A device health analysis method based on AI big model

Abstract

The present invention relates to the technical field of equipment health analysis, and discloses an equipment health analysis method based on an AI large model, the method comprising: performing multi-perspective data fusion on a collected equipment operation data sequence; performing extreme point extraction on the equipment operation fusion data sequence; performing feature extraction on the equipment operation extreme value matrix; calculating the difference between the equipment operation feature matrix and the equipment standard operation feature matrix under normal working conditions, and converting the difference into equipment health to evaluate the equipment. The present invention adopts a two-dimensional iteration method of distance iteration and cumulative difference value iteration to calculate the difference between the equipment operation feature matrix and the equipment standard operation feature matrix under normal working conditions, and based on the convolution weight parameters trained by the AI ​​large model, converts the difference into equipment health in multiple scales to evaluate the equipment, thereby realizing equipment health analysis based on the direction of equipment operation state change and multi-scale information of operation data.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment health analysis, and in particular to an equipment health analysis method based on an AI large model. Background Art

[0002] With the development of industrial intelligence, equipment health analysis is becoming increasingly important for ensuring the normal operation of equipment, improving production efficiency, and reducing maintenance costs. Traditional equipment health analysis methods are often based on manual experience, but with the rapid development of artificial intelligence technology, the use of artificial intelligence technology to quantitatively analyze equipment operating status can greatly improve the efficiency and accuracy of equipment health identification. In view of this, this patent proposes an equipment health analysis method based on a large AI model. Through a large pre-trained model, massive state data is pre-trained to learn the potential patterns and regularities in the data, thereby improving the accuracy and recognition efficiency of equipment health analysis. Summary of the Invention

[0003] In view of this, the present invention provides an equipment health analysis method based on an AI large model, the purpose of which is to: 1) perform a two-dimensional transformation combining the time series dimension and the operation dimension on the equipment operation data sequence composed of the operation status data of different operation times and different operation dimensions, extract the frequency domain information of different dimensions, and fuse the frequency domain information with the time domain information to obtain a two-dimensional transformation matrix, and realize information enhancement by extracting the real part information in the two-dimensional transformation matrix and filtering the imaginary part information, and calculate the perspective weight based on the frequency information of different dimensions to realize the fusion of equipment operation data of different operation dimensions and obtain the equipment operation fusion data sequence; 2) construct a method based on the extreme points in the fused data sequence to characterize the position of the extreme points, the data value of the extreme points and the extreme values. The equipment operation extreme value matrix of the value point label is obtained, and the equipment operation left slope change characteristics, equipment operation right slope change characteristics, operation extreme value position characteristics and equipment operation fusion data of each operation moment in the matrix are extracted to form the equipment operation feature matrix. The difference between different features is used as the feature distance between the equipment operation feature matrix and the equipment standard operation feature matrix under normal working conditions. The difference between the equipment operation feature matrix and the equipment standard operation feature matrix under normal working conditions is calculated in a two-dimensional iterative manner. The convolution weight parameters obtained by training the AI ​​large model are used to convert the difference into equipment health in a multi-scale manner to evaluate the equipment, thereby realizing equipment health analysis based on the direction of change of equipment operation status and multi-scale information of operation data.

[0004] To achieve the above objectives, the present invention provides a device health analysis method based on an AI large model, comprising the following steps:

[0005] S1: Collecting device operation data according to a preset window length to obtain a device operation data sequence, performing multi-view data fusion on the collected device operation data sequence to obtain a device operation fusion data sequence, wherein frequency domain fusion data enhancement is the main implementation method of the multi-view data fusion;

[0006] S2: Extract extreme points from the device operation fusion data sequence to obtain a device operation extreme value matrix, wherein the time series local extreme value extraction method is the main implementation method of the extreme value point extraction;

[0007] S3: Extract features from the equipment operation extreme value matrix to obtain the equipment operation feature matrix;

[0008] S4: Calculate the difference between the equipment operation characteristic matrix and the equipment standard operation characteristic matrix under normal working conditions, and convert the difference into equipment health to evaluate the equipment.

[0009] As a further improvement method of the present invention:

[0010] Optionally, in step S1, collecting device operation data according to a preset window length to obtain a device operation data sequence includes:

[0011] The equipment operation data is collected according to the preset window length to obtain the equipment operation status data at different operation times and different operation dimensions, and the collected equipment operation data is used to form an equipment operation data sequence:

[0012] L=(L1,L2,...,L n ,...,L N )

[0013] L n =(L n (1),L n (2),...,L n (m),...,L n (M))

[0014] in:

[0015] L represents the equipment operation data sequence, L n Represents the device operation data of the device at the nth operation time, N represents the preset window length, n∈[1,N];

[0016] L n (1),L n (2),...,L n (m),...,L n (M) indicates equipment operation data L n The running status data of M running dimensions, where L n (m) represents equipment operation data Ln The operating status data of the mth operating dimension in [1, M] represents the number of operating dimensions; in the embodiment of the present invention, the operating dimensions include the current, voltage, load, temperature, remaining storage space, ambient temperature, and ambient humidity of the device;

[0017] The collected equipment operation data sequence is fused from multiple perspectives to obtain the equipment operation fusion data sequence.

[0018] Optionally, performing multi-view data fusion on the collected equipment operation data sequence includes:

[0019] Multi-view data fusion is performed on the collected equipment operation data sequence L, wherein frequency domain fusion data enhancement is the main implementation method of the multi-view data fusion. The multi-view data fusion process is as follows:

[0020] S11: Perform a two-dimensional transformation on the equipment operation data sequence L, combining the time series dimension and the operation dimension:

[0021]

[0022] in:

[0023] F represents the two-dimensional transformation matrix of the equipment operation data sequence L, and F(n,m) represents the transformation value at the frequency (n,m) position in the two-dimensional transformation matrix F;

[0024] j represents the imaginary unit, j 2 =-1;

[0025] exp(·) represents an exponential function with a natural constant as the base;

[0026] S12: Convert the two-dimensional transformation matrix F into a two-dimensional transformation real part matrix:

[0027]

[0028] in:

[0029] F1 represents the real part matrix of the two-dimensional transformation;

[0030] F1(n,m) represents the transformed real part value at the frequency (n,m) position in the two-dimensional transformed real part matrix F1;

[0031] S13: Calculate the perspective weights of different operating dimensions, where the perspective weight of the mth operating dimension is w m ;

[0032] S14: Weight the real part matrix of the two-dimensional transformation based on the perspective weights of multiple operating dimensions to form a device operation fusion data sequence:

[0033] X=(X1,X2,...,X n ,...,X N )

[0034]

[0035] in:

[0036] X represents the equipment operation fusion data sequence;

[0037] X n Indicates the device operation fusion data of the device at the nth operation time.

[0038] Optionally, the formula for calculating the perspective weight of the mth operating dimension in step S13 is:

[0039]

[0040]

[0041] in:

[0042] w m represents the perspective weight of the mth operating dimension;

[0043] d m Represents the perspective information of the mth operating dimension.

[0044] Optionally, in step S2, the extreme point extraction of the device operation fusion data sequence includes:

[0045] The extreme points of the device operation fusion data sequence X are extracted to form the device operation extreme value matrix. The construction process of the device operation extreme value matrix is ​​as follows:

[0046] S21: extracting all device operation fusion data and data locations in the device operation fusion data sequence X;

[0047] S22: Calculate the extreme point type of the non-start and end equipment operation fusion data, where X1, X N The starting and ending equipment operation fusion data in the equipment operation fusion data sequence X. n Extreme point type Y n The calculation formula is:

[0048]

[0049] in:

[0050] Y n Indicates the device running fusion data X n The extreme point type, Y n =1 means the equipment runs fusion data Xn is a local minimum, Y n =-1 means the equipment runs fusion data X n is a local maximum, Y n =0 means the equipment runs fusion data X n Not a local extreme value;

[0051] S23: Construct the equipment operation extreme value matrix:

[0052]

[0053] in:

[0054] C represents the extreme value matrix of device operation, C1(n) represents the fusion data X of device operation n , C2(n) represents the equipment operation fusion data X n The data location of C3(n) represents the fusion data of device operation X n The extreme point type, where C1(n)=X n , C2(n)=n,C3(n)=Y n .

[0055] Optionally, the feature extraction of the device operation extreme value matrix in step S3 includes:

[0056] Perform feature extraction on the device operation extreme value matrix C to obtain the device operation feature matrix. The feature extraction process of the device operation extreme value matrix C is as follows:

[0057] The device operation characteristics of the device at different operation times are calculated based on the device operation extreme value matrix C, where the device operation characteristics of the device at the nth operation time are:

[0058]

[0059] in:

[0060] f n Indicates the equipment operation characteristics at the nth operation moment;

[0061] T stands for transpose;

[0062] Indicates the change characteristics of the left slope of the equipment at the nth operating moment;

[0063] Indicates the right slope change characteristics of the equipment at the nth operating moment;

[0064] Indicates the extreme position characteristics of the equipment at the nth operating moment;

[0065] The equipment operation characteristic matrix that constitutes the equipment:

[0066] f=[f1,f2,...,f n ,...,f N ]

[0067] in:

[0068] f represents the equipment operation characteristic matrix.

[0069] Optionally, in step S4, calculating the difference between the equipment operation characteristic matrix and the equipment standard operation characteristic matrix under normal working conditions includes:

[0070] Obtain the equipment standard operating characteristic matrix R under normal operating conditions:

[0071] R=[R1,R2,...,R n ,...,R N ]

[0072]

[0073] in:

[0074] R n represents the standard operating characteristics of the device under normal operating conditions at the nth operating moment; in the embodiment of the present invention, the device operating data of the device under normal operating conditions is collected and the collected device operating data is processed according to steps S1-S3 to obtain the device standard operating characteristic matrix of the device under normal operating conditions;

[0075] Indicates the equipment standard operating characteristics R n The left slope change characteristics of the equipment operation;

[0076] Indicates the equipment standard operating characteristics R n The right slope change characteristics of the equipment operation;

[0077] Indicates the equipment standard operating characteristics R n The operating extreme position characteristics in ;

[0078] R1(n) represents the standard operating characteristic of the equipment R n The equipment in the system runs fusion data;

[0079] Calculate the difference Diff(Q,R) between the equipment operation characteristic matrix f and the equipment standard operation characteristic matrix R under normal working conditions. The difference calculation process is as follows:

[0080] S41: Initialize s1=1, s2=1;

[0081] S42: Calculate the equipment operation characteristics at the s1th operation time in the equipment operation characteristic matrix f The equipment standard operation characteristics at the s2th operation time in the equipment standard operation characteristic matrix R Distance between:

[0082]

[0083] in:

[0084] Indicates the operating characteristics of the device Standard operating characteristics of the equipment the distance between them;

[0085] S43: Update to obtain the cumulative difference value:

[0086]

[0087] in:

[0088] Sum(s1,s2) represents the updated equipment operation characteristics Standard operating characteristics of the equipment The cumulative difference between

[0089] S44: Set s2 = s2 + 1, and return to step S42 until s2 = N;

[0090] S45: Set s1 = s1 + 1, and return to step S42 until s1 = N;

[0091] S46: Obtain the cumulative difference value Sum(N,N), and set the difference degree Diff(Q,R)=Sum(N,N);

[0092] Based on the AI ​​big model, the difference Diff(Q,R) is converted into device health. If the device health is lower than the preset threshold, the device needs to be replaced.

[0093] Optionally, converting the difference into device health based on the AI ​​big model includes:

[0094] Based on the AI ​​big model, the difference Diff(Q,R) is converted into device health. The conversion formula of the difference Diff(Q,R) is:

[0095]

[0096] in:

[0097] P(Q,R) represents the device health corresponding to the difference Diff(Q,R);

[0098] W represents the convolution weight matrix within the convolution kernel in the AI ​​large model;

[0099] * indicates convolution processing;

[0100] MAX indicates the preset maximum difference, and MIN indicates the preset minimum difference.

[0101] In order to solve the above problem, the present invention provides an electronic device, comprising:

[0102] a memory storing at least one instruction;

[0103] Communication interfaces to enable electronic equipment to communicate; and

[0104] The processor executes the instructions stored in the memory to implement the above-mentioned device health analysis method based on the AI ​​large model.

[0105] In order to solve the above problems, the present invention also provides a computer-readable storage medium, which stores at least one instruction, and the at least one instruction is executed by a processor in an electronic device to implement the above-mentioned device health analysis method based on the AI ​​big model.

[0106] Compared with the existing technology, this invention proposes a device health analysis method based on an AI large model, which has the following advantages:

[0107] First, this solution proposes a multi-perspective data fusion method to perform multi-perspective data fusion on the collected equipment operation data sequence L. The multi-perspective data fusion process is as follows: a two-dimensional transformation combining the time series dimension and the operation dimension is performed on the equipment operation data sequence L:

[0108]

[0109] Where: F represents the two-dimensional transformation matrix of the equipment operation data sequence L, F(n,m) represents the transformation value at the frequency (n,m) position in the two-dimensional transformation matrix F; j represents the imaginary unit, j 2 = -1; exp(·) represents an exponential function with a natural constant as the base; convert the two-dimensional transformation matrix F into a two-dimensional transformation real part matrix:

[0110]

[0111] Where: F1 represents the two-dimensional transformation real part matrix; F1(n,m) represents the transformation real part value at the frequency (n,m) position in the two-dimensional transformation real part matrix F1; the perspective weights of different operating dimensions are calculated, where the perspective weight of the mth operating dimension is w m :

[0112]

[0113] Where: w m represents the perspective weight of the mth operating dimension; d m Represents the perspective information of the mth operating dimension; the two-dimensional transformation real part matrix is ​​weighted based on the perspective weights of multiple operating dimensions to form the device operation fusion data sequence:

[0114] X=(X1,X2,...,X n ,...,X N )

[0115]

[0116] Where: X represents the equipment operation fusion data sequence; X n Represents the fused data of device operation at the nth operating moment. This solution performs a two-dimensional transformation combining the time series dimension and the operating dimension on the device operation data sequence composed of operating status data at different operating moments and different operating dimensions. It extracts frequency domain information of different dimensions and fuses the frequency domain information with the time domain information to obtain a two-dimensional transformation matrix. By extracting the real part information from the two-dimensional transformation matrix and filtering the imaginary part information, information enhancement is achieved. The perspective weight is calculated based on the frequency information of different dimensions to achieve the fusion of device operation data of different operating dimensions and obtain a fused device operation data sequence.

[0117] At the same time, this solution proposes a device operation feature extraction and health analysis method, which extracts features from the device operation extreme value matrix C to obtain the device operation feature matrix. The feature extraction process of the device operation extreme value matrix C is as follows: the device operation features of the device at different operation times are calculated based on the device operation extreme value matrix C, where the device operation features of the device at the nth operation time are:

[0118]

[0119] Where: f n It represents the equipment operation characteristics at the nth operation time; T represents transposition; Indicates the change characteristics of the left slope of the equipment at the nth operating moment; Indicates the right slope change characteristics of the equipment at the nth operating moment; Represents the extreme position characteristics of the equipment at the nth operating moment; the equipment operation characteristic matrix of the equipment is:

[0120] f=[f1,f2,...,f n ,...,f N ]

[0121] Where: f represents the equipment operation characteristic matrix. Get the equipment standard operation characteristic matrix R under normal working conditions:

[0122] R=[R1,R2,...,R n ,...,R N ]

[0123] Where: R n Represents the standard operating characteristics of the equipment at the nth operating moment under normal operating conditions; calculates the difference Diff(Q, R) between the equipment operating characteristic matrix f and the equipment standard operating characteristic matrix R under normal operating conditions, where the difference calculation process is: initialize s1 = 1, s2 = 1; calculate the equipment operating characteristics at the s1th operating moment in the equipment operating characteristic matrix f The equipment standard operation characteristics at the s2th operation time in the equipment standard operation characteristic matrix R Distance between:

[0124]

[0125] in: Indicates the operating characteristics of the device Standard operating characteristics of the equipment The distance between them; update to get the cumulative difference value:

[0126]

[0127] Where: Sum(s1,s2) represents the updated equipment operation characteristics f s1 Standard operating characteristics of the equipment The cumulative difference between them; let s2 = s2 + 1, until s2 = N; let s1 = s1 + 1, until s1 = N; obtain the cumulative difference value Sum(N, N), and let the difference Diff(Q, R) = Sum(N, N); convert the difference Diff(Q, R) into device health based on the AI ​​big model. If the device health is lower than the preset threshold, the device needs to be replaced. This solution is based on the extreme points in the fused data sequence to construct an equipment operation extreme value matrix that characterizes the extreme point position, extreme point data value, and extreme point label. It also extracts the equipment operation left slope change characteristics, equipment operation right slope change characteristics, operation extreme position characteristics, and equipment operation fusion data at each operating moment in the matrix to form an equipment operation feature matrix. The difference between different features is used as the feature distance between the equipment operation feature matrix and the equipment standard operation feature matrix under normal working conditions. The difference between the equipment operation feature matrix and the equipment standard operation feature matrix under normal working conditions is calculated using a two-dimensional iterative method. The difference is converted into equipment health in a multi-scale manner based on the convolution weight parameters trained by the AI ​​large model to evaluate the equipment, thereby realizing equipment health analysis based on the direction of change of equipment operation status and multi-scale information of operation data. BRIEF DESCRIPTION OF THE DRAWINGS

[0128] Figure 1 A flowchart of a device health analysis method based on an AI large model provided by one embodiment of the present invention;

[0129] Figure 2 A schematic diagram of the structure of an electronic device that implements a device health analysis method based on an AI large model, provided in one embodiment of the present invention.

[0130] In the figure: 1 electronic device, 10 processor, 11 memory, 12 program, 13 communication interface.

[0131] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0132] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0133] The embodiments of the present application provide a device health analysis method based on an AI large model. The execution subject of the device health analysis method based on the AI ​​large model includes, but is not limited to, at least one of electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided by the embodiments of the present application. In other words, the device health analysis method based on the AI ​​large model can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes, but is not limited to: a single server, a server cluster, a cloud server, or a cloud server cluster.

[0134] Example 1:

[0135] S1: Collect device operation data according to a preset window length to obtain a device operation data sequence, and perform multi-perspective data fusion on the collected device operation data sequence to obtain a device operation fusion data sequence.

[0136] In step S1, the device operation data is collected according to the preset window length to obtain the device operation data sequence, including:

[0137] The equipment operation data is collected according to the preset window length to obtain the equipment operation status data at different operation times and different operation dimensions, and the collected equipment operation data is used to form an equipment operation data sequence:

[0138] L=(L1,L2,...,L n ,...,L N )

[0139] L n =(L n (1),L n (2),...,L n (m),...,L n (M))

[0140] in:

[0141] L represents the equipment operation data sequence, L n Represents the device operation data of the device at the nth operation time, N represents the preset window length, n∈[1,N];

[0142] L n (1),L n (2),...,L n (m),...,L n (M) indicates equipment operation data L n The running status data of M running dimensions, where L n (m) represents equipment operation data L nThe operating status data of the mth operating dimension in [1, M] represents the number of operating dimensions; in the embodiment of the present invention, the operating dimensions include the current, voltage, load, temperature, remaining storage space, ambient temperature, and ambient humidity of the device;

[0143] The collected equipment operation data sequence is fused from multiple perspectives to obtain the equipment operation fusion data sequence.

[0144] The multi-view data fusion of the collected equipment operation data sequence includes:

[0145] Perform multi-view data fusion on the collected equipment operation data sequence L, where the multi-view data fusion process is as follows:

[0146] S11: Perform a two-dimensional transformation on the equipment operation data sequence L, combining the time series dimension and the operation dimension:

[0147]

[0148] in:

[0149] F represents the two-dimensional transformation matrix of the equipment operation data sequence L, and F(n,m) represents the transformation value at the frequency (n,m) position in the two-dimensional transformation matrix F;

[0150] j represents the imaginary unit, j 2 =-1;

[0151] exp(·) represents an exponential function with a natural constant as the base;

[0152] S12: Convert the two-dimensional transformation matrix F into a two-dimensional transformation real part matrix:

[0153]

[0154] in:

[0155] F1 represents the real part matrix of the two-dimensional transformation;

[0156] F1(n,m) represents the transformed real part value at the frequency (n,m) position in the two-dimensional transformed real part matrix F1;

[0157] S13: Calculate the perspective weights of different operating dimensions, where the perspective weight of the mth operating dimension is w m ;

[0158] S14: Weight the real part matrix of the two-dimensional transformation based on the perspective weights of multiple operating dimensions to form a device operation fusion data sequence:

[0159] X=(X1,X2,...,X n ,...,X N)

[0160]

[0161] in:

[0162] X represents the equipment operation fusion data sequence;

[0163] X n Indicates the device operation fusion data of the device at the nth operation time.

[0164] The formula for calculating the perspective weight of the mth operating dimension in step S13 is:

[0165]

[0166] in:

[0167] w m represents the perspective weight of the mth operating dimension;

[0168] d m Represents the perspective information of the mth operating dimension.

[0169] S2: Extract extreme points from the device operation fusion data sequence to obtain the device operation extreme value matrix.

[0170] In step S2, the extreme point extraction of the device operation fusion data sequence includes:

[0171] The extreme points of the device operation fusion data sequence X are extracted to form the device operation extreme value matrix. The construction process of the device operation extreme value matrix is ​​as follows:

[0172] S21: extracting all device operation fusion data and data locations in the device operation fusion data sequence X;

[0173] S22: Calculate the extreme point type of the non-start and end equipment operation fusion data, where X1, X N The starting and ending equipment operation fusion data in the equipment operation fusion data sequence X. n Extreme point type Y n The calculation formula is:

[0174]

[0175] in:

[0176] Y n Indicates the device running fusion data X n The extreme point type, Y n =1 means the equipment runs fusion data X n is a local minimum, Y n=-1 means the equipment runs fusion data X n is a local maximum, Y n =0 means the equipment runs fusion data X n Not a local extreme value;

[0177] S23: Construct the equipment operation extreme value matrix:

[0178]

[0179] in:

[0180] C represents the extreme value matrix of device operation, C1(n) represents the fusion data X of device operation n , C2(n) represents the equipment operation fusion data X n The data location of C3(n) represents the fusion data of device operation X n The extreme point type, where C1(n)=X n , C2(n

[0181] )=n,C3(n)=Y n .

[0182] S3: Extract features from the device operation extreme value matrix to obtain a device operation feature matrix.

[0183] In step S3, feature extraction of the device operation extreme value matrix is ​​performed, including:

[0184] Perform feature extraction on the device operation extreme value matrix C to obtain the device operation feature matrix. The feature extraction process of the device operation extreme value matrix C is as follows:

[0185] The device operation characteristics of the device at different operation times are calculated based on the device operation extreme value matrix C, where the device operation characteristics of the device at the nth operation time are:

[0186]

[0187] in:

[0188] f n Indicates the equipment operation characteristics at the nth operation moment;

[0189] T stands for transpose;

[0190] Indicates the change characteristics of the left slope of the equipment at the nth operating moment;

[0191] Indicates the right slope change characteristics of the equipment at the nth operating moment;

[0192] Indicates the extreme position characteristics of the equipment at the nth operating moment;

[0193] The equipment operation characteristic matrix that constitutes the equipment:

[0194] f=[f1,f2,...,f n ,...,f N ]

[0195] in:

[0196] f represents the equipment operation characteristic matrix.

[0197] S4: Calculate the difference between the equipment operation characteristic matrix and the equipment standard operation characteristic matrix under normal working conditions, and convert the difference into equipment health to evaluate the equipment.

[0198] Calculating the difference between the equipment operation characteristic matrix and the equipment standard operation characteristic matrix under normal working conditions in step S4 includes:

[0199] Obtain the equipment standard operating characteristic matrix R under normal operating conditions:

[0200] R=[R1,R2,...,R n ,...,R N ]

[0201]

[0202] in:

[0203] R n represents the standard operating characteristics of the device under normal operating conditions at the nth operating moment; in the embodiment of the present invention, the device operating data of the device under normal operating conditions is collected and the collected device operating data is processed according to steps S1-S3 to obtain the device standard operating characteristic matrix of the device under normal operating conditions;

[0204] Indicates the equipment standard operating characteristics R n The left slope change characteristics of the equipment operation;

[0205] Indicates the equipment standard operating characteristics R n The right slope change characteristics of the equipment operation;

[0206] Indicates the equipment standard operating characteristics R n The operating extreme position characteristics in ;

[0207] R1(n) represents the standard operating characteristic of the equipment R n The equipment in the system runs fusion data;

[0208] Calculate the difference Diff(Q,R) between the equipment operation characteristic matrix f and the equipment standard operation characteristic matrix R under normal working conditions. The difference calculation process is as follows:

[0209] S41: Initialize s1=1, s2=1;

[0210] S42: Calculate the equipment operation characteristics at the s1th operation time in the equipment operation characteristic matrix f The equipment standard operation characteristics at the s2th operation time in the equipment standard operation characteristic matrix R Distance between:

[0211]

[0212] in:

[0213] Indicates the operating characteristics of the device Standard operating characteristics of the equipment the distance between them;

[0214] S43: Update to obtain the cumulative difference value:

[0215]

[0216] in:

[0217] Sum(s1,s2) represents the updated equipment operation characteristics Standard operating characteristics of the equipment The cumulative difference between

[0218] S44: Set s2 = s2 + 1, and return to step S42 until s2 = N;

[0219] S45: Set s1 = s1 + 1, and return to step S42 until s1 = N;

[0220] S46: Obtain the cumulative difference value Sum(N,N), and set the difference degree Diff(Q,R)=Sum(N,N);

[0221] Based on the AI ​​big model, the difference Diff(Q,R) is converted into device health. If the device health is lower than the preset threshold, the device needs to be replaced.

[0222] The AI-based model converts the difference into device health, including:

[0223] Based on the AI ​​big model, the difference Diff(Q,R) is converted into device health. The conversion formula of the difference Diff(Q,R) is:

[0224]

[0225] in:

[0226] P(Q,R) represents the device health corresponding to the difference Diff(Q,R);

[0227] W represents the convolution weight matrix within the convolution kernel in the AI ​​large model;

[0228] * indicates convolution processing;

[0229] MAX indicates the preset maximum difference, and MIN indicates the preset minimum difference.

[0230] Example 2:

[0231] like Figure 2 , which is a schematic diagram of the structure of an electronic device for implementing a device health analysis method based on an AI large model provided by an embodiment of the present invention.

[0232] The electronic device 1 may include a processor 10 , a memory 11 , a communication interface 13 and a bus, and may further include a computer program stored in the memory 11 and executable on the processor 10 , such as a program 12 .

[0233] The memory 11 includes at least one type of readable storage medium, and the readable storage medium includes a flash memory, a mobile hard disk, a multimedia card, a card-type memory (for example, an SD or DX memory, etc.), a magnetic memory, a disk, an optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of the electronic device 1, such as a mobile hard disk of the electronic device 1. In other embodiments, the memory 11 can also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 1. Furthermore, the memory 11 can also include both an internal storage unit of the electronic device 1 and an external storage device. The memory 11 can not only be used to store application software and various types of data installed in the electronic device 1, such as the code of the program 12, etc., but can also be used to temporarily store data that has been output or is to be output.

[0234] In some embodiments, the processor 10 may be composed of an integrated circuit, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and a combination of various control chips. The processor 10 is the control core (Control Unit) of the electronic device, connecting the various components of the entire electronic device using various interfaces and lines, and executing the programs or modules stored in the memory 11 (such as the program 12 for implementing device health analysis based on the AI ​​large model) and calling the data stored in the memory 11 to perform various functions of the electronic device 1 and process data.

[0235] The communication interface 13 may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device 1 and other electronic devices, and to realize connection and communication between internal components of the electronic devices.

[0236] The bus may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The bus is configured to enable connection and communication between the memory 11 and at least one processor 10, etc.

[0237] Figure 2 Only the electronic device with components is shown, and it can be understood by those skilled in the art that Figure 2 The structure shown does not constitute a limitation on the electronic device 1 , and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0238] For example, although not shown, the electronic device 1 may further include a power source (such as a battery) for powering the various components. Preferably, the power source may be logically connected to the at least one processor 10 via a power management device, thereby implementing functions such as charging management, discharging management, and power consumption management through the power management device. The power source may further include any components such as one or more DC or AC power sources, a recharging device, a power failure detection circuit, a power converter or inverter, a power status indicator, etc. The electronic device 1 may further include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0239] Optionally, the electronic device 1 may further include a user interface, which may be a display or an input unit (such as a keyboard). Optionally, the user interface may also be a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touch device. The display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device 1 and to display a visual user interface.

[0240] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0241] The program 12 stored in the memory 11 of the electronic device 1 is a combination of multiple instructions. When executed in the processor 10, it can achieve the following:

[0242] Collecting equipment operation data according to a preset window length to obtain an equipment operation data sequence, and performing multi-view data fusion on the collected equipment operation data sequence to obtain an equipment operation fusion data sequence;

[0243] Extract extreme points from the equipment operation fusion data sequence to obtain the equipment operation extreme value matrix;

[0244] Extract features from the extreme value matrix of equipment operation to obtain the equipment operation feature matrix;

[0245] Calculate the difference between the equipment operation characteristic matrix and the equipment standard operation characteristic matrix under normal working conditions, and convert the difference into equipment health to evaluate the equipment.

[0246] Specifically, the specific implementation method of the processor 10 for the above instructions can refer to Figures 1 to 2 The description of the relevant steps in the corresponding embodiments will not be repeated here.

[0247] It should be noted that the serial numbers of the above-mentioned embodiments of the present invention are for descriptive purposes only and do not represent the advantages or disadvantages of the embodiments. In addition, the terms "including", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "including a ..." does not exclude the presence of other identical elements in the process, device, article or method comprising the element.

[0248] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better embodiment. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0249] The above are only preferred embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A device health analysis method based on AI big model, characterized in that: The method comprises: S1: Collecting equipment operation data according to a preset window length to obtain an equipment operation data sequence, and performing multi-view data fusion on the collected equipment operation data sequence to obtain an equipment operation fusion data sequence; S2: extract extreme points from the equipment operation fusion data sequence to obtain the equipment operation extreme value matrix; S3: extracting features from the equipment operation extreme value matrix to obtain the equipment operation feature matrix; S4: Calculate the difference between the equipment operation characteristic matrix and the equipment standard operation characteristic matrix under normal working conditions, and convert the difference into equipment health to evaluate the equipment; In the step S1, the device operation data is collected according to the preset window length to obtain the device operation data sequence, including: The equipment operation data is collected according to the preset window length to obtain the equipment operation status data at different operation times and different operation dimensions, and the collected equipment operation data is used to form an equipment operation data sequence: L=(L1,L2,...,L rv ,...,L N ) L n =(L n (1),L n (2),...,L n (m),...,L n (M)) in: L represents the equipment operation data sequence, L n Indicates the device operation data at the nth operation time, N represents the preset window length, L n (1), L n (2), ..., L n (m), ..., L n (M) indicates equipment operation data L n The running status data of M running dimensions, where L n (m) indicates equipment operation data L n The running status data of the mth running dimension in , where M represents the number of running dimensions, Perform multi-view data fusion on the collected equipment operation data sequence to obtain the equipment operation fusion data sequence; The multi-view data fusion of the collected equipment operation data sequence includes: The collected equipment operation data sequence L is subjected to multi-view data fusion, wherein the multi-view data fusion process is as follows: S11: Perform a two-dimensional transformation on the equipment operation data sequence L by combining the time series dimension and the operation dimension: in: F represents the two-dimensional transformation matrix of the equipment operation data sequence L, and F(n, m) represents the transformation value at the frequency (n, m) position in the two-dimensional transformation matrix F; j represents the imaginary unit, j 2 = -1; exp(·) represents an exponential function with a natural constant as the base; S12: Convert the two-dimensional transformation matrix F into a two-dimensional transformation real part matrix: in: F1 represents the real part matrix of the two-dimensional transformation; F1(n,m) represents the transformed real part value at the frequency (n,m) position in the two-dimensional transformed real part matrix F1; S13: Calculate the perspective weights of different operating dimensions, where the perspective weight of the mth operating dimension is w m ; S14: Weight the real part matrix of the two-dimensional transformation based on the perspective weights of multiple operating dimensions to form a device operation fusion data sequence: in: X represents the equipment operation fusion data sequence; X n Indicates the device operation fusion data of the device at the nth operation time.

2. The device health analysis method based on AI big model as claimed in claim 1, characterized in that: The formula for calculating the viewing angle weight of the mth operating dimension in step S13 is: in: w m represents the perspective weight of the mth operating dimension; d m Represents the viewing angle information of the mth operating dimension.

3. The device health analysis method based on AI big model as claimed in claim 1, characterized in that: In the step S2, the extreme point extraction of the device operation fusion data sequence includes: The extreme points of the equipment operation fusion data sequence X are extracted to form an equipment operation extreme value matrix. The construction process of the equipment operation extreme value matrix is ​​as follows: S21: extracting all device operation fusion data and data positions in the device operation fusion data sequence X; S22: Calculate the extreme point type of non-starting and ending equipment operation fusion data, where X1, X N is the start and end equipment operation fusion data in the equipment operation fusion data sequence X. Equipment operation fusion data X n The extreme point type Y n The calculation formula is: in: Y n Indicates the equipment running fusion data X n The extreme point type, Y n =1 means the equipment runs fusion data X n is a local minimum, Y n =-1 means the equipment runs fusion data X n is a local maximum, Y n =0 means the equipment runs fusion data X n Not a local extreme value; S23: Construct the extreme value matrix of equipment operation: in: C represents the extreme value matrix of equipment operation, C1(n) represents the fusion data X of equipment operation n , C2(n) represents the equipment operation fusion data X n The data location of C3(n) indicates the fusion data of device operation X n The extreme point type, where C1(n) = X n , C2(n)=n,C3(n)=Y n .

4. The device health analysis method based on AI big model as claimed in claim 3 is characterized in that: In the step S3, feature extraction is performed on the device operation extreme value matrix, including: Feature extraction is performed on the device operation extreme value matrix C to obtain the device operation feature matrix, where the feature extraction process of the device operation extreme value matrix C is as follows: The device operation characteristics of the device at different operation times are calculated based on the device operation extreme value matrix C, where the device operation characteristics of the device at the nth operation time are: in: f n Indicates the equipment operation characteristics of the equipment at the nth operation time; T stands for transpose; Indicates the change characteristics of the left slope of the equipment at the nth running moment; Indicates the right slope change characteristics of the equipment at the nth running moment; Indicates the extreme position characteristics of the equipment at the nth operating moment; The equipment operation characteristic matrix that constitutes the equipment: in: f represents the equipment operation characteristic matrix.

5. The device health analysis method based on AI big model as claimed in claim 4 is characterized in that: Calculating the difference between the equipment operation characteristic matrix and the equipment standard operation characteristic matrix under normal working conditions in step S4 includes: Get the equipment standard operation characteristic matrix R under normal working conditions: R=[R1,R2,...,R n ,...,R N ] in: R n Indicates the standard operating characteristics of the equipment at the nth operating moment under normal operating conditions; Indicates the standard operating characteristics of the equipment R n The left slope variation characteristics of the equipment operation; Indicates the standard operating characteristics of the equipment R n The right slope change characteristics of the equipment operation; Indicates the standard operating characteristics of the equipment R n The operating extreme position characteristics in ; R1(n) represents the standard operating characteristic of the equipment R n The equipment in the system runs fusion data; Calculate the difference Diff(Q, R) between the equipment operation characteristic matrix f and the equipment standard operation characteristic matrix R under normal working conditions, where the difference calculation process is: S41: Initialize s1=1, s2=1; S42: Calculate the equipment operation characteristics at the s1th operation time in the equipment operation characteristic matrix f The equipment standard operation characteristics at the s2th operation time in the equipment standard operation characteristic matrix R Distance between: in: Indicates the operating characteristics of the device Standard operating characteristics with equipment The distance between S43: Update to obtain the cumulative difference value: in: Sum(s1, s2) represents the updated equipment operation characteristics Standard operating characteristics with equipment The cumulative difference between S44: Set s2=s2+1, and return to step S42 until s2=N; S45: Set s1=s1+1, and return to step S42 until s1=N; S46: Obtain the cumulative difference value Sum(N, N), and set the difference degree Diff(Q, R)=Sum(N, N); Based on the AI ​​big model, the difference Diff(Q, R) is converted into equipment health. If the equipment health is lower than the preset threshold, the equipment needs to be replaced.

6. The device health analysis method based on AI big model as claimed in claim 5, characterized in that: The AI-based big model is used to convert the difference into equipment health, including: Based on the AI ​​big model, the difference Diff (Q, R) is converted into the equipment health, where the conversion formula of the difference Diff (Q, R) is: in: P(Q, R) represents the device health corresponding to the difference Diff(Q, R); W represents the convolution weight matrix in the convolution kernel in the AI ​​large model; * indicates convolution processing; MAX indicates the preset maximum difference, and MIN indicates the preset minimum difference.

Images