Ship power device fault monitoring method and system based on principal component analysis method

Abstract

The invention provides a ship power device fault monitoring method and system based on a principal component analysis method, which are applied to the technical field of ship power monitoring, and the method comprises the following steps: determining a covariance matrix of a standardized data matrix; performing eigenvalue decomposition on the covariance matrix to obtain a plurality of standard eigenvectors and a plurality of eigenvalues respectively corresponding to the plurality of standard eigenvectors; determining the number of a plurality of principal components of the principal component analysis model and the cumulative variance contribution rate of the plurality of principal components; determining the variable quantity of the test vector in the principal component space as a first statistic, and determining the squared value of the projection of the test vector in the residual space as a second statistic; when the first statistical magnitude is greater than a first statistical magnitude threshold value or the second statistical magnitude is greater than a second statistical magnitude threshold value, determining the contribution rate of each monitoring parameter in the test vector; taking the monitoring parameters of which the increase values of the contribution rates are greater than a preset contribution rate threshold value as abnormal parameters; according to the invention, abnormal parameters can be monitored in time at the early stage of an accident.

Description

Technical Field

[0001] The present invention relates to the technical field of ship power monitoring, and in particular to a ship power plant fault monitoring method and system based on principal component analysis. Background Art

[0002] The ship's power plant is a very complex system, consisting of dozens of subsystems and equipment. The system is very complex, involving hundreds of types and quantities of equipment, so the probability of failure of the power plant during operation is relatively high.

[0003] Traditional methods of monitoring accidents in ship power plants are based on sensor data design rules. However, in the actual operation of power plants, many accidents occur for the first time and when the monitoring parameters do not change much. This makes it difficult for traditional accident monitoring methods to detect abnormal parameters in a timely manner, thus causing major accidents.

[0004] It can be seen that the ship power plant fault monitoring method in the related art has technical problems of lag and low monitoring efficiency. Summary of the invention

[0005] The present invention provides a method and system for monitoring faults of a ship power plant based on principal component analysis, which are used to solve the defects of the prior art method for monitoring faults of a ship power plant, such as hysteresis and low monitoring efficiency, and to improve the probability of detecting a fault at an early stage of the fault.

[0006] The present invention provides a method for monitoring faults of a ship power plant based on principal component analysis, comprising the following steps.

[0007] Preprocessing the original measurement data to obtain a standardized data matrix, wherein the original measurement data is monitoring data of a ship power plant; determining the covariance matrix of the standardized data matrix; performing eigenvalue decomposition on the covariance matrix to obtain a plurality of standard eigenvectors and a plurality of eigenvalues ​​respectively corresponding to the plurality of standard eigenvectors; determining the number of a plurality of principal components of a principal component analysis model and the cumulative variance contribution rate of the plurality of principal components based on the plurality of standard eigenvectors and the plurality of eigenvalues; determining the principal component space and the residual space of the principal component analysis model based on the number and the cumulative variance contribution rate; determining the change of a test vector in the principal component space as a first statistic, and determining the square value of the projection of the test vector in the residual space as a second statistic; determining the contribution rate of each monitoring parameter in the test vector when the first statistic is greater than a first statistic threshold or the second statistic is greater than a second statistic threshold; and taking the monitoring parameter whose contribution rate increase value is greater than a preset contribution rate threshold as an abnormal parameter.

[0008] According to a method for monitoring faults of a ship power plant based on principal component analysis provided by the present invention, after taking the monitoring parameter whose contribution rate increase value is greater than a preset contribution rate threshold as an abnormal parameter, the method further includes: determining the device monitored by the sensor of the ship power plant corresponding to the abnormal parameter, and treating the device as a faulty device.

[0009] According to a ship power plant fault monitoring method based on principal component analysis provided by the present invention, the preprocessing of the original measurement data to obtain a standardized data matrix includes: normalizing the original measurement data by mean standard deviation to obtain a standardized data matrix, wherein each parameter in the standardized data matrix is ​​assigned an equal weight.

[0010] According to a ship power plant fault monitoring method based on principal component analysis provided by the present invention, the first statistic is the distance from the projection point of the test vector in the principal component space to the center point of the sample mean, and the first statistic is used to indicate the degree to which the test vector deviates from the principal component analysis model in terms of change trend and change amplitude; the second statistic is the Euclidean distance from the test vector to the principal component space defined by the principal component analysis model, and the second statistic is used to indicate the degree of mismatch between the test vector and the principal component analysis model.

[0011] According to a method for monitoring faults of a ship power plant based on principal component analysis provided by the present invention, the first statistic is subject to the degree of freedom and , the confidence level is The F distribution, where is the number of the pivots, is the number of parameters to be monitored, is the significance level, and the first statistic threshold is the critical value of the F distribution.

[0012] According to a ship power plant fault monitoring method based on principal component analysis method provided by the present invention, determining the contribution rate of each monitoring parameter in the test vector includes: determining a first contribution rate of each monitoring parameter in the test vector to the first statistic; determining a second contribution rate of each monitoring parameter in the test vector to the second statistic.

[0013] The present invention also provides a ship power plant fault monitoring system based on principal component analysis, comprising the following modules: a preprocessing module, used for preprocessing original measurement data to obtain a standardized data matrix, wherein the original measurement data is monitoring data of the ship power plant; a first determination module, used for determining the covariance matrix of the standardized data matrix; a decomposition module, used for performing eigenvalue decomposition on the covariance matrix to obtain a plurality of standard eigenvectors and a plurality of eigenvalues ​​respectively corresponding to the plurality of standard eigenvectors; a second determination module, used for determining the number of a plurality of principal components of the principal component analysis model based on the plurality of standard eigenvectors and the plurality of eigenvalues, and the plurality of The cumulative variance contribution rate of the principal component; a third determination module, used to determine the principal component space and the residual space of the principal component analysis model based on the quantity and the cumulative variance contribution rate; a fourth determination module, used to determine the change of the test vector in the principal component space as the first statistic, and determine the square value of the projection of the test vector in the residual space as the second statistic; a fifth determination module, used to determine the contribution rate of each monitoring parameter in the test vector when the first statistic is greater than the first statistic threshold or the second statistic is greater than the second statistic threshold; an output module, used to take the monitoring parameter whose increase in the contribution rate is greater than the preset contribution rate threshold as an abnormal parameter.

[0014] The present invention also provides an electronic 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, a ship power plant fault monitoring method based on principal component analysis as described in any one of the above is implemented.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for monitoring faults of a ship power plant based on the principal component analysis method as described above is implemented.

[0016] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned methods for monitoring faults of a ship power plant based on the principal component analysis method.

[0017] The ship power plant fault monitoring method and system based on principal component analysis provided by the present invention obtain a standardized data matrix by preprocessing the monitoring data of the ship power plant; determine the covariance matrix of the standardized data matrix; and perform data decomposition on the covariance matrix to obtain multiple standard eigenvectors and multiple corresponding eigenvalues, thereby determining the number of principal components of the principal component analysis model according to the result of eigenvalue decomposition; determine the cumulative variance contribution rate of multiple principal components of the principal component analysis model according to the multiple standard eigenvectors and the multiple eigenvalues, and determine the principal component space and residual space of the principal component analysis model, thereby, the modeling process of the principal component analysis model is completed. , principal component analysis divides the original data space into principal component space and residual space; determines the change of the test vector in the principal component space as the first statistic, and determines the square value of the projection of the test vector in the residual space as the second statistic; when the first statistic is greater than the first statistic threshold or the second statistic is greater than the second statistic threshold, determines the contribution rate of each monitoring parameter in the test vector; the monitoring parameter whose contribution rate increase value is greater than the preset contribution rate threshold is used as an abnormal parameter, thereby distinguishing the contribution rates of different monitoring parameters, and being able to monitor abnormal parameters in time at the early stage of the accident, greatly improving the response time of the fault, and leaving a lot of time for the handling of the fault. This solves the technical problem that the fault monitoring method of ship power plant in the related technology has hysteresis and low monitoring efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced one by one below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 It is a flow chart of a method for monitoring a fault of a ship power plant based on principal component analysis provided by the present invention.

[0020] Figure 2 It is a schematic diagram of a flow chart for determining the contribution rate of monitoring parameters provided by the present invention.

[0021] Figure 3 It is a system fault detection flow chart based on principal component analysis provided by the present invention.

[0022] Figure 4 It is a structural schematic diagram of a ship power plant fault monitoring system based on principal component analysis provided by the present invention.

[0023] Figure 5 It is a schematic diagram of the physical structure of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are 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.

[0025] The ship power plant is a very complex system, consisting of dozens of subsystems and equipment. The system is very complex, involving hundreds of types and quantities of equipment. Therefore, the probability of failure of the power plant during operation is relatively high. Once a typical design accident occurs, it will cause catastrophic consequences. Based on the above purpose, it is necessary to propose a more reliable ship power plant fault monitoring technology.

[0026] Traditional methods of monitoring faults in ship power plants are based on sensor data design rules. The advantage of this method is that the accident characteristics are relatively clear, and the accident can be clearly judged within the scope of the rules. However, in the actual operation of the power plant, many accidents occur when they have never occurred before and the monitoring parameters have not changed much. This brings great difficulties to the traditional methods of monitoring faults in ship power plants, and often leads to major accidents.

[0027] The embodiment of the present invention proposes a ship power plant fault monitoring method based on principal component analysis. Taking the ship power plant as the research object, power plant fault monitoring based on principal component analysis is carried out. Compared with the traditional ship power plant fault monitoring method, this method can increase the probability of detecting a fault in the early stage of occurrence.

[0028] refer to Figure 1 , Figure 1 Schematic diagram of the process of the ship power plant fault monitoring method based on principal component analysis provided by the present invention, such as Figure 1 As shown, the method includes the following steps.

[0029] Step 101, preprocessing the original measurement data to obtain a standardized data matrix, wherein the original measurement data is monitoring data of a ship power plant.

[0030] Principal component analysis (PCA) is a multivariate statistical analysis method, which was originally used to perform optimal linear and planar fitting of discrete points in space. Subsequently, PCA was used for research in data compression, fault detection, and fault identification.

[0031] In the embodiment of the present invention, the original measurement data is preprocessed by using the mean standard deviation normalization method to ensure that each parameter in the original data is assigned an equal weight.

[0032] Step 102, determining the covariance matrix of the standardized data matrix.

[0033] In the embodiment of the present invention, the standardized sample data matrix (standardized data matrix) is ,in, represents the number of samples, Represents the number of parameters to be monitored. According to statistical principles, it is approximately calculated using the following formula (1): The covariance matrix of .

[0034] (1)

[0035] in, represents the covariance matrix, represents the normalized data matrix, represents the transposed matrix of the normalized data matrix, Indicates the number of parameters to be monitored.

[0036] The covariance matrix is ​​a matrix whose elements are the covariances between the elements of each vector. The covariance matrix is ​​a natural extension from scalar random variables to high-dimensional random vectors. It represents the covariance relationship between multidimensional random variables in the form of a matrix, and has unique properties such as symmetry and semi-positive definiteness.

[0037] Step 103 , performing eigenvalue decomposition on the covariance matrix to obtain a plurality of standard eigenvectors and a plurality of eigenvalues ​​respectively corresponding to the plurality of standard eigenvectors.

[0038] In the embodiment of the present invention, the covariance matrix Perform eigenvalue decomposition to obtain the standard eigenvector of the covariance matrix and the corresponding eigenvalues These eigenvalues ​​represent the variance of the data in the direction of each principal component.

[0039] Step 104: Determine the number of principal components of the principal component analysis model and the cumulative variance contribution rates of the principal components based on the plurality of standard eigenvectors and the plurality of eigenvalues.

[0040] Next, the embodiment of the present invention needs to determine the number of principal components of the principal component analysis model This is usually done by observing the relative sizes of the eigenvalues, choosing The largest eigenvalues ​​are selected so that their cumulative variance contribution reaches a certain threshold (such as 80% or 90%).

[0041] Determine the number of principal components of the principal component analysis model based on the results of eigenvalue decomposition . The percentage of the original information contained in the principal component is the cumulative variance contribution rate of the principal component. For details, please refer to the following formula (2).

[0042] (2)

[0043] in, represents the cumulative variance contribution rate of the principal component, represents the contribution rate of a single principal component, represents the number of principal components, represents the ordinal number of the pivot element, Indicates The eigenvalues ​​corresponding to the standard eigenvectors (pivots), Indicates the sequence number of the sum index, Indicates The characteristic values ​​( Indicates calculation of all The total variance is obtained by summing the eigenvalues.

[0044] Step 105: Determine the principal component space and residual space of the principal component analysis model based on the quantity and cumulative variance contribution rate.

[0045] In an embodiment of the present invention, in principal component analysis, the eigenvectors and eigenvalues ​​obtained by eigenvalue decomposition of the covariance matrix (or the transpose of the standardized data matrix multiplied by itself) can be used to construct the principal components (principal elements) and residual space.

[0046] First, the covariance matrix Perform eigenvalue decomposition to obtain the standard eigenvector of the covariance matrix and the corresponding eigenvalues , these eigenvalues ​​are sorted from largest to smallest.

[0047] Secondly, the number of pivots The selection is usually based on the cumulative contribution rate of the eigenvalues, that is, The percentage of the sum of the eigenvalues ​​to the sum of all eigenvalues. This percentage is called the cumulative variance contribution rate. Usually, a value so that the cumulative variance contribution rate reaches a preset level (such as 80% or 90%).

[0048] The main space is from the front feature vector These eigenvectors are projections of the data in the direction of maximum variance, so they constitute the principal components of the data.

[0049] The residual space is the orthogonal complement of the principal element space, consisting of the remaining feature vector Zhang Cheng. These eigenvectors correspond to directions with smaller variance in the data and are usually considered to be noise or redundant information.

[0050] The pivots are the main directions of change in the data, and they capture most of the information in the data. The residuals contain the parts of the data that are not captured by the pivots, and usually contain noise or minor information.

[0051] Further obtain the main component and residual space of the model and , and the pivot and residual space projection matrices: and , please refer to the following formula (3) for details.

[0052] = (3)

[0053] in, represents the principal space projection matrix, represents the principal space, is the transposed matrix of the principal element space, represents the residual space projection matrix, represents the residual space, Represents the transposed matrix of the residual space.

[0054] Through the embodiments of the present invention, the data dimension can be reduced to the principal element space while retaining most of the important information and ignoring noise and redundancy.

[0055] Step 106, determining the variation of the test vector in the principal component space as the first statistic, and determining the square value of the projection of the test vector in the residual space as the second statistic.

[0056] In the embodiment of the present invention, the first statistic is Statistics: Statistics represent test vectors The mathematical meaning of the change in the principal element space is The distance from the projection point in the principal element space to the center point of the sample mean, which describes The degree of deviation from the principal component analysis model in terms of change trend and change amplitude is a measure of the internal changes in the principal component space.

[0057] In the embodiment of the present invention, the second statistic is Statistics: The statistic describes the test vector The square value of the projection in the residual space is The Euclidean distance to the principal component space defined by the model quantifies The degree of mismatch between the PCA model and the PCA model. The statistic is a measure of the internal variation of the residual space, converting multidimensional data into one-dimensional data for analysis.

[0058] Step 107 : when the first statistic is greater than the first statistic threshold or the second statistic is greater than the second statistic threshold, determine the contribution rate of each monitoring parameter in the test vector.

[0059] In the embodiment of the present invention, the first statistical threshold can be express, express Statistical threshold. When the parameter is in a normal state, its test data follows a multivariate normal distribution. The statistic falls on within the range.

[0060] In the embodiment of the present invention, the second statistical threshold can be express, yes The upper threshold of the statistic. The projection in the residual space is very small. The statistical value is When a parameter fails, the correlation between parameters is destroyed, resulting in The projection in the residual space is increased so that The Euclidean distance to the model principal component space exceeds the threshold .

[0061] Step 108: The monitoring parameter whose contribution rate increase value is greater than a preset contribution rate threshold is regarded as an abnormal parameter.

[0062] In the embodiment of the present invention, a first contribution rate of each monitoring parameter in the test vector to the first statistic is determined; and a second contribution rate of each monitoring parameter in the test vector to the second statistic is determined.

[0063] Test Vector In The total variation of the parameters in the residual space constitutes of Therefore, each monitoring parameter is calculated separately for The contribution rate of the statistic is used to evaluate the operating status of the parameter based on the size of the parameter contribution rate.

[0064] Test Vector The total change of all monitoring parameters in the principal element space is of Statistics. The change of each monitoring parameter in the principal element space is The contribution rate of the statistic indicates the operating status of this parameter.

[0065] Under normal conditions, each monitoring parameter The contribution rate of the statistic should also be the same. A parameter in When the contribution rate of An exception occurred.

[0066] Through the embodiment of the present invention, the principal component analysis method is used to monitor the fault of the ship power plant, and the contribution rate of different monitoring parameters is distinguished. When the key parameters reflecting the operation of the power plant change, the abnormal parameters can be monitored in time at the early stage of the accident, which greatly improves the response time of the fault and can leave a lot of time for the handling of the fault. Therefore, the technical problem of the ship power plant fault monitoring method in the related art, which has hysteresis and low monitoring efficiency, is solved.

[0067] According to a ship power plant fault monitoring method based on principal component analysis provided by the present invention, after taking the monitoring parameter whose contribution rate increase value is greater than a preset contribution rate threshold as an abnormal parameter, the method further includes the following steps.

[0068] The equipment monitored by the sensor of the ship power plant corresponding to the abnormal parameter is determined, and the equipment is regarded as a faulty equipment.

[0069] When a parameter in a ship's power plant is abnormal, the first thing to do is to determine which sensor detected the abnormality. This can usually be achieved through the ship's monitoring system or alarm system, which can display the data of each sensor in real time and issue an alarm when the data exceeds the normal range.

[0070] After determining the monitoring equipment, it is necessary to further analyze the operating status and performance parameters of the equipment to determine whether it has indeed failed. This requires a comprehensive analysis of the data from multiple sensors, for example, comparing whether the temperature, pressure and other parameters of different parts of the same equipment are consistent, or comparing whether the performance parameters of the equipment are significantly different from those of similar equipment.

[0071] According to a method for monitoring faults of a ship power plant based on principal component analysis provided by the present invention, raw measurement data is preprocessed to obtain a standardized data matrix, which includes the following steps.

[0072] The original measurement data is normalized by the mean and standard deviation to obtain a standardized data matrix, in which each parameter in the standardized data matrix is ​​assigned an equal weight.

[0073] In the embodiment of the present invention, the original measurement data is preprocessed by using the mean standard deviation normalization method to ensure that each parameter in the original data is assigned an equal weight. For details, please refer to the following formula (4).

[0074] (4)

[0075] in, represents the standardized data, Indicates the index of the data (for example, observations), Represents the index of a feature (for example, features or variables); Represents raw data, i.e., data that has not been standardized; Indicates The average value of a feature, that is, the mean of all observations on this feature; Indicates The standard deviation of a feature is the square root of the average of the squares of the deviations of all observations on that feature from its mean. Represents the total number of observations (parameters to be monitored).

[0076] Through the embodiments of the present invention, the comparability of data between different features is ensured, and the dimension effect in the data is removed.

[0077] According to a ship power plant fault monitoring method based on principal component analysis provided by the present invention, the first statistic is the distance from the projection point of the test vector in the principal component space to the center point of the sample mean, and the first statistic is used to indicate the degree to which the test vector deviates from the principal component analysis model in terms of change trend and change amplitude.

[0078] The second statistic is the Euclidean distance from the test vector to the principal component space defined by the principal component analysis model. The second statistic is used to indicate the degree of mismatch between the test vector and the principal component analysis model.

[0079] In the embodiment of the present invention, the first statistic can be expressed as Statistics, Statistics represent test vectors The mathematical meaning of the change in the principal element space is The distance from the projection point in the principal element space to the center point of the sample mean, which describes The degree of deviation from the principal component analysis model in terms of change trend and change amplitude is a measure of the internal changes in the principal component space. The definition of the statistic can be expressed by the following formula (5).

[0080] (5)

[0081] in, represents the first statistic, , represents a positive definite matrix, ( ) represents the eigenvalue of the principal component, represents the transposed matrix of the principal element space, represents the test vector, represents the principal space, represents the transposed matrix of the test vector, express Statistics threshold.

[0082] In the embodiment of the present invention, the second statistic can be expressed as Statistics, The statistic describes the test vector The square value of the projection in the residual space is The Euclidean distance to the principal component space defined by the model quantifies The degree of mismatch between the PCA model and the PCA model. The statistic is a measure of the internal changes in the residual space. The conversion of multidimensional data into one-dimensional data for analysis can be specifically expressed by the following formula (6):

[0083] (6)

[0084] in, represents the second statistic, Represents the test vector The Euclidean distance to the principal component space defined by the model, represents the residual space projection matrix, represents the test vector, represents the transposed matrix, represents the principal space, is the transposed matrix of the principal element space, represents the residual space, represents the transposed matrix of the residual space, Represents the identity matrix.

[0085] In the above formula (6), because is a unit orthogonal matrix, so we have ,but The statistic can be simplified as the following formula (7).

[0086] (7)

[0087] in, represents the second statistic, Represents the test vector The Euclidean distance to the principal component space defined by the model, represents the transposed matrix of the test vector, represents the identity matrix, represents the principal space, is the transposed matrix of the principal element space, represents the principal space projection matrix, Represents the second statistic threshold.

[0088] According to a method for monitoring faults of a ship power plant based on principal component analysis provided by the present invention, the first statistic is subject to the degree of freedom and , the confidence level is The F distribution, where is the number of the pivots, is the number of parameters to be monitored, is the significance level, and the first statistic threshold is the critical value of the F distribution.

[0089] In the embodiment of the present invention, when the parameters are in a normal state, the test data thereof obeys a multivariate normal distribution. The statistic falls within the first statistic threshold within the range. It can be approximately calculated using the following formula (8).

[0090] (8)

[0091] in, represents the first statistical threshold, represents the number of pivots, Indicates the number of parameters to be monitored, Indicates with and degrees of freedom and a confidence level of of Critical value of the distribution.

[0092] The F distribution is a continuous probability distribution that often appears in statistical tests such as analysis of variance and regression analysis. The F distribution has two degrees of freedom parameters: the numerator degrees of freedom (usually expressed as ) and the denominator degrees of freedom (usually expressed as ). Given a confidence level In the embodiment of the present invention, it is necessary to find the critical value of the F distribution, which is used to determine the significance of the statistical test.

[0093] In the embodiment of the present invention, the second statistical threshold yes The upper threshold of the statistic. The projection in the residual space is very small. The statistical value is When a parameter fails, the correlation between parameters is destroyed, resulting in The projection in the residual space is increased so that The Euclidean distance to the model principal component space exceeds the threshold . It can be approximately calculated using the following formula (9)-formula (11).

[0094] (9)

[0095] in, Indicates that based on The power sum of eigenvalues, is the covariance matrix No. The characteristic value Power, here , and the sum is from the first ( ) eigenvalues, that is, only the eigenvalues ​​in the residual space are considered. represents the index number of the eigenvalue in the residual space, Representation and Related functions are used for subsequent calculations.

[0096] Under certain assumptions, It approximately obeys the normal distribution, that is, refer to the following formula (10).

[0097] (10)

[0098] in, represents the second statistical threshold, represents the sum of powers based on the first eigenvalue, Based on the power sum of the second eigenvalue, Representation and Related functions, Represents a normal distribution.

[0099] The selected confidence level is When Statistics upper threshold For the calculation expression of , please refer to the following formula (11).

[0100] (11)

[0101] in, represents the second statistical threshold, represents the sum of powers based on the first eigenvalue, Based on the power sum of the second eigenvalue, Representation and (Based on The power of the eigenvalues ​​and the function related to Represents the standard normal distribution Confidence limits.

[0102] refer to Figure 2 , Figure 2 It is a schematic diagram of a flow chart for determining the contribution rate of monitoring parameters provided by the present invention.

[0103] According to a method for monitoring faults of a ship power plant based on principal component analysis provided by the present invention, the contribution rate of each monitoring parameter in a test vector is determined, comprising the following steps: Step 201: determine a first contribution rate of each monitoring parameter in a test vector to a first statistic.

[0104] Step 202: Determine a second contribution rate of each monitoring parameter in the test vector to a second statistic.

[0105] In the embodiment of the present invention, a Contribution analysis method of statistics (second statistics), test vector In The total variation of the parameters in the residual space constitutes of Therefore, each monitoring parameter is calculated separately for The contribution rate of the statistic is used to evaluate the operating status of the parameter according to the size of the parameter contribution rate. Middle parameter pairs The contribution rate of the statistic can be specifically referred to the following formula (12).

[0106] (12)

[0107] in, Indicates parameter pairs The contribution rate of the statistic, is the test vector The projection in the residual space, yes No. elements; represents the principal space, is the transposed matrix of the principal element space, represents the test vector, represents the identity matrix, represents the principal space projection matrix, represents the residual space projection matrix, Represents the transposed matrix of the test vector.

[0108] Under normal conditions, each monitoring parameter The contribution of the statistic should be the same. A parameter in When the contribution rate of increases, it proves that the parameter An exception occurred.

[0109] In the embodiment of the present invention, a Contribution analysis method of statistics (first statistics), test vector The total change of all monitoring parameters in the principal element space is of Statistics. The change of each monitoring parameter in the principal element space is The contribution rate of the statistic indicates the operating status of this parameter. The monitoring parameters in The contribution rate of the statistic can be obtained by referring to the following formulas (13) to (14).

[0110] Calculate monitoring parameters For score vector Contribution rate: (13)

[0111] in, Indicates The test vector The contribution rate of each monitoring parameter, Indicates score vector, Represents the covariance matrix The eigenvector of No. elements, Indicates eigenvalues, Indicates test vectors.

[0112] In the formula, express The score vector of ; The eigenvector representing ∑ No. elements; , will soon Middle elements remain unchanged and the rest are set to zero.

[0113] Calculate monitoring parameters right Contribution rate of statistics: (14)

[0114] in, Indicates Test vectors Statistics, is the number of parameters to be monitored, Indicates the index number of the eigenvalue, Represents the index number of the test vector, Indicates score vector, Represents the covariance matrix The eigenvector of No. elements, Indicates eigenvalues, Indicates test vectors.

[0115] Under normal conditions, each monitoring parameter The contribution rate of the statistic should also be the same. A parameter in When the contribution rate of An exception occurred.

[0116] refer to Figure 3 , Figure 3 It is a system fault detection flow chart based on principal component analysis provided by the present invention, which includes PCA model establishment and sensor monitoring.

[0117] The establishment of the principal component analysis (PCA) model includes: obtaining the original operating data, standardizing the data matrix, calculating the eigenvalues ​​and eigenvectors, and calculating the number of principal components ( ), calculate monitoring parameters ( ), calculate monitoring parameters ( ).

[0118] Sensor monitoring includes: acquiring test vectors, standardizing test vectors, calculating real-time , Calculate real-time , determine whether or , if not, return to execute the standardized test vector; if yes, continue to execute abnormal alarm, calculate the contribution rate of each sensor, and determine the high contribution rate variable.

[0119] The principal component analysis model is established, and the process of fault detection and fault identification is described as follows: Figure 3 As shown in the figure, when analyzing the contribution rate, the main The contribution rate of the statistic is mainly distributed in Statistics are supplementary. This is because Statistics are calculated through the principal component space and represent the system state, while The statistics are calculated from the residual space and mainly reflect the system noise and disturbance. The contribution rate distribution of the statistic can more reasonably reflect the contribution rate of the monitoring parameters to the fault.

[0120] The present invention monitors the fault of the ship power plant through the principal component analysis method, distinguishes the contribution rate of different monitoring parameters, and can timely monitor the abnormal parameters at the early stage of the accident when the key parameters reflecting the operation change, greatly improving the response time of the fault, and leaving a lot of time for the handling of the fault. This solves the technical problems of the ship power plant fault monitoring method in the related art, which has hysteresis and low monitoring efficiency.

[0121] The following is a description of the ship power plant fault monitoring system based on the principal component analysis method provided by the present invention. The ship power plant fault monitoring system based on the principal component analysis method described below and the ship power plant fault monitoring method based on the principal component analysis method described above can be referenced to each other.

[0122] refer to Figure 4 , Figure 4 It is a structural schematic diagram of the ship power plant fault monitoring system based on the principal component analysis method provided by the present invention, which includes: a preprocessing module 401, a first determination module 402, a decomposition module 403, a second determination module 404, a third determination module 405, a fourth determination module 406, a fifth determination module 407 and an output module 408.

[0123] The preprocessing module 401 is used to preprocess the original measurement data to obtain a standardized data matrix, wherein the original measurement data is monitoring data of a ship power plant.

[0124] The first determination module 402 is used to determine the covariance matrix of the standardized data matrix.

[0125] The decomposition module 403 is used to perform eigenvalue decomposition on the covariance matrix to obtain a plurality of standard eigenvectors and a plurality of eigenvalues ​​respectively corresponding to the plurality of standard eigenvectors.

[0126] The second determination module 404 is used to determine the number of a plurality of principal components of the principal component analysis model and the cumulative variance contribution rate of the plurality of principal components based on the plurality of standard eigenvectors and the plurality of eigenvalues.

[0127] The third determination module 405 is used to determine the principal component space and the residual space of the principal component analysis model based on the cumulative variance contribution rate.

[0128] The fourth determination module 406 is used to determine the variation of the test vector in the principal component space as the first statistic, and determine the square value of the projection of the test vector in the residual space as the second statistic.

[0129] The fifth determination module 407 is configured to determine a contribution rate of each monitoring parameter in the test vector when the first statistic is greater than a first statistic threshold or the second statistic is greater than a second statistic threshold.

[0130] The output module 408 is used to take the monitoring parameter whose contribution rate increase value is greater than a preset contribution rate threshold as an abnormal parameter.

[0131] Specifically, the above-mentioned ship power plant fault monitoring system based on principal component analysis method provided by the present invention can implement all the method steps implemented by the above-mentioned ship power plant fault monitoring method embodiment based on principal component analysis method, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as the method embodiment will not be described in detail here.

[0132] Figure 5 is a schematic diagram of the physical structure of the electronic device provided by the present invention, such as Figure 5As shown, the electronic device may include: a processor (processor) 510, a communication interface (Communications Interface) 520, a memory (memory) 530 and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other through the communication bus 540. The processor 510 can call the logic instructions in the memory 530 to execute a ship power plant fault monitoring method based on the principal component analysis method, the method comprising: preprocessing the original measurement data to obtain a standardized data matrix, wherein the original measurement data is the monitoring data of the ship power plant; determining the covariance matrix of the standardized data matrix; performing eigenvalue decomposition on the covariance matrix to obtain multiple standard eigenvectors and multiple eigenvalues ​​corresponding to the multiple standard eigenvectors; based on the multiple standard eigenvectors and the multiple eigenvalues, determining the number of multiple principal components of the principal component analysis model and the cumulative variance contribution rate of the multiple principal components; based on the cumulative variance contribution rate, determining the principal component space and the residual space of the principal component analysis model; determining the change of the test vector in the principal component space as the first statistic, and determining the square value of the projection of the test vector in the residual space as the second statistic; when the first statistic is greater than the first statistic threshold or the second statistic is greater than the second statistic threshold, determining the contribution rate of each monitoring parameter in the test vector; and taking the monitoring parameter whose contribution rate increase value is greater than the preset contribution rate threshold as an abnormal parameter.

[0133] In addition, the logic instructions in the above-mentioned memory 530 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0134] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the ship power plant fault monitoring method based on the principal component analysis method provided by the above methods. The method includes: preprocessing the original measurement data to obtain a standardized data matrix, wherein the original measurement data is the monitoring data of the ship power plant; determining the covariance matrix of the standardized data matrix; performing eigenvalue decomposition on the covariance matrix to obtain multiple standard eigenvectors and corresponding to the multiple standard eigenvectors respectively. multiple eigenvalues; based on multiple standard eigenvectors and multiple eigenvalues, determine the number of multiple principal components of the principal component analysis model and the cumulative variance contribution rate of the multiple principal components; based on the cumulative variance contribution rate, determine the principal component space and the residual space of the principal component analysis model; determine the change of the test vector in the principal component space as the first statistic, and determine the square value of the projection of the test vector in the residual space as the second statistic; when the first statistic is greater than the first statistic threshold or the second statistic is greater than the second statistic threshold, determine the contribution rate of each monitoring parameter in the test vector; take the monitoring parameter whose contribution rate increase value is greater than the preset contribution rate threshold as an abnormal parameter.

[0135] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the ship power plant fault monitoring method based on principal component analysis provided by the above methods, the method comprising: preprocessing the original measurement data to obtain a standardized data matrix, wherein the original measurement data is the monitoring data of the ship power plant; determining the covariance matrix of the standardized data matrix; performing eigenvalue decomposition on the covariance matrix to obtain a plurality of standard eigenvectors and a plurality of eigenvalues ​​respectively corresponding to the plurality of standard eigenvectors; based on the plurality of standard eigenvectors, The eigenvector and multiple eigenvalues ​​are used to determine the number of multiple principal components of the principal component analysis model and the cumulative variance contribution rate of the multiple principal components; based on the cumulative variance contribution rate, the principal component space and the residual space of the principal component analysis model are determined; the change of the test vector in the principal component space is determined as the first statistic, and the square value of the projection of the test vector in the residual space is determined as the second statistic; when the first statistic is greater than the first statistic threshold or the second statistic is greater than the second statistic threshold, the contribution rate of each monitoring parameter in the test vector is determined; the monitoring parameter whose contribution rate increase value is greater than the preset contribution rate threshold is taken as an abnormal parameter.

[0136] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0137] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0138] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for monitoring faults of a ship power plant based on principal component analysis, characterized in that: include: Preprocessing the original measurement data to obtain a standardized data matrix, wherein the original measurement data is monitoring data of a ship power plant; determining a covariance matrix of the standardized data matrix; Performing eigenvalue decomposition on the covariance matrix to obtain a plurality of standard eigenvectors and a plurality of eigenvalues ​​respectively corresponding to the plurality of standard eigenvectors; Determining the number of a plurality of principal components of a principal component analysis model and the cumulative variance contribution rates of the plurality of principal components based on the plurality of standard eigenvectors and the plurality of eigenvalues; Based on the quantity and the cumulative variance contribution rate, determining the principal component space and the residual space of the principal component analysis model; Determine a variation of the test vector in the principal component space as a first statistic, and determine a square value of a projection of the test vector in the residual space as a second statistic; When the first statistic is greater than a first statistic threshold or the second statistic is greater than a second statistic threshold, determining a contribution rate of each monitoring parameter in the test vector; The monitoring parameter whose contribution rate increase value is greater than a preset contribution rate threshold is regarded as an abnormal parameter.

2. The method for monitoring faults of a ship power plant based on principal component analysis according to claim 1, characterized in that: After taking the monitoring parameter whose contribution rate increase value is greater than a preset contribution rate threshold as an abnormal parameter, the method further includes: The equipment monitored by the sensor of the ship power plant corresponding to the abnormal parameter is determined, and the equipment is regarded as a faulty equipment.

3. The method for monitoring faults of a ship power plant based on principal component analysis according to claim 1, characterized in that: The raw measurement data is preprocessed to obtain a standardized data matrix, including: The original measurement data is normalized by the mean and standard deviation to obtain a standardized data matrix, wherein each parameter in the standardized data matrix is ​​assigned an equal weight.

4. The method for monitoring faults of a ship power plant based on principal component analysis according to claim 1, characterized in that: The first statistic is the distance from the projection point of the test vector in the principal component space to the center point of the sample mean, and the first statistic is used to indicate the degree to which the test vector deviates from the principal component analysis model in terms of change trend and change amplitude; The second statistic is the Euclidean distance from the test vector to the principal component space defined by the principal component analysis model, and the second statistic is used to indicate the degree of mismatch between the test vector and the principal component analysis model.

5. The method for monitoring faults of a ship power plant based on principal component analysis according to claim 1, characterized in that: The first statistic has the degrees of freedom and , the confidence level is The F distribution, where is the number of the pivots, is the number of parameters to be monitored, is the significance level, and the first statistic threshold is the critical value of the F distribution.

6. The method for monitoring faults of a ship power plant based on principal component analysis according to claim 1, characterized in that: Determining the contribution rate of each monitoring parameter in the test vector includes: Determine a first contribution rate of each monitoring parameter in the test vector to the first statistic; Determine a second contribution rate of each monitoring parameter in the test vector to the second statistic.

7. A ship power plant fault monitoring system based on principal component analysis, characterized in that: include: A preprocessing module, used for preprocessing the original measurement data to obtain a standardized data matrix, wherein the original measurement data is monitoring data of a ship power plant; A first determination module, used to determine the covariance matrix of the standardized data matrix; A decomposition module, used for performing eigenvalue decomposition on the covariance matrix to obtain a plurality of standard eigenvectors and a plurality of eigenvalues ​​respectively corresponding to the plurality of standard eigenvectors; A second determination module is used to determine the number of a plurality of principal components of the principal component analysis model and the cumulative variance contribution rate of the plurality of principal components based on the plurality of standard eigenvectors and the plurality of eigenvalues; A third determination module is used to determine the principal component space and the residual space of the principal component analysis model based on the quantity and the cumulative variance contribution rate; a fourth determination module, configured to determine a variation of a test vector in the principal component space as a first statistic, and to determine a square value of a projection of the test vector in the residual space as a second statistic; a fifth determination module, configured to determine a contribution rate of each monitoring parameter in the test vector when the first statistic is greater than a first statistic threshold or the second statistic is greater than a second statistic threshold; The output module is used to take the monitoring parameter whose contribution rate increase value is greater than a preset contribution rate threshold as an abnormal parameter.

8. An electronic 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 computer program, the method for monitoring faults of a ship power plant based on principal component analysis as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for monitoring faults of a ship power plant based on principal component analysis as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for monitoring faults of a ship power plant based on principal component analysis as described in any one of claims 1 to 6 is implemented.

Images