extractor comes from the source domain sample, and the domain discrimination loss is obtained;

[0027] Based on the comprehensive classification loss, the CORAL loss and the domain discrimination loss, an optimization objective function is established;

[0028] Based on the optimization objective function, cross-domain feature distribution alignment from the simulation domain to the real domain is achieved.

[0029] The present invention also provides a marine engine lubricating oil system fault diagnosis system based on transfer learning from simulation domain to real domain, which is used to implement the method, and the system comprises:

[0030] The fault simulation module is used to simulate faults under different working conditions based on the constructed marine engine lubricating oil system simulation model and obtain simulated fault signal responses;

[0031] A real domain data acquisition module is used to obtain real domain fault data of different working conditions of the engine lubricating oil system based on the engine test platform;

[0032] A simulation domain data acquisition module, used to obtain simulation domain fault data based on the real domain fault data and the simulation fault response signal by adopting a feature fusion structure based on an encoder-decoder;

[0033] A fault diagnosis model building module is used to perform cross-domain feature distribution alignment from the simulation domain to the real domain based on a transfer learning strategy, and train a neural network based on the simulation domain fault data and the real domain fault data to obtain an engine lubricating oil system fault diagnosis model based on dual transfer learning of data migration and algorithm migration;

[0034] The diagnosis result acquisition module is used to perform real-time fault diagnosis on the marine engine lubricating oil system based on the engine lubricating oil fault diagnosis model to obtain the diagnosis result.

[0035] Preferably, the fault simulation module includes a simulation model building unit for building a simulation model of the marine engine lubricating oil system; wherein the marine engine lubricating oil system simulation model includes a pressure model and a temperature model; and the process of building the marine engine lubricating oil system simulation model includes:

[0036] Collecting engine lubricating oil system data; wherein the engine lubricating oil system data includes pipeline data, fluid density and lubricating oil pump characteristic curve;

[0037] The node-based modeling strategy is adopted to divide the engine oil system into five computing nodes: oil pan, oil pump, heat exchanger, filter, and engine body.

[0038] Based on the engine lubricating oil system data, the pressure model and the temperature model of each calculation node are calculated respectively, and all the calculation nodes are connected according to a logical relationship to obtain the marine engine lubricating oil system simulation model.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] 1. Through the encoder-decoder structure, the feature distribution of simulation data and actual data is deeply integrated to generate twin simulation data with smaller information distribution differences, providing reliable and comprehensive feature representation for subsequent cross-domain fault diagnosis.

[0041] 2. A densely connected neural network model is used to efficiently extract features from the input data, and combined with a transfer learning strategy, the domain transmission error between the simulation domain and the real domain is reduced, achieving accurate cross-domain fault diagnosis of the engine from the simulation domain to the real domain.

[0042] 3. By generating high-quality twin simulation data, the diagnostic model’s excessive dependence on actual measurement data is reduced.

[0043] 4. The present invention has the ability to diagnose faults across load conditions, from load conditions (M1, M2, M3) to load conditions (M4, M5, M6), wherein M1, M2, M3 are not limited to the 70%, 85%, 100% loads adopted by the present invention, and M4, M5, M6 are not limited to the 25%, 50% and 75% loads adopted by the present invention, that is, the ability of the fault diagnosis model of the present invention to diagnose faults across load conditions is not limited to from high load to general load. The present invention further expands the scope of application of the model and improves its practicality and robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in 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 creative labor.

[0045] Figure 1 This is a flow chart of a method for diagnosing faults in a marine engine lubricating oil system based on transfer learning from a simulation domain to a real domain according to an embodiment of the present invention;

[0046] Figure 2 This is a flow chart of twin simulation data acquisition in an embodiment of the present invention;

[0047] Figure 3 This is a comparison diagram of data migration effects in an embodiment of the present invention;

[0048] Figure 4 A diagram of a deep learning model used in the algorithm migration of an embodiment of the present invention;

[0049] Figure 5 This is a comparison chart of the diagnosis results of the embodiment of the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0052] Embodiment 1

[0053] like Figure 1 As shown, a method for fault diagnosis of a marine engine lubricating oil system based on transfer learning from a simulation domain to a real domain comprises:

[0054] S1: Based on the constructed marine engine lubricating oil system simulation model, fault simulations under different working conditions are performed to obtain the simulated fault signal response S r ; A further implementation method is that the marine engine lubricating oil system simulation model includes a pressure model and a temperature model; specifically, the pressure model is calculated by the following formula:

[0055]

[0056] Where z1 and z2 are the inlet and outlet heights of the pipeline, p1 and p2 are the inlet and outlet pressures of the pipeline, v1 and v2 are the inlet and outlet flow rates of the pipeline, α1 and α2 are the kinetic energy correction coefficients, 2 for laminar flow calculation, 1 for turbulent flow calculation, and h v It is the total energy loss caused by the unit gravity liquid flowing from pipe section 1 to pipe section 2.

[0057] The temperature model is as follows:

[0058] Q t =KAΔ m ,

[0059] In the formula, Q t is the total heat transfer, K is the total heat transfer coefficient of the heat exchanger, A is the total heat transfer area, Δ m is the average temperature difference of the heat transfer medium.

[0060] The method of constructing a simulation model of a marine engine lubricating oil system includes:

[0061] The engine lubricating oil system data is collected; the engine lubricating oil system data includes pipeline data, fluid density and lubricating oil pump characteristic curve; in this embodiment, the engine model referenced in the lubricating oil system modeling process is the 16V396TE engine. The operating conditions include three high-load conditions of 70%, 85% and 100%.

[0062] The node-based modeling strategy is adopted to divide the engine oil system into five computing nodes: oil pan, oil pump, heat exchanger, filter, and engine body.

[0063] Based on the engine lubricating oil system data, the pressure model and temperature model of each calculation node are calculated respectively, and all the calculation nodes are connected according to the logical relationship to obtain the simulation model of the marine engine lubricating oil system.

[0064] Specifically, the parameters calculated by the engine lubricating oil system temperature model and the lubricating oil system pressure model include eight parameters: temperature after the lubricating oil pump, temperature before the lubricating oil filter, temperature after the lubricating oil filter, temperature of the lubricating oil main oil channel, pressure after the lubricating oil pump, pressure before the lubricating oil filter, pressure after the lubricating oil filter and pressure of the lubricating oil main oil channel.

[0065] A further implementation method is that, based on the constructed marine engine lubricating oil system simulation model, the fault types for fault simulation include five fault modes: lubricating oil filter blockage F1, lubricating oil cooler fouling F2, lubricating oil shortage F3, lubricating oil leakage F4 and bypass valve leakage F5. In this embodiment, the fault modes are not limited to the above five modes, and only the above five modes are used as cases.

[0066] Combined with the calibration parameters provided in the engine bench test manual, the simulation model is initially parameterized and logically verified, and the key parameters in the simulation model (such as conductance coefficient, heat transfer efficiency coefficient, etc.) are gradually adjusted to make the output of the simulation model more consistent with the bench test data.

[0067] The fault injection is implemented as follows: the lubricating oil filter is blocked F1 fault injected, the boundary pressure coefficient is set to 1.4-1.6 to simulate the reduction of the inlet oil pressure, the boundary heat coefficient is set to 1.1-1.4 to simulate the overall temperature increase of the lubricating oil, and the lubricating oil filter conductance coefficient is set to 0.6-0.9 to simulate the blockage of the lubricating oil filter structure.

[0068] The oil cooler fouling F2 fault injection sets the boundary pressure coefficient to 1.0-1.1 to simulate the pressure caused by the temperature increase, sets the boundary heat coefficient to 1.5-2.0 to simulate the increase in the overall temperature of the lubricating oil, and sets the oil cooler flow coefficient to 1.0-1.1 to simulate the fouling of the cooler structure.

[0069] The lubricating oil shortage F3 fault injection sets the boundary heat coefficient to 1.0-1.1 to simulate the increase of lubricating oil temperature, sets the boundary flow coefficient to 0.98-1.0, and sets the boundary pressure coefficient to 0.9-1.0 to simulate the decrease of boundary pressure.

[0070] The lubricating oil leakage F4 fault injection sets the boundary heat coefficient to 1.1-1.2 to simulate the increase of lubricating oil temperature, sets the boundary flow coefficient to 0.98-1.0 to simulate the decrease of lubricating oil flow, sets the boundary pressure coefficient to 0.7-0.9 to simulate the decrease of boundary pressure, and sets the lubricating oil inlet pipe flow coefficient to 0.9-1.0 to simulate the lubricating oil leakage at the inlet pipe.

[0071] The bypass valve leakage F5 fault injection sets the boundary flow coefficient to 0.98-1.0 to simulate the reduction of lubricating oil flow, sets the boundary pressure coefficient to 0.4-0.6 to simulate the decrease of boundary pressure, sets the boundary heat coefficient to 1.1-1.3 to simulate the increase of overall temperature, sets the lubricating oil filter flow coefficient to 0.5-0.6, and sets the lubricating oil inlet pipe flow coefficient to 0.6-0.7 to simulate the leakage of the bypass valve structure.

[0072] S2: Based on the engine test platform, obtain the real domain fault data of the engine lubricating oil system under different working conditions; in this embodiment, the engine test platform model is 8V396TE engine, using WZP-2080 / 139 temperature sensor and GYPM10 pressure transmitter to obtain the actual fault data of the engine lubricating oil system.

[0073] The actual fault data parameters of the engine test platform are obtained, including 8 parameters: temperature after the lubricating oil pump, temperature before the lubricating oil filter, temperature after the lubricating oil filter, temperature of the lubricating oil main oil channel, pressure after the lubricating oil pump, pressure before the lubricating oil filter, pressure after the lubricating oil filter and pressure of the lubricating oil main oil channel. Specifically, the engine lubricating oil system data is collected at 25%, 50% and 75% load conditions respectively.

[0074] S3: Based on the real domain fault data and the simulated fault response signal, a feature fusion structure based on an encoder-decoder is used to obtain the simulated domain fault data; a further implementation method is that the simulated domain fault data includes the after-lubricating oil pump temperature, the before-lubricating oil filter temperature, the after-lubricating oil filter temperature, the lubricating oil main oil channel temperature, the after-lubricating oil pump pressure, the before-lubricating oil filter pressure, the after-lubricating oil filter pressure and the lubricating oil main oil channel pressure.

[0075] like Figure 2 As shown, a further implementation method is that the method for obtaining simulation domain fault data includes:

[0076] Based on the encoder, the real domain fault data is decomposed to obtain high-frequency noise and low-frequency noise;

[0077] Based on high-frequency noise, noise distribution characteristics are obtained; wherein, the noise distribution characteristics include noise mean and noise standard deviation; specifically, the actual measurement data (real domain fault data) obtained is input into the encoder for signal decomposition, and the noise distribution parameter mean μ of the actual measurement data is obtained. r and variance σ r 2 , the distribution can be expressed as follows:

[0078]

[0079] The noise distribution characteristics are reconstructed based on the decoder, and the reconstructed noise distribution characteristics are compared with the simulated fault signal response S r Fusion, to obtain twin simulation data; specifically, in the decoder, the noise data is reconstructed according to the noise distribution parameters, the noise data N re Can be rebuilt as:

[0080]

[0081] Finally, the simulation signal S r With the reconstruction noise N re Fusion generates twin data D f :

[0082] D f =S r +N re .

[0083] Based on the twin simulation data, the simulation domain fault data is obtained.

[0084] S4: Based on the transfer learning strategy, the cross-domain feature distribution alignment from the simulation domain to the real domain is performed, and the neural network is trained based on the simulation domain fault data and the real domain fault data to obtain the engine lubricating oil system fault diagnosis model based on dual transfer learning of data migration and algorithm migration. Each dense block contains 3 convolutional layers, and each convolutional layer shares feature knowledge with all previous convolutional layers. This design makes the input of each layer include not only the output of the previous layer, but also the output of all previous layers, which improves the utilization of data features. The transition block is located between the two dense blocks. Its main purpose is to reduce the size of the features and the number of channels to reduce the computational cost. Figure 3 , Figure 4 shown.

[0085] In this embodiment, the neural network used is a densely connected neural network, and the architecture is composed of two dense blocks and a transition block stacked alternately.

[0086] The specific densely connected neural network model parameters are shown in Table 1.

[0087] Table 1

[0088]

[0089]

[0090] The transfer learning strategy is that the neural network model achieves effective alignment of cross-domain feature distribution through comprehensive classification loss, CORAL loss and adversarial learning strategy of domain discriminator.

[0091] A further implementation method is that the method for aligning cross-domain feature distribution from the simulation domain to the real domain includes:

[0092] The feature extractor and label classifier are trained using the simulated domain fault data and the real domain fault data to obtain the fault type prediction probability, and the comprehensive classification loss is obtained based on the fault type prediction probability; the feature extractor includes the initial layer, dense block-1, transition block and dense block-2. The label classifier includes an adaptive average pooling layer and a linear layer.

[0093] Specifically, the classification loss is used to train the feature extractor and label classifier, using the cross entropy loss, and the calculation formula is as follows:

[0094]

[0095] Where N is the number of data samples and K is the total number of fault categories. The true category label of sample i. The classifier's predicted probability for the class of sample i.

[0096] The trained feature extractor is used to extract features from source domain samples and target domain samples, obtain feature representation, and calculate CORAL loss; the source domain samples include simulation domain fault data; the target domain samples include real domain fault data; specifically, CORAL loss is used to minimize the difference between the source domain and target domain feature distributions, which is defined as:

[0097]

[0098] Among them, H s H represents the output of the source domain sample after the feature extractor F. t It is the output of the target domain sample after the feature extractor F. is the Frobenius norm squared of the matrix. d is the dimension size of the feature. Σ s Represents the covariance matrix of the source domain data, Σ t Represents the covariance matrix of the target domain data.

[0099] Based on the discriminator, the feature representation extracted by the feature extractor is predicted to be from the source domain sample, and the domain discrimination loss is obtained; specifically, the domain discrimination loss extracts domain-invariant features through domain adversarial training. The discriminator D predicts whether the sample comes from the source domain, and the feature extractor F tries to generate domain confusion features. The binary cross entropy loss is defined as:

[0100]

[0101] Where N = N s +N t , d i represents the domain label of sample i, It represents the predicted probability of the discriminator D that sample i belongs to the source domain.

[0102] The domain-invariant features mentioned above refer to features that are consistently distributed in the source domain and the target domain. By extracting domain-invariant features, the model can be trained on the source domain and generalize well to the target domain. Domain confusion features are features generated by the feature extractor during domain adversarial training. Their main function is to make it difficult for the discriminator D to distinguish whether the sample comes from the source domain or the target domain, and to make the feature distribution of the source domain and the target domain overlap as much as possible.

[0103] The common goal of both is to reduce the distribution difference between the source domain and the target domain. The process of generating domain confusion features is actually a means to achieve domain invariant feature extraction.

[0104] Based on the comprehensive classification loss, CORAL loss and domain discrimination loss, an optimization objective function is established; specifically, the overall optimization goal is the weighted sum of the comprehensive classification loss, CORAL loss and domain discrimination loss, defined as:

[0105] L total =L cls +λ coral L coral -λ adv L adv ,

[0106] Among them, λ coral is the weight coefficient of CORAL loss. adv is the weight coefficient of the domain discrimination loss, and the negative sign indicates the adversarial target.

[0107] Based on the optimization objective function, cross-domain feature distribution alignment from the simulation domain to the real domain is achieved.

[0108] In this embodiment, the twin simulation data used to train the model includes five fault modes, each fault mode includes 36,000 data items, and the actual data used to train the model also includes five fault modes, each fault mode includes 18,000 data items.

[0109] S5: Based on the engine lubricating oil fault diagnosis model, perform real-time fault diagnosis on the marine engine lubricating oil system and obtain the diagnosis results. Figure 5 shown.

[0110] In this embodiment, the real-time fault diagnosis obtains the fault diagnosis result by sampling the parameters of the engine oil system running under any actual working condition in real time and inputting the parameters into the fault diagnosis model.

[0111] Specifically, it involves 9 transfer learning tasks in the engine lubricating oil system dataset, marked as Task1 to Task9. Taking Task 1 as an example, Task 1 uses the data under 70% load of the twin simulation data (simulation domain fault data) as the source domain data, and uses the data under 25% load of the actual data (real domain fault data) as the target domain data for testing.

[0112] In a specific embodiment, Task 1: 70% → 25% migration task; Task 2: 70% → 50% migration task; Task 3: 70% → 75% migration task; Task 4: 85% → 25% migration task; Task 5: 85% → 50% migration task; Task 6: 85% → 75% migration task; Task 7: 100% → 25% migration task; Task 8: 100% → 50% migration task; Task 9: 100% → 75% migration task.

[0113] The engine cross-domain fault diagnosis model based on dual transfer learning of data migration and algorithm migration constructed by the present invention has good accuracy for engine fault diagnosis and performs well in 9 migration tasks. Figure 5 As shown in the figure, in the 9 migration tasks, the diagnostic accuracy is guaranteed to be above 89.68%, and the average accuracy can reach 97.70%, which has good generalization.

[0114] The present invention has the ability to diagnose faults across load conditions, from load conditions (M1, M2, M3) to load conditions (M4, M5, M6), wherein M1, M2, M3 are not limited to the 70%, 85%, 100% loads adopted by the present invention, and M4, M5, M6 are not limited to the 25%, 50% and 75% loads adopted by the present invention, that is, the ability of the fault diagnosis model of the present invention to diagnose faults across load conditions is not limited to from high load to general load. The present invention further expands the scope of application of the model and improves its practicality and robustness.

[0115] Embodiment 2

[0116] The present invention also provides a marine engine lubricating oil system fault diagnosis system based on transfer learning from simulation domain to real domain, which is used to implement the method, and the system includes:

[0117] The fault simulation module is used to simulate faults under different working conditions based on the constructed marine engine lubricating oil system simulation model and obtain simulated fault signal responses;

[0118] A real domain data acquisition module is used to obtain real domain fault data of different working conditions of the engine lubricating oil system based on the engine test platform;

[0119] A simulation domain data acquisition module is used to obtain simulation domain fault data based on real domain fault data and simulated fault response signals by adopting a feature fusion structure based on an encoder-decoder;

[0120] The fault diagnosis model building module is used to align the cross-domain feature distribution from the simulation domain to the real domain based on the transfer learning strategy, and train the neural network based on the fault data in the simulation domain and the fault data in the real domain to obtain the engine lubricating oil system fault diagnosis model based on the dual transfer learning of data transfer and algorithm transfer;

[0121] The diagnosis result acquisition module is used to perform real-time fault diagnosis on the lubricating oil system of a marine engine based on the engine lubricating oil fault diagnosis model to obtain the diagnosis result.

[0122] A further implementation method is that the fault simulation module includes a simulation model building unit for building a simulation model of a marine engine lubricating oil system; wherein the simulation model of the marine engine lubricating oil system includes a pressure model and a temperature model; and the process of building the simulation model of the marine engine lubricating oil system includes:

[0123] Collecting engine lubricating oil system data; wherein the engine lubricating oil system data includes pipeline data, fluid density and lubricating oil pump characteristic curve;

[0124] The node-based modeling strategy is adopted to divide the engine oil system into five computing nodes: oil pan, oil pump, heat exchanger, filter, and engine body.

[0125] Based on the engine lubricating oil system data, the pressure model and temperature model of each calculation node are calculated respectively, and all the calculation nodes are connected according to the logical relationship to obtain the simulation model of the marine engine lubricating oil system.

[0126] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for fault diagnosis of a marine engine lubricating oil system based on transfer learning from simulation domain to real domain, characterized in that: The method comprises: Based on the constructed marine engine lubricating oil system simulation model, fault simulations under different working conditions are performed to obtain simulated fault signal responses; Based on the engine test platform, obtain real domain fault data of engine lubricating oil system under different working conditions; Based on the real domain fault data and the simulated fault response signal, a feature fusion structure based on an encoder-decoder is adopted to obtain simulated domain fault data; Based on the transfer learning strategy, the cross-domain feature distribution alignment from the simulation domain to the real domain is performed, and the neural network is trained based on the simulation domain fault data and the real domain fault data to obtain the engine lubricating oil system fault diagnosis model based on dual transfer learning of data migration and algorithm migration; Based on the engine lubricating oil fault diagnosis model, real-time fault diagnosis is performed on the lubricating oil system of the marine engine to obtain a diagnosis result.

2. The method according to claim 1, characterized in that The marine engine lubricating oil system simulation model includes a pressure model and a temperature model; and the method for constructing the marine engine lubricating oil system simulation model includes: Collecting engine lubricating oil system data; wherein the engine lubricating oil system data includes pipeline data, fluid density and lubricating oil pump characteristic curve; The node-based modeling strategy is adopted to divide the engine oil system into five computing nodes: oil pan, oil pump, heat exchanger, filter, and engine body. Based on the engine lubricating oil system data, the pressure model and the temperature model of each calculation node are calculated respectively, and all the calculation nodes are connected according to a logical relationship to obtain the marine engine lubricating oil system simulation model.

3. The method according to claim 1, characterized in that Based on the constructed marine engine lubricating oil system simulation model, the fault types simulated include five fault modes: lubricating oil filter blockage F1, lubricating oil cooler fouling F2, lubricating oil shortage F3, lubricating oil leakage F4 and bypass valve leakage F5.

4. The method according to claim 1, characterized in that The simulation domain fault data includes the temperature after the lubricating oil pump, the temperature before the lubricating oil filter, the temperature after the lubricating oil filter, the temperature of the lubricating oil main oil channel, the pressure after the lubricating oil pump, the pressure before the lubricating oil filter, the pressure after the lubricating oil filter and the pressure of the lubricating oil main oil channel.

5. The method according to claim 1, characterized in that: Methods for obtaining simulation domain fault data include: Performing signal decomposition on the real domain fault data based on an encoder to obtain high frequency noise; Based on the high-frequency noise, a noise distribution characteristic is obtained; wherein the noise distribution characteristic includes a noise mean and a noise standard deviation; Reconstructing the noise distribution characteristics based on the decoder, and fusing the reconstructed noise distribution characteristics with the simulated fault signal response to obtain twin simulation data; Based on the twin simulation data, the simulation domain fault data is obtained.

6. The method according to claim 1, characterized in that Methods for aligning cross-domain feature distribution from the simulation domain to the real domain include: Using the simulation domain fault data and the real domain fault data to train a feature extractor and a label classifier to obtain a fault type prediction probability, and obtaining a comprehensive classification loss based on the fault type prediction probability; Using the trained feature extractor to extract features from source domain samples and target domain samples, obtain feature representation, and calculate CORAL loss; wherein the source domain samples include the simulation domain fault data; the target domain samples include the real domain fault data; Based on the discriminator, it is predicted whether the feature representation extracted by the feature extractor comes from the source domain sample, and the domain discrimination loss is obtained; Based on the comprehensive classification loss, the CORAL loss and the domain discrimination loss, an optimization objective function is established; Based on the optimization objective function, cross-domain feature distribution alignment from the simulation domain to the real domain is achieved.

7. A marine engine lubricating oil system fault diagnosis system based on transfer learning from simulation domain to real domain, used to implement the method described in any one of claims 1 to 6, characterized in that: The system comprises: The fault simulation module is used to simulate faults under different working conditions based on the constructed marine engine lubricating oil system simulation model and obtain simulated fault signal responses; A real domain data acquisition module is used to obtain real domain fault data of different working conditions of the engine lubricating oil system based on the engine test platform; A simulation domain data acquisition module, used to obtain simulation domain fault data based on the real domain fault data and the simulation fault response signal by adopting a feature fusion structure based on an encoder-decoder; A fault diagnosis model building module is used to perform cross-domain feature distribution alignment from the simulation domain to the real domain based on a transfer learning strategy, and train a neural network based on the simulation domain fault data and the real domain fault data to obtain an engine lubricating oil system fault diagnosis model based on dual transfer learning of data migration and algorithm migration; The diagnosis result acquisition module is used to perform real-time fault diagnosis on the marine engine lubricating oil system based on the engine lubricating oil fault diagnosis model to obtain the diagnosis result.

8. The system according to claim 7, characterized in that The fault simulation module includes a simulation model building unit, which is used to build the simulation model of the marine engine lubricating oil system; wherein the simulation model of the marine engine lubricating oil system includes a pressure model and a temperature model; the process of building the simulation model of the marine engine lubricating oil system includes: Collecting engine lubricating oil system data; wherein the engine lubricating oil system data includes pipeline data, fluid density and lubricating oil pump characteristic curve; The node-based modeling strategy is adopted to divide the engine oil system into five computing nodes: oil pan, oil pump, heat exchanger, filter, and engine body. Based on the engine lubricating oil system data, the pressure model and the temperature model of each calculation node are calculated respectively, and all the calculation nodes are connected according to a logical relationship to obtain the marine engine lubricating oil system simulation model.

Images