Engine operating state monitoring method and system based on multi-model integration metrics

Abstract

The present invention relates to the technical field of engine operating state monitoring, and provides an engine operating state monitoring method and system based on multi-model integrated metrics, including: obtaining monitoring data of the engine at different sampling frequencies at a certain sampling point; after grouping the monitoring data according to the sampling frequency and variable type, performing feature extraction to obtain a monitoring data matrix and feature data corresponding to each group; for the monitoring data matrix and feature data of each group, respectively obtaining monitoring metric values through the monitoring data model and feature data model corresponding to the group; based on each monitoring metric value, calculating the conditional probabilities of the normal state and the fault state, and combining the probabilities of the sampling point having a fault under different models, obtaining the multi-model integrated monitoring metric of the sampling point; based on the multi-model integrated monitoring metric of the sampling point, determining whether the sampling point is in an abnormal state. It can give a comprehensive evaluation of the engine operating state and improve the effectiveness of model monitoring.

Description

Technical Field

[0001] The present invention belongs to the technical field of engine operating state monitoring, and particularly relates to an engine operating state monitoring method and system based on multi-model integrated indicators. Background Technique

[0002] The statements in this part only provide background technical information related to the present invention and do not necessarily constitute prior art.

[0003] In the existing engine operating state monitoring technology, generally, the engine operating state is monitored by monitoring whether a single variable exceeds a threshold or by monitoring the spectral characteristics of vibration signals, without comprehensively using the collected data.

[0004] When the existing technology monitors the engine operating state through a single variable, the information provided by multiple engine variables is not utilized. When monitoring the engine operating state by solely monitoring the engine vibration signal, the information contained in a large number of state variables during engine operation is ignored. This results in insufficient utilization of engine operating state data and a problem that the established monitoring model has poor fault sensitivity. Summary of the Invention

[0005] In order to solve the technical problems existing in the above background technique, the present invention provides an engine operating state monitoring method and system based on multi-model integrated indicators. The multi-model integrated monitoring indicators are jointly determined by multiple engine operating state monitoring models, which can give a comprehensive evaluation of the engine operating state and improve the effectiveness of model monitoring.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] The first aspect of the present invention provides an engine operating state monitoring method based on multi-model integrated indicators, which includes:

[0008] Obtain the monitoring data of the engine at different sampling frequencies at a certain sampling point;

[0009] After grouping the monitoring data according to the sampling frequency and variable type, perform feature extraction to obtain the monitoring data matrix and feature data corresponding to each group;

[0010] For the monitoring data matrix and feature data of each group, respectively obtain the monitoring index values through the monitoring data model and feature data model corresponding to the group;

[0011] Based on each monitoring index value, calculate the conditional probabilities of the normal state and the fault state, and combine the probabilities of the sampling point failing under different models to obtain the multi-model integrated monitoring index of the sampling point;

[0012] Based on the multi-model integrated monitoring metrics of the sampling points, determine whether the sampling point is in an abnormal state.

[0013] Further, feature extraction is performed on the monitoring data matrix in the form of a sliding window.

[0014] Further, the monitoring data includes: engine operating state data and engine vibration data.

[0015] Further, the feature data includes: the mean and variance extracted from the operating state data; the time-domain features and frequency-domain features extracted from the vibration data.

[0016] Further, the multi-model integrated monitoring metric for sampling point i is:

[0017]

[0018] where N represents the number of models, is the monitoring metric value of sampling point i under the k-th model, is the data of sampling point i corresponding to the k-th model, F represents the fault state, is the conditional probability of the fault state, is the probability of a fault occurring at sampling point i under the k-th model.

[0019] Further, after normalizing the monitoring data matrix and feature data of each group, they are input into the monitoring data model and feature data model.

[0020] Further, if the square of the multi-model integrated monitoring metric of the sampling point is greater than the control limit, the sampling point is in an abnormal state.

[0021] The second aspect of the present invention provides an engine operating state monitoring system based on multi-model integrated metrics, which includes:

[0022] A data acquisition module, which is configured to: acquire monitoring data of the engine at different sampling frequencies at a certain sampling point;

[0023] A feature extraction module, which is configured to: group the monitoring data according to the sampling frequency and variable type, and then perform feature extraction to obtain the monitoring data matrix and feature data corresponding to each group;

[0024] A monitoring metric calculation module, which is configured to: for the monitoring data matrix and feature data of each group, respectively obtain the monitoring metric values through the monitoring data model and feature data model corresponding to the group;

[0025] An integrated module, which is configured to: calculate the conditional probabilities of the normal state and the fault state based on each monitoring index value, and combine the probabilities of the sampling points having faults under different models to obtain the multi-model integrated monitoring index of the sampling points;

[0026] A monitoring module, which is configured to: determine whether the sampling point is in an abnormal state based on the multi-model integrated monitoring index of the sampling point.

[0027] The third aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps in the engine operating state monitoring method based on the multi-model integrated index as described above are implemented.

[0028] The fourth aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps in the engine operating state monitoring method based on the multi-model integrated index as described above are implemented.

[0029] Compared with the prior art, the beneficial effects of the present invention are:

[0030] The engine operating state monitoring method based on the multi-model integrated index provided by the present invention respectively establishes monitoring models for different types of collected data, and respectively extracts features from different types of data and establishes monitoring models, making more full use of the state information of the engine operation contained in the collected data.

[0031] The engine operating state monitoring method based on the multi-model integrated index provided by the present invention, in order to comprehensively judge the engine operation state by multiple monitoring models, proposes a multi-model integrated monitoring index, which is jointly determined by the established multiple engine operating state monitoring models, and can give a comprehensive evaluation of the engine operation state, improving the effectiveness of model monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The specification drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention.

[0033] Figure 1 is a flowchart of the engine operating state monitoring method based on the multi-model integrated index in the first embodiment of the present invention;

[0034] Figure 2 is a schematic diagram of the monitoring data model and the feature data model in the first embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0035] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0036] It should be noted that the following detailed description is illustrative and is intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0037] Term Explanation:

[0038] Integrated Index: A monitoring index jointly constructed by multiple model monitoring indexes, used to indicate the operating state of the engine.

[0039] Control Limit: A limit value jointly determined by training data, test data, and a monitoring model. If the monitoring index exceeds this limit value, it indicates that the engine operation is abnormal.

[0040] Embodiment 1

[0041] This embodiment provides a method for monitoring the operating state of an engine based on a multi-model integrated index. The monitoring data of the engine collected is divided into data groups according to the sampling frequency and variable type; feature extraction is performed on the data of different groups respectively, and each group obtains a monitoring data matrix (the original data without feature extraction) and a feature data matrix; models are established for the obtained monitoring data matrix and feature data matrix respectively, and the control limit values are calculated; finally, an integrated monitoring index of multiple models is established based on the monitoring index and control limit value of each sample under each model, and this index is used to measure the overall operating state of the engine.

[0042] The method for monitoring the operating state of an engine based on a multi-model integrated index provided by this embodiment, as Figure 1 and Figure 2 shown, specifically includes the following steps:

[0043] Step 1: Obtain the monitoring data of the engine collected online to obtain the monitoring data of the engine at different sampling frequencies at a certain sampling point.

[0044] Step 2: Group the monitoring data according to the sampling frequency and variable type to obtain N * groups, and perform feature extraction to obtain the monitoring data matrix and feature data corresponding to each group.

[0045] For example, the data of type 1 is denoted as X 1 , that is, the monitoring data matrix of the first group, then the monitoring data matrix of the k-th group is denoted as X k , k = 1, 2,..., N * .

[0046] In this embodiment, considering the actual situation, the variables (monitoring data) are divided according to the sampling frequency and variable type, and the specific description is as follows: For example, the monitoring data of the engine at a certain sampling point includes: The engine operating state data (i.e., the process variables during engine operation) includes: fuel injection amount sampling frequency 100 Hz, engine intake air amount sampling frequency 100 Hz, accelerator pedal opening sampling frequency 50 Hz, engine speed sampling frequency 50 Hz, engine torque sampling frequency 50 Hz, etc.; The engine vibration data includes: engine vibration signal sampling frequency 1 MHz, engine vibration signal sampling frequency 5 MHz, engine vibration signal sampling frequency 10 MHz. At this time, the monitoring data can be divided into two categories: process variables and vibration signal variables according to the variable type; then, according to the sampling frequency, the monitoring data belonging to the process variables is divided into a 100 Hz variable group and a 50 Hz variable group, and the monitoring data belonging to the vibration signal variables is divided into a 1 MHz variable group, a 5 MHz variable group, and a 10 MHz variable group.

[0047] As an implementation manner, the method for feature extraction of the monitoring data matrix corresponding to a certain group is:

[0048] Through a sliding window in the form of 1 to perform feature extraction on the monitoring data matrix X 1 corresponding to a certain group, the feature matrix of the monitoring data matrix X

[0049]

[0050]

[0051]

[0052] where w in it is the width of the sliding window, which needs to be reasonably set according to the actual data; is the feature matrix of the monitoring data matrix obtained after feature extraction; n1 represents the number of data in the monitoring data matrix X 1 corresponding to a certain group, m1 represents the dimension of the monitoring data matrix X 1 n1 > n2, m1 < m2, where w represents the size of the sliding window, n2 represents the number of rows of the feature matrix, and m2 represents the dimension of the feature matrix.

[0053] Similarly, through the sliding window corresponding to the kth group in the form of k to perform feature extraction on the monitoring data matrix X k the feature matrix

[0054] Taking the variable group 1 as an example, assume that the size of the sliding window is w = 30, that is, 30 sample points are selected from the original variable matrix each time (30 rows, and each row of data represents a sample point); the original data is the matrix X 1 , and the sliding window is (where w = 30; i = 1, …, (n1 – w + 1)):

[0055]

[0056] The obtained feature matrix is as follows:

[0057]

[0058] Taking the calculation of the mean feature as an example, in Similarly, other features of the data are calculated.

[0059] Specifically, the monitoring data includes: engine operating state data and engine vibration data.

[0060] The feature data includes: mean and variance features extracted from the engine operating state data; variance, mean square value, root mean square value, skewness, kurtosis, waveform index, margin index, pulse index, peak index, and kurtosis index, etc. time domain features extracted from the engine vibration data, and average frequency, root mean square frequency, standard deviation frequency, kurtosis frequency, etc. frequency domain features.

[0061] Step 3: Standardize the monitoring data matrix and feature data of each group respectively.

[0062] Step 4: For the direct data and feature data of each group, obtain the monitoring index values respectively through the direct data model and feature data model corresponding to the group.

[0063] Specifically, for the i-th group, calculate the monitoring indexes of each model respectively to obtain where N = 2N * , N is the number of sub-monitoring models established, N * represents the number of groups, represents the vector composed of the monitoring index values of the i-th group under the k-th model; in a certain vector , the monitoring index value of the sample point i under the k-th model is

[0064] Each group (each type) corresponds to 2 multivariate statistical monitoring models, one is the monitoring data model and the other is the feature data model.

[0065] For example, the control limit of model 1 is T 1L, Similarly, the control limit of the i-th model is T iL .

[0066] Step 5: Calculate the value of the integrated index. Based on each monitoring index value, calculate the conditional probabilities of the normal state and the fault state, and combine the probabilities of the sampling point failing under different models to obtain the multi-model integrated monitoring index of the sampling point.

[0067] Assume that the data of the sample point (sampling point) i corresponding to the k-th model is The probability that the sample point i fails under the k-th model can be written in the following form:

[0068]

[0069] where, N # and F represent the normal state and the fault state respectively; and are 1-α and α respectively; α is the control limit confidence level, which is set according to the actual situation and generally can be taken as 0.9 - 0.98; the conditional probabilities of the fault state and the normal state and are defined respectively as:

[0070]

[0071]

[0072] The multi-model integrated monitoring index of the sample point i can be written in the following form:

[0073]

[0074] where, k represents the value calculated under the k-th model, k = 1,..., N, and N is the number of sub-monitoring models established.

[0075] Step 6: Based on the multi-model integrated monitoring index of the sampling point, determine whether the sampling point is in an abnormal state. Determine whether the of the i-th sampling point is greater than the control limit B L . If so, an abnormal state is detected; otherwise, return to Step 1.

[0076] where, the control limit B L of the integrated monitoring index B is determined according to the test data, and the steps to obtain the control limit B L are as follows:

[0077] (1) Obtain the original training data.

[0078] (2) Group the collected data according to the data type and sampling frequency.

[0079] (3) Obtain the data matrix and data feature matrix of different groups of data.

[0080] (4) Calculate the mean, variance, etc. of the variables in each matrix respectively, and perform standardization.

[0081] (5) Perform eigenvalue decomposition on each matrix respectively to obtain a set of spatial basis vectors.

[0082] (6) Calculate the projection values of each data matrix on its corresponding spatial basis vectors respectively, and obtain the projection value data matrix respectively.

[0083] (7) Determine the principal components of each monitoring model according to the corresponding projection value data matrix.

[0084] (8) Calculate the control limits T 1L , T 2L , …, T NL .

[0085] (9) Establish the multi-model integrated monitoring index B.

[0086] In step (5), the spatial basis vectors are obtained as follows:

[0087] For the sake of convenience of explanation, a certain obtained feature data matrix X F is used as the general processing matrix X ori ∈R n×m , where n and m respectively represent the number of samples and measurement variables in the feature data matrix X F ; X * is the standardized data matrix of X ori ; that is:

[0088]

[0089]

[0090] where, μ ori ∈R 1×m is the column mean vector of the sample matrix X ori ; Σ ori ∈R m×m is the standard deviation matrix of the sample matrix X ori ; I n =[1, 1, …, 1] ∈R n .

[0091] The covariance matrix of X * is , which is a Hermitian matrix:

[0092]

[0093] Perform singular value decomposition

[0094]

[0095] where the diagonal matrix Λ = diag(λ1, λ2, …, λ m ) is the eigenvalue, and P ∈ R m×m is the eigenvector corresponding to the eigenvalue. The matrix P composed of m eigenvectors of

[0096] is a set of basis vectors. In step (6), calculate the projection values of the data matrix on its corresponding spatial basis vectors respectively, and obtain the projection value data matrix, specifically:

[0097] The data matrix X * projects onto the basis vector P as follows:

[0098] T = X * P (12a)

[0099]

[0100] t j = X * × p j (12c)

[0101] where T ∈ R n×m is the projection of the data matrix X * on the basis vector. Each row of T represents the new coordinates of the normalized sampling points. t j is the j-th column of the matrix T, which represents the coverage range of the data matrix X * on the j-th basis vector p j .

[0102] After determining the projection transformation vector, it is necessary to perform projection conversion on the process measurement data. Construct a single-step sliding window X k ∈ R w×m to select the normalized process measurement data, where w is the window width.

[0103]

[0104] By projecting the selected data X k ∈ R w×m onto the basis vector P, the projection transformation components (PTCs) of the window data can be obtained and recorded as The projection of the window data on the basis vector can be defined as:

[0105]

[0106] Similarly, denote the coverage of the k-th window data X on the basis vectors. k .

[0107] According to the relevant properties of the matrix, are orthogonal and can be quantized and transformed into the following form:

[0108]

[0109] Take the elements on the diagonal (the projection values of the sample points on this vector dimension) as a row. As the sliding window moves, the projection value data matrix is finally obtained.

[0110] In step (7), according to the corresponding projection value data matrix, determine the principal components of each monitoring model, specifically:

[0111] For the obtained Solve its covariance matrix S and perform eigenvalue decomposition:

[0112]

[0113] where Λ is a diagonal matrix with diagonal elements being the eigenvalues λ of the matrix . Arrange its diagonal elements in descending order. At this time, the direction matrix P corresponding to the diagonal elements after the descending order arrangement (obtained by adjusting the order of the vectors in the matrix V).

[0114] Select the principal components according to the cumulative contribution rate, that is, according to the ratio of the sum of the first r diagonal elements to the sum of all diagonal elements. Generally, take the ratio between 0.8 and 0.95. The first r direction vectors p in the corresponding P matrix are the principal components (specifically refer to PCA).

[0115] In step (8), calculate the control limits T 1L , T 2L , …, T NL , specifically:

[0116] Use the T2 statistic in the PCA method to calculate each row of data in the projection value data matrix (the characteristics of the original sample data).

[0117]

[0118] Among them, denotes the i-th row data in the matrix. A set of monitoring values obtained by processing the data used for modeling by the first model through steps (1) to (8). Arrange this set of values in ascending order and take the position of the first α% values. The value at this position is denoted as T 1L, if the value is 95, that is, T 1L is greater than 95% of the normal modeling data monitoring value.

[0119] (9) Establish the multi-model integrated monitoring index B, as shown in equations (4), (5), (6), and (7).

[0120] (10) The calculation form of the integrated monitoring index B is equation (7), and the determination rule of B L is: Substitute the actually collected normal samples into the model (method) proposed in the present invention, and calculate the integrated monitoring index B of each sample i (i is the sample label), and sort the obtained B i in ascending order. Confirm how much confidence level is used to obtain the value of B according to the actual usage requirements, that is, based on the B L after ascending order and the set confidence level, obtain the value of B i . Obtain B with a confidence level of 0.95 L , which means that B L is greater than 95% of the data in the B L data sequence. i data sequence.

[0121] The engine operating state monitoring method based on the multi-model integrated index provided in this embodiment respectively establishes an original data monitoring model and a data feature monitoring model for the samples with different sampling frequencies collected, and respectively determines the control limits of the models, and integrates the monitoring results of multiple models into a monitoring index to monitor whether the engine operating state is abnormal using the integrated index.

[0122] The engine operating state monitoring method based on the multi-model integrated index provided in this embodiment proposes to establish multiple monitoring models for the engine operating state, perform modeling according to the data sampling frequency and data type, and respectively perform feature extraction on each type of data collected and establish a monitoring model. The engine operating state is monitored simultaneously through the established multiple monitoring models of the engine operating state.

[0123] The engine operating state monitoring method based on the multi-model integrated index provided in this embodiment proposes an integrated index for the multiple established engine operating state monitoring models, and this index comprehensively combines the monitoring results of the multiple established models for the engine operating state.

[0124] The engine operating state monitoring method based on multi-model integrated metrics provided in this embodiment establishes monitoring models for different types of collected data respectively, and extracts features from different types of data and establishes monitoring models respectively. It makes more full use of the state information of the engine operation contained in the collected data. In order to comprehensively judge the engine operation state by multiple monitoring models, a monitoring metric for multi-model integration is proposed. This metric is jointly determined by multiple established engine operating state monitoring models, and can give a comprehensive evaluation of the engine operating state, improving the effectiveness of model monitoring.

[0125] Embodiment 2

[0126] This embodiment provides an engine operating state monitoring system based on multi-model integrated metrics, which specifically includes:

[0127] A data acquisition module, which is configured to: acquire monitoring data of the engine at different sampling frequencies at a certain sampling point;

[0128] A feature extraction module, which is configured to: after grouping the monitoring data according to the sampling frequency and variable type, perform feature extraction to obtain a monitoring data matrix and feature data corresponding to each group;

[0129] A monitoring metric calculation module, which is configured to: for the monitoring data matrix and feature data of each group, respectively obtain monitoring metric values through the monitoring data model and feature data model corresponding to the group;

[0130] An integration module, which is configured to: based on each monitoring metric value, calculate the conditional probabilities of the normal state and the fault state, and combine the probabilities of the sampling point having a fault under different models to obtain the multi-model integrated monitoring metric of the sampling point;

[0131] A monitoring module, which is configured to: based on the multi-model integrated monitoring metric of the sampling point, judge whether the sampling point is in an abnormal state.

[0132] It should be noted here that each module in this embodiment corresponds to each step in Embodiment 1 one by one, and its specific implementation process is the same, so it will not be repeated here.

[0133] Embodiment 3

[0134] This embodiment provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the steps in the engine operating state monitoring method based on multi-model integrated metrics as described in Embodiment 1 above.

[0135] Embodiment 4

[0136] This embodiment provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps in the engine operating state monitoring method based on multi-model integration metrics as described in the above-mentioned Embodiment 1 are implemented.

[0137] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a hardware embodiment, a software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories and optical memories, etc.) containing computer-usable program code.

[0138] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0139] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0140] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0141] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), or the like.

[0142] The foregoing is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An engine operating state monitoring method based on multi-model integration metrics, characterized in that, Including: Obtain the monitoring data of the engine at different sampling frequencies at a certain sampling point; After grouping the monitoring data according to the sampling frequency and variable type, perform feature extraction to obtain the monitoring data matrix and feature data corresponding to each group; For the monitoring data matrix and feature data of each group, respectively obtain the monitoring index values through the monitoring data model and feature data model corresponding to the group; Based on each monitoring index value, calculate the conditional probabilities of the normal state and the fault state, and combine the probabilities of the sampling point having a fault under different models to obtain the multi-model integrated monitoring index of the sampling point; Based on the multi-model integrated monitoring index of the sampling point, determine whether the sampling point is in an abnormal state.

2. The engine operating state monitoring method based on multi-model integration metrics according to claim 1, wherein Perform feature extraction on the monitoring data matrix in the form of a sliding window.

3. The engine operating condition monitoring method based on multi-model integration metrics according to claim 1, characterized in that The monitoring data includes: engine operating state data and engine vibration data.

4. The engine operating state monitoring method based on multi-model integration metrics according to claim 3, wherein The feature data includes: the mean and variance extracted from the operating state data; the time-domain features and frequency-domain features extracted from the vibration data.

5. The engine operating state monitoring method based on multi-model integration metrics according to claim 1, wherein The multi-model integrated monitoring index of sampling point i is: where N represents the number of models, is the monitoring index value of sampling point i under the k-th model, is the data of sampling point i corresponding to the k-th model, and F represents the fault state, is the conditional probability of the fault state, is the probability of a fault occurring at sampling point i under the k-th model.

6. The engine operating state monitoring method based on multi-model integrated metrics according to claim 1, characterized in that, After standardizing the monitoring data matrix and feature data of each group, input them into the monitoring data model and feature data model.

7. The engine operating state monitoring method based on multi-model integration metrics according to claim 1, characterized in that If the square of the multi-model integrated monitoring index of the sampling point is greater than the control limit, the sampling point is in an abnormal state.

8. An engine operating condition monitoring system based on multi-model integration metrics, characterized in that, Including: A data acquisition module, which is configured to: obtain the monitoring data of the engine at different sampling frequencies at a certain sampling point; A feature extraction module, which is configured to: after grouping the monitoring data according to the sampling frequency and variable type, perform feature extraction to obtain the monitoring data matrix and feature data corresponding to each group; A monitoring index calculation module, which is configured to: for the monitoring data matrix and feature data of each group, respectively obtain the monitoring index values through the monitoring data model and feature data model corresponding to the group; An integration module, which is configured to: based on each monitoring index value, calculate the conditional probabilities of the normal state and the fault state, and combine the probabilities of the sampling point having a fault under different models to obtain the multi-model integrated monitoring index of the sampling point; A monitoring module, which is configured to: based on the multi-model integrated monitoring index of the sampling point, determine whether the sampling point is in an abnormal state.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by a processor, it implements the steps in the engine operating state monitoring method based on multi-model integrated indicators described in any one of claims 1-7.

10. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the engine operating state monitoring method based on multi-model integrated indicators described in any one of claims 1-7.

Images