Fault detection method and device, computer equipment, readable storage medium and program product

Abstract

The invention relates to a fault detection method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring sample monitoring data collected in a preset historical time period for entity equipment; obtaining first sample monitoring data and second sample monitoring data; determining a covariance matrix according to the first sample monitoring data and the second sample monitoring data so as to further obtain a conversion matrix; acquiring third sample monitoring data and fourth sample monitoring data acquired at the current moment; according to the conversion matrix, the third sample monitoring data and the fourth sample monitoring data, calculating a standard variable matrix, and calculating an information increment matrix and a first mean value of element values in the information increment matrix; and according to the first mean value and a preset threshold coefficient, calculating a dynamic threshold value, and according to a relationship between the dynamic threshold value and the first mean value, judging whether the physical equipment has a fault. By adopting the method, the accuracy of fault detection can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of fault detection, and in particular to a fault detection method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Art

[0002] With the development of sensor technology and computer technology, it has become possible to collect and store a large amount of data in civilian industrial processes such as chemical production and military equipment such as missile weapons. In order to ensure their safe and stable operation, experts and scholars have proposed multivariate statistical process monitoring (MSPM) technology. MSPM technology mainly includes principal component analysis, partial least squares, independent component analysis (ICA) and canonical variate analysis (CVA). Among them, the basic independent component analysis, principal component analysis and partial least squares algorithms all perform fault detection under the assumption that the process data has no dynamic characteristics, while CVA is born on the basis of optimal statistical reasoning to consider the dynamic characteristics of process data.

[0003] Although the control limits obtained by the existing technology through distribution can distinguish whether the operation is in a normal state or an abnormal state, they cannot reflect the dynamic changes of process data in real time, resulting in inaccurate fault judgment results. Summary of the invention

[0004] Based on this, it is necessary to provide a fault detection method, apparatus, computer equipment, computer-readable storage medium and computer program product that can improve the accuracy of fault detection results in response to the above technical problems.

[0005] In a first aspect, the present application provides a fault detection method, comprising:

[0006] For physical devices, obtain sample monitoring data collected within a preset historical period;

[0007] For a historical moment within the preset historical period, the sample monitoring data collected before the historical moment and within the preset historical period is used as the first sample monitoring data; the sample monitoring data collected after the historical moment and within the preset historical period is used as the second sample monitoring data;

[0008] Determine a first Hankel matrix based on the first sample monitoring data, and determine a second Hankel matrix based on the second sample monitoring data;

[0009] Based on the first Hankel matrix and the second Hankel matrix, a covariance matrix is ​​calculated, and a singular value decomposition is performed on the covariance matrix to obtain a conversion matrix;

[0010] Acquire the third sample monitoring data collected at the current moment, and the fourth sample monitoring data collected within a period corresponding to a preset time length before the current moment;

[0011] Calculating a canonical variate matrix according to the conversion matrix, the third sample monitoring data, and the fourth sample monitoring data;

[0012] Based on the canonical variable matrix and the third sample monitoring data, calculating an information increment matrix and a first mean value of element values ​​in the information increment matrix;

[0013] A dynamic threshold is calculated according to the first mean value and a preset threshold coefficient, and whether the physical device is faulty is determined according to a relationship between the dynamic threshold value and the first mean value.

[0014] In one embodiment, determining a first Hankel matrix based on the first sample monitoring data and determining a second Hankel matrix based on the second sample monitoring data includes:

[0015] performing standardization processing on the first sample monitoring data and the second sample monitoring data respectively;

[0016] Constructing a plurality of first vectors according to the standardized first sample monitoring data, and constructing a plurality of second vectors according to the standardized second sample monitoring data;

[0017] A first average vector of the plurality of first vectors and a second average vector of the plurality of second vectors are calculated, and a first Hankel matrix is ​​determined according to the plurality of first vectors and the first average vector; and a second Hankel matrix is ​​determined according to the plurality of second vectors and the second average vector.

[0018] In one embodiment, calculating a canonical variable matrix according to the conversion matrix, the third sample monitoring data and the fourth sample monitoring data comprises:

[0019] performing standardization processing on the third sample monitoring data and the fourth sample monitoring data respectively;

[0020] Constructing a plurality of third vectors according to the standardized third sample monitoring data and the standardized fourth sample monitoring data, constructing a fourth vector according to the plurality of third vectors, and calculating a third average vector of the plurality of third vectors;

[0021] The third mean vector is subtracted from the fourth vector and the resultant result is multiplied by the transformation matrix to obtain a canonical variable matrix.

[0022] In one embodiment, the calculating the information increment matrix based on the canonical variable matrix and the third sample monitoring data includes:

[0023] Based on the degree of difference between each element value in the canonical variable matrix and the average value of the element values ​​in the row to which each element value belongs, obtaining a preprocessed canonical variable matrix;

[0024] Determine a first local matrix at a current moment and a second local matrix at a moment next to the current moment according to the third sample monitoring data and the preprocessed canonical variate matrix;

[0025] An information increment matrix is ​​calculated according to the first local matrix and the second local matrix.

[0026] In one embodiment, calculating the information increment matrix according to the first local matrix and the second local matrix includes:

[0027] constructing a third local matrix according to the vectors corresponding to the overlapping moments between the first local matrix and the second local matrix, and calculating the third average vectors corresponding to the vectors in the third local matrix;

[0028] Calculate a first covariance matrix at a current moment and a second covariance matrix at a moment next to the current moment according to the first local matrix, the second local matrix and the third average vector;

[0029] The first covariance matrix is ​​subtracted from the second covariance matrix to obtain an information increment matrix.

[0030] In one embodiment, judging whether the physical device is faulty according to the relationship between the dynamic threshold and the first mean value includes:

[0031] When the dynamic threshold is smaller than the first average value, the physical device is in a fault state;

[0032] When the dynamic thresholds are all greater than the first average value, the physical device is in a normal state.

[0033] In a second aspect, the present application also provides a fault detection device, comprising:

[0034] An acquisition module is used to acquire sample monitoring data collected within a preset historical period for a physical device;

[0035] The acquisition module is further configured to, for a historical moment within the preset historical period, use the sample monitoring data collected before the historical moment and within the preset historical period as the first sample monitoring data; and use the sample monitoring data collected after the historical moment and within the preset historical period as the second sample monitoring data;

[0036] A determination module, configured to determine a first Hankel matrix based on the first sample monitoring data, and to determine a second Hankel matrix based on the second sample monitoring data;

[0037] The determining module is further used to calculate a covariance matrix based on the first Hankel matrix and the second Hankel matrix, and perform singular value decomposition on the covariance matrix to obtain a conversion matrix;

[0038] The acquisition module is further used to acquire the third sample monitoring data collected at the current moment and the fourth sample monitoring data collected within a period corresponding to a preset time length before the current moment;

[0039] A calculation module, used for calculating a canonical variable matrix according to the conversion matrix, the third sample monitoring data and the fourth sample monitoring data;

[0040] The calculation module is further used to calculate the information increment matrix and the first mean of the element values ​​in the information increment matrix based on the standard variable matrix and the third sample monitoring data;

[0041] The judgment module is used to calculate a dynamic threshold value according to the first mean value and a preset threshold coefficient, and judge whether the physical device is faulty according to the relationship between the dynamic threshold value and the first mean value.

[0042] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0043] For physical devices, obtain sample monitoring data collected within a preset historical period;

[0044] For a historical moment within the preset historical period, the sample monitoring data collected before the historical moment and within the preset historical period is used as the first sample monitoring data; the sample monitoring data collected after the historical moment and within the preset historical period is used as the second sample monitoring data;

[0045] Determine a first Hankel matrix based on the first sample monitoring data, and determine a second Hankel matrix based on the second sample monitoring data;

[0046] Based on the first Hankel matrix and the second Hankel matrix, a covariance matrix is ​​calculated, and a singular value decomposition is performed on the covariance matrix to obtain a conversion matrix;

[0047] Acquire the third sample monitoring data collected at the current moment, and the fourth sample monitoring data collected within a period corresponding to a preset time length before the current moment;

[0048] Calculating a canonical variate matrix according to the conversion matrix, the third sample monitoring data, and the fourth sample monitoring data;

[0049] Based on the canonical variable matrix and the third sample monitoring data, calculating an information increment matrix and a first mean value of element values ​​in the information increment matrix;

[0050] A dynamic threshold is calculated according to the first mean value and a preset threshold coefficient, and whether the physical device is faulty is determined according to a relationship between the dynamic threshold value and the first mean value.

[0051] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0052] For physical devices, obtain sample monitoring data collected within a preset historical period;

[0053] For a historical moment within the preset historical period, the sample monitoring data collected before the historical moment and within the preset historical period is used as the first sample monitoring data; the sample monitoring data collected after the historical moment and within the preset historical period is used as the second sample monitoring data;

[0054] Determine a first Hankel matrix based on the first sample monitoring data, and determine a second Hankel matrix based on the second sample monitoring data;

[0055] Based on the first Hankel matrix and the second Hankel matrix, a covariance matrix is ​​calculated, and a singular value decomposition is performed on the covariance matrix to obtain a conversion matrix;

[0056] Acquire the third sample monitoring data collected at the current moment, and the fourth sample monitoring data collected within a period corresponding to a preset time length before the current moment;

[0057] Calculating a canonical variate matrix according to the conversion matrix, the third sample monitoring data, and the fourth sample monitoring data;

[0058] Based on the canonical variable matrix and the third sample monitoring data, calculating an information increment matrix and a first mean value of element values ​​in the information increment matrix;

[0059] A dynamic threshold is calculated according to the first mean value and a preset threshold coefficient, and whether the physical device is faulty is determined according to a relationship between the dynamic threshold value and the first mean value.

[0060] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0061] For physical devices, obtain sample monitoring data collected within a preset historical period;

[0062] For a historical moment within the preset historical period, the sample monitoring data collected before the historical moment and within the preset historical period is used as the first sample monitoring data; the sample monitoring data collected after the historical moment and within the preset historical period is used as the second sample monitoring data;

[0063] Determine a first Hankel matrix based on the first sample monitoring data, and determine a second Hankel matrix based on the second sample monitoring data;

[0064] Based on the first Hankel matrix and the second Hankel matrix, a covariance matrix is ​​calculated, and a singular value decomposition is performed on the covariance matrix to obtain a conversion matrix;

[0065] Acquire the third sample monitoring data collected at the current moment, and the fourth sample monitoring data collected within a period corresponding to a preset time length before the current moment;

[0066] Calculating a canonical variate matrix according to the conversion matrix, the third sample monitoring data, and the fourth sample monitoring data;

[0067] Based on the canonical variable matrix and the third sample monitoring data, calculating an information increment matrix and a first mean value of element values ​​in the information increment matrix;

[0068] A dynamic threshold is calculated according to the first mean value and a preset threshold coefficient, and whether the physical device is faulty is determined according to a relationship between the dynamic threshold value and the first mean value.

[0069] The above fault detection method, device, computer equipment, computer-readable storage medium and computer program product first calculate the canonical variable matrix of the third monitoring data and the fourth monitoring data of the real-time sample, and then use the canonical variable matrix as input to introduce the information increment matrix of the improved covariance matrix. Then calculate the mean of the matrix, and set the dynamic threshold according to the state of the previous sample (normal or faulty). Finally, the current working state is judged by comparing the size of the mean and the dynamic threshold, and the dynamic changes of the process data are combined to make the fault detection result more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0071] Figure 1 An application environment diagram of a fault detection method in an embodiment;

[0072] Figure 2 is a flow chart of a fault detection method in one embodiment;

[0073] Figure 3 is a flow chart of a fault detection method in another embodiment;

[0074] Figure 4 is a structural block diagram of a fault detection device in one embodiment;

[0075] Figure 5 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0076] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0077] The fault detection method provided in the embodiment of the present application can be applied to Figure 1In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the sample monitoring data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. Among them, the terminal 102 can be, but is not limited to, various personal computers, laptops, smart phones, tablets, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart car devices, projection devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The head-mounted device can be a virtual reality (VR) device, an augmented reality (AR) device, smart glasses, etc. The server 104 can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services.

[0078] In an exemplary embodiment, Figure 2 As shown, a fault detection method is provided, which is applied to Figure 1 The terminal 102 in the example is used as an example to illustrate, including the following steps 202 to 216. Among them:

[0079] Step 202: acquiring sample monitoring data collected within a preset historical period for a physical device.

[0080] Among them, the physical equipment can be physical equipment used in civil industrial processes such as chemical production.

[0081] In some embodiments, sample monitoring data of the physical device is collected within a preset historical period, and the multivariate state space model of the physical device is as shown in formula (1) and formula (2).

[0082]

[0083]

[0084] Among them, the data matrix X∈ , A is the state matrix, C is the output matrix, x(k) is the state of the physical device at time k, y(k) is the output of the physical device at time k, w(k) and e(k) are modeling errors that obey non-Gaussian distribution.

[0085] Step 204, for a historical moment within a preset historical period, the sample monitoring data collected before the historical moment and within the preset historical period is used as the first sample monitoring data; the sample monitoring data collected after the historical moment and within the preset historical period is used as the second sample monitoring data.

[0086] In some embodiments, a historical moment is set in a preset historical period, and q sample monitoring data y(k) collected before the historical moment and within the preset historical period are used as first sample monitoring data; and q sample monitoring data y(k) collected after the historical moment and within the preset historical period are used as second sample monitoring data.

[0087] Step 206: determine a first Hankel matrix based on the first sample monitoring data, and determine a second Hankel matrix based on the second sample monitoring data.

[0088] The Hankel matrix is ​​a matrix in which the elements on each sub-diagonal are equal.

[0089] In some embodiments, the first sample monitoring data and the second sample monitoring data are standardized respectively; a plurality of first vectors are constructed according to the standardized first sample monitoring data, and a plurality of second vectors are constructed according to the standardized second sample monitoring data; a first average vector of the plurality of first vectors and a second average vector of the plurality of second vectors are calculated, and a first Hankel matrix is ​​determined according to the plurality of first vectors and the first average vector. ; Determine the second Hankel matrix based on the multiple second vectors and the second average vector .

[0090] Step 208: Calculate a covariance matrix based on the first Hankel matrix and the second Hankel matrix, and perform singular value decomposition on the covariance matrix to obtain a conversion matrix.

[0091] Among them, the covariance matrix is ​​a matrix that describes the relationship between multidimensional data features; singular value decomposition is to decompose a matrix into the product of an orthogonal matrix, a diagonal matrix and an orthogonal matrix.

[0092] In some embodiments, according to the first Hankel matrix and the second Hankel matrix , respectively calculate the covariance matrix and the cross-covariance matrix. The specific calculation formulas are shown in formula (3), formula (4) and formula (5).

[0093]

[0094]

[0095]

[0096] Where, N=n-2q+1, n is the number of samples, for The transposed matrix of for The transposed matrix of .

[0097] The covariance matrix is ​​decomposed by singular values, and the specific calculation is shown in formula (6). The canonical variable matrix of historical data is obtained, and the specific calculation is shown in formula (7). Then, the transformation matrix is ​​further obtained, and the specific calculation is shown in formula (8).

[0098]

[0099]

[0100]

[0101] in, , , is the transposed matrix of V.

[0102] Step 210, obtaining the third sample monitoring data collected at the current moment and the fourth sample monitoring data collected within a period corresponding to a preset time length before the current moment.

[0103] In some embodiments, the third sample monitoring data at the current moment and the fourth sample monitoring data within a time period corresponding to a preset time length before the current moment are collected.

[0104] Step 212, calculating a canonical variate matrix according to the conversion matrix, the third sample monitoring data and the fourth sample monitoring data.

[0105] In some embodiments, the third sample monitoring data and the fourth sample monitoring data are standardized, respectively; multiple third vectors are constructed based on the standardized third sample monitoring data and the standardized fourth sample monitoring data, and a fourth vector is constructed based on the multiple third vectors, and a third average vector of the multiple third vectors is calculated; the third average vector is subtracted from the fourth vector, and the result is multiplied by the transformation matrix to obtain a normalized variable matrix.

[0106] Step 214, based on the canonical variable matrix and the third sample monitoring data, calculate the information increment matrix and the first mean of the element values ​​in the information increment matrix.

[0107] In some embodiments, based on the degree of difference between each element value in the canonical variable matrix and the average value of the element values ​​in the row to which each element value belongs, a preprocessed canonical variable matrix is ​​obtained; based on the third sample monitoring data and the preprocessed canonical variable matrix, a first local matrix at the current moment and a second local matrix at the next moment after the current moment are determined.

[0108] According to the vectors corresponding to the overlapping moments between the first local matrix and the second local matrix, a third local matrix is ​​constructed, and the third mean vector corresponding to the vectors in the third local matrix is ​​calculated; according to the first local matrix, the second local matrix and the third mean vector, the first covariance matrix at the current moment and the second covariance matrix at the next moment of the current moment are calculated; the first covariance matrix is ​​subtracted from the second covariance matrix to obtain the information increment matrix, and the first mean of the elements in the information increment matrix is ​​calculated. The specific calculation formula of the first mean is shown in (9).

[0109]

[0110] in, is the information increment matrix.

[0111] Step 216, calculating a dynamic threshold value according to the first mean value and a preset threshold coefficient, and determining whether the physical device is faulty according to the relationship between the dynamic threshold value and the first mean value.

[0112] In some embodiments, a dynamic threshold is calculated based on a first mean value and a preset threshold coefficient. The specific calculation formulas are shown in (10) and (11). When a dynamic threshold value is less than the first mean value, the physical device is in a fault state; when the dynamic threshold values ​​are all greater than the first mean value, the physical device is in a normal state.

[0113]

[0114]

[0115] in, For different threshold coefficients, generally ; Represents the first mean of the elements in the information increment matrix at time s.

[0116] In the above fault detection method, the canonical variable matrix of the third monitoring data and the fourth monitoring data of the real-time sample is calculated, and then the canonical variable matrix is ​​used as input to introduce the information increment matrix of the improved covariance matrix. Then the mean of the matrix is ​​calculated, and the dynamic threshold is set according to the state of the previous sample (normal or faulty). Finally, the current working state is judged by comparing the size of the mean and the dynamic threshold, and the dynamic changes of the process data are combined to make the fault detection result more accurate.

[0117] In an exemplary embodiment, a first Hankel matrix is ​​determined based on first sample monitoring data, and a second Hankel matrix is ​​determined based on second sample monitoring data, including: standardizing the first sample monitoring data and the second sample monitoring data, respectively; constructing multiple first vectors based on the standardized first sample monitoring data, and constructing multiple second vectors based on the standardized second sample monitoring data; calculating a first average vector of the multiple first vectors and a second average vector of the multiple second vectors, and determining a first Hankel matrix based on the multiple first vectors and the first average vector; and determining a second Hankel matrix based on the multiple second vectors and the second average vector.

[0118] In some embodiments, the first sample monitoring data and the second sample monitoring data are standardized respectively, and a plurality of first vectors are constructed according to the standardized first sample monitoring data, and a plurality of second vectors are constructed according to the standardized second sample monitoring data. The plurality of first vectors constitute The specific expression is as shown in formula (12), multiple second vectors constitute The specific expression is as shown in formula (13), calculating the first average vector of multiple first vectors and the second average vector of the plurality of second vectors , subtract the first average vector from each first vector to form an element in the first Hankel matrix. The specific calculation formula is shown in (14). Subtract the second average vector from each second vector to form an element in the second Hankel matrix. The specific calculation formula is shown in (15).

[0119]

[0120]

[0121]

[0122]

[0123] Comprehensive first Hankel matrix The elements in and the second Hankel matrix Each element in gets the first Hankel matrix and the second Hankel matrix , the specific calculation formulas are shown in formula (16) and formula (17).

[0124]

[0125]

[0126] In the above embodiment, the sample monitoring data is standardized to avoid the influence of data of different units and eliminate the dimensional effect of multiple variables. The conversion matrix is ​​obtained by calculating the Hankel matrix of the past sample monitoring data, so that the result of judgment using the conversion matrix is ​​more accurate.

[0127] In some exemplary embodiments, a canonical variable matrix is ​​calculated based on a transformation matrix, third sample monitoring data, and fourth sample monitoring data, including: standardizing the third sample monitoring data and the fourth sample monitoring data, respectively; constructing a plurality of third vectors based on the standardized third sample monitoring data and the standardized fourth sample monitoring data, and constructing a fourth vector based on the plurality of third vectors, and calculating a third average vector of the plurality of third vectors; and subtracting the third average vector from the fourth vector and multiplying the result by the transformation matrix to obtain a canonical variable matrix.

[0128] In some embodiments, the third sample monitoring data and the fourth sample monitoring data are standardized, a plurality of third vectors are constructed according to the standardized third sample monitoring data and the standardized fourth sample monitoring data, and a fourth vector is constructed according to the plurality of third vectors. , calculate the third average vector of multiple third vectors , subtract the third mean vector from the fourth vector and multiply it by the transformation matrix to get the canonical variable matrix , the specific calculation formula is shown in (18).

[0129]

[0130] in, The fourth vector represents a total of q third sample monitoring data and fourth sample monitoring data collected within a period corresponding to a preset time length from the current moment, The canonical variate matrix representing the data at the current moment.

[0131] In the above embodiment, the canonical variable matrix is ​​obtained by combining the monitoring data at the current moment with the conversion matrix obtained from the historical data, so that the canonical variable matrix is ​​more in line with the actual situation and the final result is more accurate.

[0132] In some exemplary embodiments, an information increment matrix is ​​calculated based on a canonical variable matrix and third sample monitoring data, including: obtaining a preprocessed canonical variable matrix based on the degree of difference between each element value in the canonical variable matrix and the average value of the element values ​​in the row to which each element value belongs; determining a first local matrix at a current moment and a second local matrix at a moment next to the current moment based on the third sample monitoring data and the preprocessed canonical variable matrix; and calculating an information increment matrix based on the first local matrix and the second local matrix.

[0133] In actual implementation, For an m*n matrix, get The mean vector of The specific calculation formula is shown in (19). According to the average value of each element in the standard variable matrix and the average value of the element in the row to which each element belongs, that is, the mean vector The degree of difference between them is determined by preprocessing the canonical variable matrix. The specific preprocessing formula is shown in (20).

[0134]

[0135]

[0136] in, , for The transpose of is the preprocessed canonical variate matrix.

[0137] Select monitoring data of L normal samples with a duration of L from the third sample monitoring data, and obtain the first local matrix at the current moment and the second local matrix at the next moment of the current moment. The specific expression formulas are shown in (21) and (22).

[0138]

[0139]

[0140] in, Represents the first local matrix at the current moment, represent The monitoring data collected at all times, Represents the second local matrix at the next moment of the current moment.

[0141] Furthermore, an information increment matrix is ​​calculated based on the first local matrix and the second local matrix.

[0142] In the above embodiment, data from multiple samples and multiple moments are fully considered to calculate the local matrix, so that the final result is more accurate.

[0143] In some exemplary embodiments, an information increment matrix is ​​calculated based on a first local matrix and a second local matrix, including: constructing a third local matrix based on vectors corresponding to overlapping moments between the first local matrix and the second local matrix, and calculating a third average vector corresponding to the vectors in the third local matrix; calculating a first covariance matrix at a current moment and a second covariance matrix at a moment next to the current moment based on the first local matrix, the second local matrix and the third average vector; and subtracting the first covariance matrix from the second covariance matrix to obtain an information increment matrix.

[0144] In actual implementation, the first local matrix is ​​compared with the second local matrix, and the third local matrix is ​​constructed according to the vector corresponding to the overlapping moment between the two. , calculate the third average vector corresponding to the vector in the third local matrix. The specific calculation formula is shown in (23).

[0145]

[0146] in, .

[0147] According to the first local matrix, the second local matrix and the third average vector, the first covariance matrix at the current moment and the second covariance matrix at the next moment of the current moment are calculated. The specific calculation formulas are shown in (24) and (25).

[0148]

[0149]

[0150] in, is the first covariance matrix at the current moment, is the second covariance matrix of the next moment from the current moment, is the mean vector at the current moment, which is specifically expressed as , similarly, we can get the mean vector of the next moment at the current moment , for the convenience of calculation, , .

[0151] Subtract the first covariance matrix from the second covariance matrix to obtain the information increment matrix , the specific calculation formula is shown in (26).

[0152]

[0153] In the above embodiment, the information increment matrix of the monitoring data is obtained by calculation, and a large amount of information in the monitoring data is presented in the form of a matrix, so that subsequent processing is simpler and more efficient.

[0154] In some exemplary embodiments, whether the physical device is faulty is determined based on the relationship between the dynamic threshold and the first mean, including: when there is a dynamic threshold less than the first mean, the physical device is in a faulty state; when the dynamic thresholds are all greater than the first mean, the physical device is in a normal state.

[0155] In actual implementation, when , When there is a dynamic threshold value less than the first mean value, the physical device is in a fault state. , When all dynamic means in are greater than the first mean, the physical device is in a normal state.

[0156] In the above embodiment, the state of the entity device is determined by using multiple dynamic thresholds, so that the final determination result is more accurate and the occurrence of erroneous determination is reduced.

[0157] To explain the fault detection method in the present application in detail, an embodiment is used for explanation below. The specific flow chart is shown in FIG3. By way of example, the present application is explained with a fault detection method in a specific scenario.

[0158] First, sample monitoring data of the physical device is collected within a preset historical period, and the multivariate state space model of the physical device is shown in formula (1) and formula (2). A historical moment is set in the preset historical period, and the q sample monitoring data y(k) collected before the historical moment and within the preset historical period are used as the first sample monitoring data; the q sample monitoring data y(k) collected after the historical moment and within the preset historical period are used as the second sample monitoring data.

[0159] The first sample monitoring data and the second sample monitoring data are standardized respectively, and a plurality of first vectors are constructed according to the standardized first sample monitoring data, and a plurality of second vectors are constructed according to the standardized second sample monitoring data. The plurality of first vectors constitute The specific expression is as shown in formula (12), multiple second vectors constitute The specific expression is as shown in formula (13), calculating the first average vector of multiple first vectors and the second average vector of the plurality of second vectors , subtract the first average vector from each first vector to form an element in the first Hankel matrix. The specific calculation formula is shown in (14). Subtract the second average vector from each second vector to form an element in the second Hankel matrix. The specific calculation formula is shown in (15).

[0160] Comprehensive first Hankel matrix The elements in and the second Hankel matrix Each element in gets the first Hankel matrix and the second Hankel matrix , the specific calculation formulas are shown in formula (16) and formula (17).

[0161] According to the first Hankel matrix and the second Hankel matrix , respectively calculate the covariance matrix and the cross-covariance matrix, the specific calculation formulas are shown in formula (3), formula (4) and formula (5). Decompose the covariance matrix through singular values, the specific calculation is shown in formula (6), and obtain the canonical variable matrix of historical data, the specific calculation is shown in formula (7), and then further obtain the transformation matrix, the specific calculation is shown in formula (8).

[0162] The third sample monitoring data and the fourth sample monitoring data are standardized, a plurality of third vectors are constructed according to the standardized third sample monitoring data and the standardized fourth sample monitoring data, and a fourth vector is constructed according to the plurality of third vectors. , calculate the third average vector of multiple third vectors , subtract the third mean vector from the fourth vector and multiply it by the transformation matrix to get the canonical variable matrix , the specific calculation formula is shown in (18).

[0163] make For an m*n matrix, get The mean vector of The specific calculation formula is shown in (19). According to the average value of each element in the standard variable matrix and the average value of the element in the row to which each element belongs, that is, the mean vector The degree of difference between them is used to preprocess the canonical variable matrix. The specific preprocessing formula is shown in (20). In the third sample monitoring data, the monitoring data of L normal samples with a duration of L are selected to obtain the first local matrix at the current moment and the second local matrix at the next moment of the current moment. The specific expression formulas are shown in (21) and (22). The first local matrix and the second local matrix are compared, and the third local matrix is ​​constructed according to the vector corresponding to the overlapping moment between the two. , calculate the third average vector corresponding to the vector in the third local matrix. The specific calculation formula is shown in (23).

[0164] According to the first local matrix, the second local matrix and the third average vector, the first covariance matrix at the current moment and the second covariance matrix at the next moment of the current moment are calculated. The specific calculation formulas are shown in (24) and (25). The information increment matrix is ​​obtained by subtracting the first covariance matrix from the second covariance matrix. , the specific calculation formula is shown in (26), and the first mean of the elements in the information increment matrix is ​​calculated. The specific calculation formula of the first mean is shown in (9).

[0165] The dynamic threshold is calculated based on the first mean value and the preset threshold coefficient. The specific calculation formulas are shown in (10) and (11). , When there is a dynamic threshold value less than the first mean value, the physical device is in a fault state. , When all dynamic means in are greater than the first mean, the physical device is in a normal state.

[0166] Based on the existing fault detection methods, this application fully considers the shortcomings of obtaining statistical control limits through distribution, and realizes a more accurate fault detection method that can reflect real-time changes in data from the perspective of real-time reflection of dynamic changes in process data, thereby improving the accuracy of fault detection.

[0167] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0168] Based on the same inventive concept, the embodiment of the present application also provides a fault detection device for implementing the fault detection method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more fault detection device embodiments provided below can refer to the limitations on the fault detection method above, and will not be repeated here.

[0169] In an exemplary embodiment, Figure 4 As shown, a fault detection device is provided, including: an acquisition module 401, a determination module 402, a calculation module 403 and a judgment module 404, wherein:

[0170] The acquisition module is used to acquire sample monitoring data collected within a preset historical period for a physical device.

[0171] The acquisition module is also used to use, for a historical moment within the preset historical period, the sample monitoring data collected before the historical moment and within the preset historical period as the first sample monitoring data; and to use the sample monitoring data collected after the historical moment and within the preset historical period as the second sample monitoring data.

[0172] A determination module is used to determine a first Hankel matrix based on the first sample monitoring data, and to determine a second Hankel matrix based on the second sample monitoring data.

[0173] The determination module is further used to calculate a covariance matrix based on the first Hankel matrix and the second Hankel matrix, and perform singular value decomposition on the covariance matrix to obtain a transformation matrix.

[0174] The acquisition module is further used to acquire the third sample monitoring data collected at the current moment and the fourth sample monitoring data collected within a period corresponding to a preset time length before the current moment.

[0175] A calculation module is used to calculate a canonical variable matrix according to the conversion matrix, the third sample monitoring data and the fourth sample monitoring data.

[0176] The calculation module is further used to calculate the information increment matrix and the first mean of the element values ​​in the information increment matrix based on the standard variable matrix and the third sample monitoring data.

[0177] The judgment module is used to calculate a dynamic threshold value according to the first mean value and a preset threshold coefficient, and judge whether the physical device is faulty according to the relationship between the dynamic threshold value and the first mean value.

[0178] In some embodiments, the above-mentioned determination module is also used to perform standardization processing on the first sample monitoring data and the second sample monitoring data, respectively; construct multiple first vectors based on the standardized first sample monitoring data, and construct multiple second vectors based on the standardized second sample monitoring data; calculate the first average vector of the multiple first vectors and the second average vector of the multiple second vectors, and determine the first Hankel matrix based on the multiple first vectors and the first average vector; determine the second Hankel matrix based on the multiple second vectors and the second average vector.

[0179] In some embodiments, the above-mentioned calculation module is also used to perform standardization processing on the third sample monitoring data and the fourth sample monitoring data, respectively; construct multiple third vectors based on the standardized third sample monitoring data and the standardized fourth sample monitoring data, and construct a fourth vector based on the multiple third vectors, and calculate the third average vector of the multiple third vectors; subtract the third average vector from the fourth vector, multiply it by the transformation matrix, and obtain the normalized variable matrix.

[0180] In some embodiments, the above-mentioned calculation module is also used to obtain a preprocessed canonical variable matrix based on the degree of difference between each element value in the canonical variable matrix and the average value of the element values ​​in the row to which each element value belongs; determine the first local matrix at the current moment and the second local matrix at the next moment after the current moment according to the third sample monitoring data and the preprocessed canonical variable matrix; and calculate the information increment matrix according to the first local matrix and the second local matrix.

[0181] In some embodiments, the above-mentioned calculation module is also used to construct a third local matrix based on the vectors corresponding to the overlapping moments between the first local matrix and the second local matrix, and calculate the third average vector corresponding to the vectors in the third local matrix; calculate the first covariance matrix at the current moment and the second covariance matrix at the next moment of the current moment based on the first local matrix, the second local matrix and the third average vector; subtract the first covariance matrix from the second covariance matrix to obtain the information increment matrix.

[0182] In some embodiments, the above-mentioned judgment module is also used to judge that when there is a dynamic threshold value less than the first mean value, the physical device is in a fault state; when the dynamic threshold values ​​are all greater than the first mean value, the physical device is in a normal state.

[0183] Each module in the above fault detection device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0184] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Figure 5 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface, a display unit and an input device. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store sample monitoring data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a fault detection method is implemented.

[0185] The display unit of the computer device is used to form a visually visible picture, and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a key, trackball, or touchpad provided on the computer device housing, or an external keyboard, touchpad, or mouse, etc.

[0186] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0187] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0188] For physical devices, obtain sample monitoring data collected within a preset historical period;

[0189] For a historical moment within the preset historical period, the sample monitoring data collected before the historical moment and within the preset historical period is used as the first sample monitoring data; the sample monitoring data collected after the historical moment and within the preset historical period is used as the second sample monitoring data;

[0190] Determine a first Hankel matrix based on the first sample monitoring data, and determine a second Hankel matrix based on the second sample monitoring data;

[0191] Based on the first Hankel matrix and the second Hankel matrix, a covariance matrix is ​​calculated, and a singular value decomposition is performed on the covariance matrix to obtain a conversion matrix;

[0192] Acquire the third sample monitoring data collected at the current moment, and the fourth sample monitoring data collected within a period corresponding to a preset time length before the current moment;

[0193] Calculating a canonical variate matrix according to the conversion matrix, the third sample monitoring data, and the fourth sample monitoring data;

[0194] Based on the canonical variable matrix and the third sample monitoring data, calculating an information increment matrix and a first mean value of element values ​​in the information increment matrix;

[0195] A dynamic threshold is calculated according to the first mean value and a preset threshold coefficient, and whether the physical device is faulty is determined according to a relationship between the dynamic threshold value and the first mean value.

[0196] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0197] For physical devices, obtain sample monitoring data collected within a preset historical period;

[0198] For a historical moment within the preset historical period, the sample monitoring data collected before the historical moment and within the preset historical period is used as the first sample monitoring data; the sample monitoring data collected after the historical moment and within the preset historical period is used as the second sample monitoring data;

[0199] Determine a first Hankel matrix based on the first sample monitoring data, and determine a second Hankel matrix based on the second sample monitoring data;

[0200] Based on the first Hankel matrix and the second Hankel matrix, a covariance matrix is ​​calculated, and a singular value decomposition is performed on the covariance matrix to obtain a conversion matrix;

[0201] Acquire the third sample monitoring data collected at the current moment, and the fourth sample monitoring data collected within a period corresponding to a preset time length before the current moment;

[0202] Calculating a canonical variate matrix according to the conversion matrix, the third sample monitoring data, and the fourth sample monitoring data;

[0203] Based on the canonical variable matrix and the third sample monitoring data, calculating an information increment matrix and a first mean value of element values ​​in the information increment matrix;

[0204] A dynamic threshold is calculated according to the first mean value and a preset threshold coefficient, and whether the physical device is faulty is determined according to a relationship between the dynamic threshold value and the first mean value.

[0205] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0206] For physical devices, obtain sample monitoring data collected within a preset historical period;

[0207] For a historical moment within the preset historical period, the sample monitoring data collected before the historical moment and within the preset historical period is used as the first sample monitoring data; the sample monitoring data collected after the historical moment and within the preset historical period is used as the second sample monitoring data;

[0208] Determine a first Hankel matrix based on the first sample monitoring data, and determine a second Hankel matrix based on the second sample monitoring data;

[0209] Based on the first Hankel matrix and the second Hankel matrix, a covariance matrix is ​​calculated, and a singular value decomposition is performed on the covariance matrix to obtain a conversion matrix;

[0210] Acquire the third sample monitoring data collected at the current moment, and the fourth sample monitoring data collected within a period corresponding to a preset time length before the current moment;

[0211] Calculating a canonical variate matrix according to the conversion matrix, the third sample monitoring data, and the fourth sample monitoring data;

[0212] Based on the canonical variable matrix and the third sample monitoring data, calculating an information increment matrix and a first mean value of element values ​​in the information increment matrix;

[0213] A dynamic threshold is calculated according to the first mean value and a preset threshold coefficient, and whether the physical device is faulty is determined according to a relationship between the dynamic threshold value and the first mean value.

[0214] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0215] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0216] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0217] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A fault detection method, characterized in that: The method comprises: For physical devices, obtain sample monitoring data collected within a preset historical period; For a historical moment within the preset historical period, the sample monitoring data collected before the historical moment and within the preset historical period is used as the first sample monitoring data; the sample monitoring data collected after the historical moment and within the preset historical period is used as the second sample monitoring data; Determine a first Hankel matrix based on the first sample monitoring data, and determine a second Hankel matrix based on the second sample monitoring data; Based on the first Hankel matrix and the second Hankel matrix, a covariance matrix is ​​calculated, and a singular value decomposition is performed on the covariance matrix to obtain a conversion matrix; Acquire the third sample monitoring data collected at the current moment, and the fourth sample monitoring data collected within a period corresponding to a preset time length before the current moment; Calculating a canonical variate matrix according to the conversion matrix, the third sample monitoring data, and the fourth sample monitoring data; Based on the canonical variable matrix and the third sample monitoring data, calculating an information increment matrix and a first mean value of element values ​​in the information increment matrix; A dynamic threshold is calculated according to the first mean value and a preset threshold coefficient, and whether the physical device is faulty is determined according to a relationship between the dynamic threshold value and the first mean value.

2. The method according to claim 1, characterized in that: The determining a first Hankel matrix based on the first sample monitoring data and determining a second Hankel matrix based on the second sample monitoring data includes: performing standardization processing on the first sample monitoring data and the second sample monitoring data respectively; Constructing a plurality of first vectors according to the standardized first sample monitoring data, and constructing a plurality of second vectors according to the standardized second sample monitoring data; A first average vector of the plurality of first vectors and a second average vector of the plurality of second vectors are calculated, and a first Hankel matrix is ​​determined according to the plurality of first vectors and the first average vector; and a second Hankel matrix is ​​determined according to the plurality of second vectors and the second average vector.

3. The method according to claim 1, characterized in that The calculating the canonical variable matrix according to the conversion matrix, the third sample monitoring data and the fourth sample monitoring data comprises: performing standardization processing on the third sample monitoring data and the fourth sample monitoring data respectively; Constructing a plurality of third vectors according to the standardized third sample monitoring data and the standardized fourth sample monitoring data, constructing a fourth vector according to the plurality of third vectors, and calculating a third average vector of the plurality of third vectors; The third mean vector is subtracted from the fourth vector and the resultant result is multiplied by the transformation matrix to obtain a canonical variable matrix.

4. The method according to claim 1, characterized in that The calculating the information increment matrix based on the canonical variable matrix and the third sample monitoring data includes: Based on the degree of difference between each element value in the canonical variable matrix and the average value of the element values ​​in the row to which each element value belongs, obtaining a preprocessed canonical variable matrix; Determine a first local matrix at a current moment and a second local matrix at a moment next to the current moment according to the third sample monitoring data and the preprocessed canonical variate matrix; An information increment matrix is ​​calculated according to the first local matrix and the second local matrix.

5. The method according to claim 4, characterized in that The step of calculating the information increment matrix according to the first local matrix and the second local matrix includes: constructing a third local matrix according to the vectors corresponding to the overlapping moments between the first local matrix and the second local matrix, and calculating the third average vectors corresponding to the vectors in the third local matrix; Calculate a first covariance matrix at a current moment and a second covariance matrix at a moment next to the current moment according to the first local matrix, the second local matrix and the third average vector; The first covariance matrix is ​​subtracted from the second covariance matrix to obtain an information increment matrix.

6. The method according to claim 1, characterized in that The determining whether the physical device is faulty according to the relationship between the dynamic threshold and the first mean value includes: When the dynamic threshold is smaller than the first average value, the physical device is in a fault state; When the dynamic thresholds are all greater than the first average value, the physical device is in a normal state.

7. A fault detection device, characterized in that: The device comprises: An acquisition module is used to acquire sample monitoring data collected within a preset historical period for a physical device; The acquisition module is further configured to, for a historical moment within the preset historical period, use the sample monitoring data collected before the historical moment and within the preset historical period as the first sample monitoring data; and use the sample monitoring data collected after the historical moment and within the preset historical period as the second sample monitoring data; A determination module, configured to determine a first Hankel matrix based on the first sample monitoring data, and to determine a second Hankel matrix based on the second sample monitoring data; The determining module is further used to calculate a covariance matrix based on the first Hankel matrix and the second Hankel matrix, and perform singular value decomposition on the covariance matrix to obtain a conversion matrix; The acquisition module is further used to acquire the third sample monitoring data collected at the current moment and the fourth sample monitoring data collected within a period corresponding to a preset time length before the current moment; A calculation module, used for calculating a canonical variable matrix according to the conversion matrix, the third sample monitoring data and the fourth sample monitoring data; The calculation module is further used to calculate the information increment matrix and the first mean of the element values ​​in the information increment matrix based on the standard variable matrix and the third sample monitoring data; The judgment module is used to calculate a dynamic threshold value according to the first mean value and a preset threshold coefficient, and judge whether the physical device is faulty according to the relationship between the dynamic threshold value and the first mean value.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images