Ship power system fault monitoring method

Abstract

The invention relates to a ship power system fault monitoring method, which comprises the following steps of: acquiring a relevant parameter # imgabs1 # of a nearest # imgabs0 # ship power system, preprocessing the relevant parameter, constructing a fault monitoring model based on an LSTM (Long Short Term Memory) model, adjusting a learning rate # imgabs2 # of the fault monitoring model based on Bayesian optimization, and acquiring optimal parameter configuration of a learning rate # imgabs3 # of the fault monitoring model. The optimized fault monitoring model is used for outputting a fault prediction result, screening key parameters by comprehensively adopting ship power system parameters and utilizing Pearson's correlation coefficients, extracting feature parameters through PCA dimensionality reduction, constructing the fault monitoring model in combination with an LSTM model, and meanwhile, optimizing a learning rate by adopting Bayesian, so that the fault prediction result is obtained. According to the method, the accuracy and efficiency of fault prediction are remarkably improved, the calculation complexity is reduced, the generalization ability of a fault monitoring model is enhanced, the downtime and maintenance cost caused by faults are reduced, and the safety and reliability of ship operation are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of ship power systems, and in particular to a ship power system fault monitoring method. Background Art

[0002] The ship's power system plays an extremely important role on the ship. It is an independent and movable power system. The reliability, continuity and power supply quality of the system directly affect the economic indicators, technical indicators and vitality of the ship. Due to the complex structure of the ship and the scattered distribution of electrical equipment, the failure of the lines and equipment of various new ships constantly challenges the increasingly "aging" knowledge system of ship operators. When faults occur in parallel, it is difficult to quickly find the fault and eliminate it.

[0003] After searching, a method for power system equipment status monitoring and fault diagnosis was proposed according to the Chinese invention application number CN202410743775.1. The invention content recorded that "the data analysis and processing method extracts feature vectors reflecting the status of power system equipment from the power equipment operation data through a feature extraction module, a feature vector normalization module and a feature vector dimension reduction module, realizes the normalization and dimension reduction processing of the feature vector, and improves the accuracy of subsequent power system equipment status monitoring. The decision tree model training algorithm trains the feature vector and the power equipment operation data to generate a status monitoring model. The status monitoring model realizes abnormal and normal status monitoring of power system equipment through a model classification and discrimination method."

[0004] Although the power system equipment status monitoring and fault diagnosis method can predict whether the power system equipment status is abnormal or normal according to the power equipment operation data, and the intelligent diagnosis system diagnoses the abnormal status of the power equipment through the fault location module, the fault analysis module and the fault elimination module, and realizes the fault location, analysis and elimination of the abnormal status of the power equipment, the parameters affecting the ship's power system include not only voltage signals, current signals and vibration frequency signals, but also the correlation between the optimization model method and the importance of the parameters is weak, which affects the monitoring effect of the ship's power system faults and needs to be improved. Therefore, a power system equipment status monitoring and fault diagnosis method is proposed to solve the above problems. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a method for monitoring faults in a ship power system, which has the advantages of good monitoring effect, etc., and solves the problem that the parameters affecting the ship power system include not only voltage signals, current signals and vibration frequency signals, but also the optimization model method is weakly correlated with the importance of the parameters, thus affecting the monitoring effect of the ship power system faults.

[0006] To achieve the above object, the present invention provides the following technical solution: a method for monitoring a fault in a ship power system, comprising the following specific steps: S1: Get the latest Related parameter matrix of ship power system during the ship's voyage , , , create a set of related parameter matrices , and preprocess the relevant parameter matrix to obtain the preprocessed relevant parameter matrix set ; The S1 comprises the following sub-steps: S1.1: Get the latest Related parameter matrix of secondary ship power system , , , create a set of related parameter matrices ; S1.2: Based on 3 Principle to remove relevant parameter matrix Outliers in ; S1.3: Normalization method for the correlation parameter matrix Perform normalization to obtain the relevant parameter matrix set after preprocessing ; The relevant parameter matrix Including voltage fluctuation characteristic vector, frequency change characteristic vector, current mutation characteristic vector, harmonic content characteristic vector, three-phase imbalance characteristic vector; S2: Calculate the correlation parameter matrix based on the Pearson correlation coefficient The correlation between them is used to select the key parameter matrix Based on PCA dimensionality reduction, the key parameter matrix after dimensionality reduction is As the characteristic parameter matrix , create a feature parameter matrix set ; The S2 comprises the following sub-steps: S2.1: Calculate the correlation parameter matrix based on the Pearson correlation coefficient The correlation between them is calculated as: ; In the formula, and Two different related parameters The i-th value of and Two different related parameters The mean of For related parameters and related parameters The correlation coefficient between i is the eth first related parameter, f i is the fth second related parameter, is the mean of the first correlation parameter, is the mean of the second correlation parameter, ; S2.2: Setting relevance thresholds , using the judgment formula , remove the correlation parameter matrix with correlation below the correlation threshold , retain the correlation parameter matrix with correlations above the correlation threshold As a key parameter matrix ; S2.3: Use PCA to transform the key parameter matrix The high-dimensional feature vector in is converted into a low-dimensional feature vector to obtain the feature parameter matrix , create a feature parameter matrix set ; S3: Build a fault monitoring model FMM based on the LSTM model, and transform the feature parameter matrix into The fault monitoring model FMM is trained as a training set, and the fault monitoring model FMM is optimized based on the cross entropy loss function to obtain the fault monitoring model FMM'. Finally, the accuracy, precision, recall and F1 score are used to evaluate the performance of the fault monitoring model FMM'. The S3 comprises the following sub-steps: S3.1: Construct a fault monitoring model FMM based on the LSTM algorithm, including a forget gate, an input gate, and an output gate. The forget gate determines which information is discarded from the previous cell state. The input gate determines which new information will be stored in the cell state. The cell state is updated by combining the results of the forget gate and the input gate to update the cell state. The output gate is used to decide which information is output from the cell state. S3.2: input the training set into the fault monitoring model FMM for training; S3.3: Based on the cross entropy loss function, the fault monitoring model FMM is optimized to obtain the fault monitoring model FMM'. The calculation expression is: ; In the formula, y b Indicates the bth characteristic parameter ,l(y b =d) takes a value of 0 or 1. When it takes a value of 1, it means that when y b Correctly classified as type d fault type, when the value is 0, it means y b Not correctly classified as a Class D fault type, is the number of training samples, C c Indicates the type of fault involved, Indicates the bth characteristic parameter The probability of being correctly classified as a type d fault, is the cross entropy loss; S3.4: Use accuracy, precision, recall and F1 score to evaluate the performance of the fault monitoring model FMM'; S4: Learning rate of fault monitoring model FMM' based on Bayesian optimization Adjust to obtain the learning rate of the fault monitoring model FMM' The optimal parameter configuration is used to obtain the optimized fault monitoring model FMM'', and the optimized fault monitoring model FMM'' is used to reconstruct and predict the ship power related signals in the ship power system, so as to realize the ship power system fault identification and the backtracking and analysis of the fault process; The S4 comprises the following specific steps: S4.1: Characteristic parameter matrix The value range is , learning rate The value range is , the characteristic parameters and learning rate All are set to v levels, and the calculation expression is: ; ; ; ; In the formula, is the characteristic parameter matrix The interval between each gear, For the The boundary of the file, is the learning rate The interval between each gear, For the The boundaries of the files; S4.2: Create a two-dimensional parameter space, where each row represents a feature parameter matrix level and each column represents a learning rate level, and initialize a database to store the sampling points and their corresponding objective function values ​​during the Bayesian optimization process; S4.3: Randomly select a set of initial sampling points from the two-dimensional parameter space, and use the initial sampling points to train an initial proxy model to approximate the performance of the fault monitoring model FMM'; S4.4: Calculate the expected improvement value of each combination of feature parameter matrix gear and learning rate gear; S4.5: Select the sampling point with the highest expected improvement value as the next evaluation point; S4.6: Train the fault monitoring model on the selected sampling points and evaluate its true objective function value; S4.7: Add the newly obtained sampling points and their corresponding objective function values ​​to the existing sampling point set; S4.8: Determine whether the Bayesian optimization has reached the preset number of iterations. If yes, proceed to S4.9; if no, proceed to S4.4; S4.9: Stop the optimization process and retrain the proxy model using the updated set of sampling points to improve its approximation accuracy; S4.10: The parameters in the trained proxy model are used as the optimal parameters of the optimized fault monitoring model FMM''; S4.11: Use the optimized fault monitoring model FMM'' to reconstruct and predict the ship power related signals in the ship power system, and compare them with the real-time collected ship power system operation data to realize the ship power system fault identification; S5: Establishing the characteristic parameter matrix The specific location of the fault L f The index relationship between them, that is, when the fault monitoring model FMM'' detects a fault in the ship's power system, the specific location of the fault in the ship's power system is traced in real time; The S5 comprises the following specific steps: Get recent The specific location where the fault occurred during the voyage of the ship L f , create a set of related parameter matrices The specific location of the fault L f The index relationship between , and then establish the relevant parameter matrix And the related parameter matrix The transformation relationship between , and then establish the relevant parameter matrix With key parameter matrix The transformation relationship between , and then establish the key parameter matrix With the characteristic parameter matrix The dimensionality reduction relationship between , and finally establish the characteristic parameter matrix The specific location of the fault L f Index relationship between That is, when the fault monitoring model FMM'' detects a fault in the ship's power system, the specific location of the fault in the ship's power system is traced in real time.

[0007] Compared with the prior art, the technical solution of this application has the following beneficial effects: The ship power system fault monitoring method adopts the ship power system parameters comprehensively, uses the Pearson correlation coefficient to screen key parameters, extracts feature parameters through PCA dimensionality reduction, builds a fault monitoring model in combination with the LSTM algorithm, and adopts the Bayesian optimization learning rate. It significantly improves the accuracy and efficiency of fault prediction, reduces the computational complexity, and enhances the generalization ability of the fault monitoring model, which helps to reduce the downtime and maintenance costs caused by faults and improve the safety and reliability of ship operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 It is the overall flow chart of the present invention; Figure 2 This is a flow chart of the learning rate adjustment of the present invention. DETAILED DESCRIPTION

[0009] 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.

[0010] See also Figure 1-2 In this embodiment, a method for monitoring a fault in a ship power system includes the following specific steps: S1: Get the latest Related parameter matrix of ship power system during the ship's voyage , , , create a set of related parameter matrices , and preprocess the relevant parameter matrix to obtain the preprocessed relevant parameter matrix set ; The S1 comprises the following sub-steps: S1.1: Get the latest Related parameter matrix of secondary ship power system , , , create a set of related parameter matrices ; S1.2: Based on 3 Principle to remove relevant parameter matrix Outliers in ; S1.3: Normalization method for the correlation parameter matrix Perform normalization to obtain the relevant parameter matrix set after preprocessing ; The relevant parameter matrix Including voltage fluctuation characteristic vector, frequency change characteristic vector, current mutation characteristic vector, harmonic content characteristic vector, three-phase imbalance characteristic vector; The acquisition of the voltage fluctuation characteristic vector, the frequency change characteristic vector, and the current mutation characteristic vector is common knowledge for those skilled in the art, so the acquisition process is not described in detail. The acquisition process of the harmonic content characteristic vector is as follows: after discrete sampling of the signal, the fast Fourier transform (FFT) algorithm is applied to convert the time domain signal into the frequency domain to obtain a spectrum diagram, and the amplitude and phase information of each harmonic are extracted from the spectrum diagram to form a harmonic content characteristic vector; and the acquisition process of the three-phase imbalance characteristic vector is as follows: in each of the three-phase power system of the ship, Install current or voltage transformers on each phase to measure the current or voltage value of each phase in real time. Calculate the maximum and minimum values ​​of the three-phase current or voltage based on the measured data. Use the three-phase imbalance calculation formula, such as (maximum phase load - minimum phase load) / maximum phase load × 100% or MAX (phase current - three-phase average current) / three-phase average current × 100% to calculate the three-phase imbalance. Use the calculated three-phase imbalance as a component of the feature vector, and then use the current or voltage value, average value, standard deviation and other statistics of each phase as other components of the feature vector. S2: Calculate the correlation parameter matrix based on the Pearson correlation coefficient The correlation between them is used to select the key parameter matrix Based on PCA dimensionality reduction, the key parameter matrix after dimensionality reduction is As the characteristic parameter matrix , create a feature parameter matrix set ; The S2 comprises the following sub-steps: S2.1: Calculate the correlation parameter matrix based on the Pearson correlation coefficient The correlation between them is calculated as: ; In the formula, and Two different related parameters The i-th value of and Two different related parameters The mean of For related parameters and related parameters The correlation coefficient between i is the eth first related parameter, f i is the fth second related parameter, is the mean of the first correlation parameter, is the mean of the second correlation parameter, ; S2.2: Setting relevance thresholds , using the judgment formula , remove the correlation parameter matrix with correlation below the correlation threshold , retain the correlation parameter matrix with correlations above the correlation threshold As a key parameter matrix ; S2.3: Use PCA to transform the key parameter matrix The high-dimensional feature vector in is converted into a low-dimensional feature vector to obtain the feature parameter matrix , create a feature parameter matrix set ; The specific steps of S2.3 are as follows: S2.3.1 Key parameter matrix Standardize and get the standardized key parameter matrix , the calculation expression is: ; ; ; In the formula, is the number of key parameters, is the i-th value of the key parameter, is the mean value of the key parameter, is the key parameter and mean The square root of the mean of the squares of the differences is the standard deviation. is the standardized key parameter matrix; S2.3.2: Calculate the covariance matrix , the calculation expression is: ; In the formula, for The transposed matrix of is the covariance matrix; S2.3.3: Solving for the eigenvalues ​​of the covariance matrix and the eigenvector , the calculation expression is: ; In the formula, For the eigenvalues, For The corresponding eigenvector; S2.3.4: Based on the characteristic value Size, select The eigenvector corresponding to the largest eigenvalue Become a new basis vector and form the principal component space; S2.3.5: Calculate the principal component score. The calculation expression is: ; In the formula, is the principal component score matrix.

[0011] S2.3.6: Convert the principal component score matrix The key parameters after dimensionality reduction are used as feature parameters , create a feature parameter set , used for fault monitoring model training; S3: Build a fault monitoring model FMM based on the LSTM model, and transform the feature parameter matrix into The fault monitoring model FMM is trained as a training set, and the fault monitoring model FMM is optimized based on the cross entropy loss function to obtain the fault monitoring model FMM'. Finally, the accuracy, precision, recall and F1 score are used to evaluate the performance of the fault monitoring model FMM'. The S3 comprises the following sub-steps: S3.1: Construct fault monitoring model FMM based on LSTM algorithm; S3.2: input the training set into the fault monitoring model FMM for training; S3.3: Based on the cross entropy loss function, the fault monitoring model FMM is optimized to obtain the fault monitoring model FMM'. The calculation expression is: ; In the formula, y b Indicates the bth characteristic parameter ,l(y b =d) takes a value of 0 or 1. When it takes a value of 1, it means that when y b Correctly classified as type d fault type, when the value is 0, it means y b Not correctly classified as a Class D fault type, is the number of training samples, C c Indicates the type of fault involved, Indicates the bth characteristic parameter The probability of being correctly classified as a type d fault, is the cross entropy loss; S3.4: Use accuracy, precision, recall and F1 score to evaluate the performance of the fault monitoring model FMM'; S4: Learning rate of fault monitoring model FMM' based on Bayesian optimization Adjust to obtain the learning rate of the fault monitoring model FMM' The optimal parameter configuration is used to obtain the optimized fault monitoring model FMM'', and the optimized fault monitoring model FMM'' is used to reconstruct and predict the ship power related signals in the ship power system, so as to realize the ship power system fault identification and the backtracking and analysis of the fault process; The S4 comprises the following specific steps: S4.1: Characteristic parameter matrix The value range is , learning rate The value range is , the characteristic parameters and learning rate All are set to v levels, and the calculation expression is: ; ; ; ; In the formula, is the characteristic parameter matrix The interval between each gear, For the The boundary of the file, is the learning rate The interval between each gear, For the The boundaries of the files; S4.2: Create a two-dimensional parameter space, where each row represents a feature parameter matrix level and each column represents a learning rate level, and initialize a database to store the sampling points and their corresponding objective function values ​​during the Bayesian optimization process; S4.3: Randomly select a set of initial sampling points from the two-dimensional parameter space, and use the initial sampling points to train an initial proxy model to approximate the performance of the fault monitoring model FMM'; S4.4: Calculate the expected improvement value of each combination of feature parameter matrix gear and learning rate gear; S4.5: Select the sampling point with the highest expected improvement value as the next evaluation point; S4.6: Train the fault monitoring model on the selected sampling points and evaluate its true objective function value; S4.7: Add the newly obtained sampling points and their corresponding objective function values ​​to the existing sampling point set; S4.8: Determine whether the Bayesian optimization has reached the preset number of iterations. If yes, proceed to S4.9; if no, proceed to S4.4; S4.9: Stop the optimization process and retrain the proxy model using the updated set of sampling points to improve its approximation accuracy; S4.10: The parameters in the trained proxy model are used as the optimal parameters of the optimized fault monitoring model FMM''; S4.11: Use the optimized fault monitoring model FMM'' to reconstruct and predict the ship power related signals in the ship power system, and compare them with the real-time collected ship power system operation data to realize the ship power system fault identification; S5: Establishing the characteristic parameter matrix The specific location of the fault L f The index relationship between them, that is, when the fault monitoring model FMM'' detects a fault in the ship's power system, the specific location of the fault in the ship's power system is traced in real time; The S5 comprises the following specific steps: Get recent The specific location where the fault occurred during the voyage of the ship L f , create a set of related parameter matrices The specific location of the fault L f The index relationship between , and then establish the relevant parameter matrix And the related parameter matrix The transformation relationship between , and then establish the relevant parameter matrix With key parameter matrix The transformation relationship between , and then establish the key parameter matrix With the characteristic parameter matrix The dimensionality reduction relationship between , and finally establish the characteristic parameter matrix The specific location of the fault L f Index relationship between That is, when the fault monitoring model FMM'' detects a fault in the ship's power system, the specific location of the fault in the ship's power system is traced in real time.

[0012] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0013] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring a fault in a ship power system, characterized in that: The specific steps include: S1: Get the latest Related parameter matrix of ship power system during the ship's voyage , , , create a set of related parameter matrices , and preprocess the relevant parameter matrix to obtain the preprocessed relevant parameter matrix set ; S2: Calculate the correlation parameter matrix based on the Pearson correlation coefficient The correlation between them is used to screen out the key parameter matrix Based on PCA dimensionality reduction, the key parameter matrix after dimensionality reduction is As the characteristic parameter matrix , create a feature parameter matrix set ; S3: Build a fault monitoring model FMM based on the LSTM model, and transform the feature parameter matrix into The fault monitoring model FMM is trained as a training set, and the fault monitoring model FMM is optimized based on the cross entropy loss function to obtain the fault monitoring model FMM'. Finally, the accuracy, precision, recall and F1 score are used to evaluate the performance of the fault monitoring model FMM'. S4: Learning rate of fault monitoring model FMM' based on Bayesian optimization Adjust to obtain the learning rate of the fault monitoring model FMM' The optimal parameter configuration is used to obtain the optimized fault monitoring model FMM'', and the optimized fault monitoring model FMM'' is used to reconstruct and predict the ship power related signals in the ship power system, so as to realize the ship power system fault identification and the backtracking and analysis of the fault process; S5: Establishing the characteristic parameter matrix The specific location of the fault L f The index relationship between them, that is, when the fault monitoring model FMM'' detects a ship power system fault, the specific location of the ship power system fault is traced in real time.

2. A method for monitoring a fault in a ship power system according to claim 1, characterized in that: The relevant parameter matrix It includes voltage fluctuation characteristic vector, frequency change characteristic vector, current mutation characteristic vector, harmonic content characteristic vector and three-phase imbalance characteristic vector.

3. A method for monitoring faults in a ship power system according to claim 1, characterized in that: The S1 comprises the following sub-steps: S1.1: Get the latest Related parameter matrix of secondary ship power system , , , create a set of related parameter matrices ; S1.2: Based on 3 Principle to remove relevant parameter matrix Outliers in ; S1.3: Normalization method for the correlation parameter matrix Perform normalization to obtain the relevant parameter matrix set after preprocessing .

4. A method for monitoring a fault in a ship power system according to claim 2, characterized in that: The S2 comprises the following sub-steps: S2.1: Calculate the correlation parameter matrix based on the Pearson correlation coefficient The correlation between them is calculated as: ； In the formula, and Two different related parameters The i-th value of and Two different related parameters The mean of For related parameters and related parameters The correlation coefficient between i is the eth first related parameter, f i is the fth second related parameter, is the mean of the first correlation parameter, is the mean of the second correlation parameter, ; S2.2: Setting relevance thresholds , using the judgment formula , remove the correlation parameter matrix with correlation below the correlation threshold , retain the correlation parameter matrix with correlations above the correlation threshold As a key parameter matrix ; S2.3: Use PCA to transform the key parameter matrix The high-dimensional feature vector in is converted into a low-dimensional feature vector to obtain the feature parameter matrix , create a feature parameter matrix set .

5. A method for monitoring faults in a ship power system according to claim 3, characterized in that: The S3 comprises the following sub-steps: S3.1: Construct fault monitoring model FMM based on LSTM algorithm; S3.2: input the training set into the fault monitoring model FMM for training; S3.3: Based on the cross entropy loss function, the fault monitoring model FMM is optimized to obtain the fault monitoring model FMM'. The calculation expression is: ； In the formula, y b Indicates the bth characteristic parameter ,l(y b =d) takes a value of 0 or 1. When it takes a value of 1, it means that when y b Correctly classified as type d fault type, when the value is 0, it means y b Not correctly classified as a Class D fault type, is the number of training samples, C c Indicates the type of fault involved, Indicates the bth characteristic parameter The probability of being correctly classified as a type d fault, is the cross entropy loss; S3.4: Use accuracy, precision, recall and F1 score to evaluate the performance of the fault monitoring model FMM'.

6. A method for monitoring a fault in a ship power system according to claim 1, characterized in that: The S4 comprises the following specific steps: S4.1: Characteristic parameter matrix The value range is , learning rate The value range is , the characteristic parameters and learning rate All are set to v levels, and the calculation expression is: ； ； ； ； In the formula, is the characteristic parameter matrix The interval between each gear, For the The boundary of the file, is the learning rate The interval between each gear, For the The boundaries of the files; S4.2: Create a two-dimensional parameter space, where each row represents a feature parameter matrix level and each column represents a learning rate level, and initialize a database to store the sampling points and their corresponding objective function values ​​during the Bayesian optimization process; S4.3: Randomly select a set of initial sampling points from the two-dimensional parameter space, and use the initial sampling points to train an initial proxy model to approximate the performance of the fault monitoring model FMM'; S4.4: Calculate the expected improvement value of each combination of feature parameter matrix gear and learning rate gear; S4.5: Select the sampling point with the highest expected improvement value as the next evaluation point; S4.6: Train the fault monitoring model on the selected sampling points and evaluate its true objective function value; S4.7: Add the newly obtained sampling points and their corresponding objective function values ​​to the existing sampling point set; S4.8: Determine whether the Bayesian optimization has reached the preset number of iterations. If yes, proceed to S4.9; if no, proceed to S4.4; S4.9: Stop the optimization process and retrain the proxy model using the updated set of sampling points to improve its approximation accuracy; S4.10: The parameters in the trained proxy model are used as the optimal parameters of the optimized fault monitoring model FMM''; S4.11: Use the optimized fault monitoring model FMM'' to reconstruct and predict the ship power related signals in the ship power system, and compare them with the real-time collected ship power system operation data to realize the ship power system fault identification; S4.12: If there is no abnormal signal before the fault occurs, the optimized fault monitoring model FMM'' will be used to reconstruct all operating signals of the ship's power system and trace back and analyze the fault.

7. A method for monitoring faults in a ship power system according to claim 1, characterized in that: The S5 comprises the following specific steps: Get recent The specific location where the fault occurred during the voyage of the ship L f , create a set of related parameter matrices The specific location of the fault L f The index relationship between , and then establish the relevant parameter matrix And the related parameter matrix The transformation relationship between , and then establish the relevant parameter matrix With key parameter matrix The transformation relationship between , and then establish the key parameter matrix With the characteristic parameter matrix The dimensionality reduction relationship between , and finally establish the characteristic parameter matrix The specific location of the fault L f Index relationship between That is, when the fault monitoring model FMM'' detects a fault in the ship's power system, the specific location of the fault in the ship's power system is traced in real time.

Images