CN113469833A - Remote intelligent detection and processing method supporting industrial internet equipment - Google Patents

Remote intelligent detection and processing method supporting industrial internet equipment Download PDF

Info

Publication number
CN113469833A
CN113469833A CN202110670108.1A CN202110670108A CN113469833A CN 113469833 A CN113469833 A CN 113469833A CN 202110670108 A CN202110670108 A CN 202110670108A CN 113469833 A CN113469833 A CN 113469833A
Authority
CN
China
Prior art keywords
fault
industrial internet
temp
code
equipment
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202110670108.1A
Other languages
Chinese (zh)
Other versions
CN113469833B (en
Inventor
窦万春
周维
孙玉虎
张松
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University
Original Assignee
Nanjing University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanjing University filed Critical Nanjing University
Priority to CN202110670108.1A priority Critical patent/CN113469833B/en
Publication of CN113469833A publication Critical patent/CN113469833A/en
Application granted granted Critical
Publication of CN113469833B publication Critical patent/CN113469833B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/04—Manufacturing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0637—Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
    • G16Y40/00—IoT characterised by the purpose of the information processing
    • G16Y40/10—Detection; Monitoring
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Educational Administration (AREA)
  • Theoretical Computer Science (AREA)
  • Marketing (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Physics & Mathematics (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Manufacturing & Machinery (AREA)
  • Development Economics (AREA)
  • Primary Health Care (AREA)
  • Game Theory and Decision Science (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Computing Systems (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The invention provides a remote intelligent detection and processing method supporting industrial internet equipment, which comprises the following steps: step 1, constructing a fault library, a method library and a fault detection model of equipment; step 2, after the industrial internet equipment breaks down, the fault description information is sent to an industrial internet platform; step 3, after the industrial internet platform obtains the fault description information, obtaining a fault solution in the method library through a fault detection model, and feeding back the fault solution to the industrial internet equipment in the form of a two-dimensional code and a fault code; step 4, a user can obtain a fault solution by scanning the two-dimensional code or inputting the fault code; and 5, processing the fault of the industrial Internet equipment by the user through the fault solution, feeding the fault information back to the industrial Internet platform by the user when the fault solution cannot solve the fault, and starting the optimization of the fault detection model after the platform collects enough information.

Description

Remote intelligent detection and processing method supporting industrial internet equipment
Technical Field
The invention relates to a remote intelligent detection and processing method supporting industrial internet equipment.
Background
With the rapid development of the industrial manufacturing industry, the informatization of industrial production is greatly promoted by governments and enterprises. The continuous development of information technology enables the internet to be popularized and popularized continuously. At present, the concept of the internet and the advanced manufacturing industry is deepened, and an industrial internet concept is proposed. Ten years ago, the general electric company first proposed the concept of industrial interconnection, which idea is to connect related industrial elements such as production equipment, employees, customers, products, suppliers, industrial chains, etc. through an industrial-oriented network. In 2019, China issued guidance on industrial internet network construction and popularization, and the guidance plays a promoting role in industrial internet development. Industrial internet is essentially the internet of things and a particular manifestation of the internet in the industrial world, which allows instrumented objects (devices, products, systems, etc.) and people to collaborate, share information, collaborate, and make decisions in an industrial environment. Nowadays, the industrial internet provides a method for converting industrial operation flow by integrating a machine sensor, middleware, software and a back-end cloud computing and storing system, and corresponding results obtained by technologies such as artificial intelligence and the like are used as feedback to improve the overall efficiency. Through the industrial internet, intelligent interconnected objects and personnel can exchange information, continuously communicate with each other and cooperatively work to enhance the production environment.
Under the current industrial internet environment, by selecting a proper state monitoring sensor, the data acquisition of state signals of all mechanical parts of the industrial internet equipment can be continuously and parallelly carried out. These data roles of the device are to be remotely transmitted to the industrial internet platform when the device fails. The equipment fault detection is an important component of a predictive maintenance technology system, and is characterized by a feature extraction algorithm and a fault identification method. The choice of a suitable condition monitoring sensor is emphasized because the feature extraction algorithm is extracting the valid content of the raw signal. The effective information collected by the appropriate sensor is more, so that the fault type identification is more facilitated, the fault confirmation is carried out, and the early warning information is generated. The practical significance of equipment fault diagnosis in predictive maintenance is to remind equipment users of removing faults in time and enable the equipment to enter a stable operation period again.
However, after industrial equipment leaves a factory, the actual use environment may be very harsh, for example, some large machines (excavators and the like) are usually operated at a site such as a construction site, and maintenance personnel cannot arrive at the site in a short time. When industrial internet equipment breaks down, equipment users usually need to look up a large amount of professional data to overhaul the equipment, so that a large amount of time can be wasted, important influences are generated on construction period and the like, and even serious consequences are caused. During actual device usage, many errors can be resolved by some necessary instructions or guidance to the user himself, but the difficulty in obtaining guidance programs results in the user having to wait for the merchant to provide help. However, waiting for the merchant to provide help often wastes a significant amount of time, which is unacceptable in situations where the project is tight or the equipment is in high demand for use. At this time, how to remotely support the user to quickly solve the fault becomes a problem to be solved urgently.
The advent of industrial internet and artificial intelligence technologies such as sensors, networking systems, powerful and inexpensive data storage systems, wireless communications, powerful data analysis capabilities, provide fertile soil for detecting and handling failure events. This also makes it possible to perform remote detection and processing of industrial internet devices in an industrial internet environment, thereby improving production efficiency.
Disclosure of Invention
The purpose of the invention is as follows: the invention aims to provide a remote intelligent detection and processing method supporting industrial internet equipment, aiming at the problem that the industrial internet equipment lacks a remote fault detection and processing scheme under the industrial internet environment.
The invention discloses a remote intelligent detection and processing method supporting industrial internet equipment, which comprises the following steps:
step 1, constructing a fault library M, a method library W and a fault detection model G of equipment based on equipment delivery information and equipment test data of industrial internet equipment; the fault library is the fault description information of the equipment and the fault category to which the fault belongs. The method library is a set of fault solutions, each fault solution is composed of two parts, one part is a fault code, and the other part is a specific description of the fault code corresponding to the problem solution, generally multimedia content;
and 2, after the industrial internet equipment breaks down, sending the fault description information to an industrial internet platform, wherein the industrial internet equipment is generally provided with a sensor to collect the information of the equipment in real time. When equipment fails, the sensor collects internal equipment information to form a fault vector mtempAnd transmitting the information to the industrial Internet platform at the nearest position. The industrial internet platform stores a fault library M, a method library W and a fault detection model G;
and 3, after the industrial internet platform obtains the fault description information, obtaining a fault solution in the method library through the fault detection model, and feeding back the fault solution to the industrial internet equipment in the form of the two-dimensional code and the fault code. Generally, the fault detection model G can calculate the category to which the fault belongs through a fault vector transmitted to the industrial internet platform, and the industrial internet platform obtains a fault code corresponding to the fault category to obtain a solution to the fault. In addition, the industrial internet platform can also generate two-dimension code information for multimedia contents describing a fault solution, and simultaneously send the two-dimension code information and the fault code to the industrial internet equipment;
and 4, the user can obtain a fault solution by scanning the two-dimensional code or inputting the fault code. Because the multimedia content is included in the two-dimensional code, a user can obtain the description content of the fault solution by scanning the two-dimensional code, when the content comprises video and text content, the video and the text are both returned to the user, otherwise, only the text content is returned. Meanwhile, the invention considers some industrial internet devices with limited display content, therefore, the invention also allows users to obtain a fault solution by inputting fault codes;
and step 5, obtaining a fault library and a method library built in the remote industrial Internet platform through testing by a manufacturer before equipment leaves a factory, but the user may have no problem when the manufacturer tests in the actual use process. Therefore, the trouble code and trouble solution transmitted from the remote industrial internet platform are not always able to solve the problem occurred to the user,
in step 1, the fault library M is fault description information of the device and a fault category to which the fault belongs, and includes a plurality of fault description information and fault categories to which the fault belongs, and M ═1,m2,m3… }, where the ith fault description information mi＝{mi,1,mi,2,…,mi,n,yiOne piece of fault description information consists of n +1 parameters, and the first n parameters are description information of n dimensions of the current industrial internet equipment, namely mi,nDescription information of nth dimension for industrial internet device, yiIs the category to which the fault belongs;
the method library W contains solutions corresponding to the faults, and each fault solution is composed of two parts, one part is a fault code, and the other part is a detailed description of the problem solution corresponding to the fault code, and is generally multimedia content. The library of methods is generally denoted as W ═ W1,w2,…wlH, wherein the ith failure solution wi＝{eri,mediIt can be seen that one solution consists of two parts, eriIndicating a fault code which is a string of characters; mediRepresenting a piece of multimedia content, said multimedia content presenting an eriThe specific solution to the described failure; fault code deviceiAnd the category y to which the fault belongsiThe corresponding relation exists, and the fault detection model G can describe the information m according to the faultiCalculating the fault category y of the current faultiFind outCorresponding fault eriAnd positioning a corresponding fault solution in the method library, wherein the fault detection model G is a current existing model.
The present invention uses G to represent the fault detection model. The fault detection model can be based on the fault description information miCalculating the fault category to which the current fault belongs and a solution corresponding to the fault;
it should be noted that, for an industrial internet device, the information is possessed by the device when the device leaves a factory, and a manufacturer performs a relevant test when the device leaves the factory and obtains corresponding device data;
in step 2, the industrial internet equipment is generally equipped with a sensor to collect information of the equipment in real time. When the equipment has faults, the sensor collects the internal equipment information to form a fault vector mtempAnd transmitting the data to the industrial internet platform at the nearest position. The industrial internet platform stores a fault library M, a method library W and a fault detection model G.
In step 3, the fault detection model G can calculate a solution corresponding to the fault through the fault vector transmitted to the industrial internet platform. The industrial internet platform collects the information, generates two-dimension code information for describing the multimedia content of the fault solution method, and simultaneously sends the two-dimension code information and the fault code to the industrial internet equipment.
Step 4 comprises the following steps: the two-dimension code comprises multimedia content, a user can obtain description content of a fault solution by scanning the two-dimension code, when the description content comprises video and text content, the video and the text are returned to the user, and if the description content only comprises the text content, the text content is only returned; if the industrial internet device can only display the fault code, the user obtains a fault solution by inputting the fault code.
In step 5, the fault library and the method library built in the remote industrial internet platform are obtained by a manufacturer through testing before the equipment leaves a factory, but a user may have no problem when the manufacturer tests in the actual use process. Therefore, the fault code and the fault solution transmitted by the remote industrial internet platform may not always solve the problem of the user, and in particular, the user processes the fault of the industrial internet device through the fault solution, including the following two cases:
in the first case, a user obtains a fault solution by inputting a fault code or scanning a two-dimensional code, and successfully solves the problem of the industrial internet equipment through the fault solution, when the equipment actually works, a part of faults cannot be solved by the user (for example, parts need to be replaced), and the manufacturer needs to be informed to dispatch a professional to solve the problem. Among all solutions obtained by scanning the two-dimensional code or inputting the fault code, the present invention considers that the current fault solution is effective if the fault occurred by the user is successfully solved. After the solution pushed by the industrial internet platform successfully solves the problem of the fault existing in the industrial internet equipment, the fault detection and processing flow ends;
in the second situation, the problem that the equipment can not be solved by a fault solution obtained by inputting a fault code or scanning a two-dimensional code by a user is solved, the two-dimensional code pushed by the industrial internet platform to the user contains the fault solution, and a window facilitating the user to feed back is arranged. After the remote industrial internet platform obtains user feedback, a manufacturer is firstly informed to send a professional to go to the place where the industrial internet equipment is located to overhaul the equipment, and the fault vector m is markedtemp. After the professional detects and eliminates the fault, the solution of the fault is submitted to an industrial internet control platform through a background. The industrial internet platform can generate fault codes er for the solutions of the faultstempAnd multimedia content medtemp. Then the industrial internet platform can be processed according to the following steps:
step 5-1, the industrial internet platform can convert the fault vector mtempFault code devicetempAnd multimedia content medtempOne note of compositionAdding the record into a temporary sample library, wherein the record style is mtemp,ertemp,medtemp,counter]Therein, counterThe number of records corresponding to the current fault code is recorded; since the error is a failure that does not exist in the factory test, the failure code should also be a new string;
step 5-2, the industrial Internet platform can continuously collect the use condition of the user and record the fault which can not be solved, when the number of the existing records is countederWhen the value is greater than the threshold value gamma, the value is 100 in the invention, and the industrial internet platform starts the optimization process of the fault detection model G;
and 5-3, adding the records meeting the requirements in the temporary sample library into a fault library by the industrial internet platform to form a record { m }temp,1,mtemp,2,…,mtemp,n,ytempThen the industrial internet platform starts the optimization process of the fault detection model G, wherein mtemp,nFor this record, the description information of the nth dimension of the industrial Internet equipment, ytempThe category of the industrial internet equipment fault in the record is recorded; it should be noted that y istempAnd ertempIs in one-to-one correspondence with ytempTypically the existing maximum yiValue plus 1, i.e. ytemp＝maxyi+1；
Step 5-4, for ease of consideration, the present invention uses a log-probability regression model. Meanwhile, fault detection in an industrial internet environment generally refers to a multi-classification problem, and therefore, the invention adopts Error Correcting Output Codes (ECOCs) to solve the multi-classification problem. After the data in the temporary sample library is added into the fault library, setting the current common K-type faults, and dividing the K-type faults into two types, wherein the K-type faults are
Figure BDA0003118834170000051
Sample as a good example, the rest
Figure BDA0003118834170000052
The individual example is a counterexample; the error correction output code is a "Many-to-Many" (Many vs. Many, MvM) splitA common splitting technology in a strategy is a common technology for solving the problem of multi-classification in machine learning.
Step 5-5, utilizing error correction output code to make final output
Figure BDA0003118834170000053
For each partition, constructing a two-classification model, wherein the conditional probability distribution of the two-term log probability regression model is as follows:
Figure BDA0003118834170000054
Figure BDA0003118834170000055
wherein, P (y)i＝1|xi) When the sample is xiA probability that the category is a first category;
P(yi＝0|xi) When the sample is xiThen, the class is the probability of the second class;
xi＝(mi,1,mi,2,…,mi,n1) is equivalent to miI.e. the ith fault description information, yiFor the category to which the fault belongs, w ═ w1,w2,…,wNB) is a model parameter, wNThe model parameter of the Nth sample is N, the total number of samples is N, and b is a constant parameter;
the following steps are mainly to find the parameter w, where x ═ mi,1,mi,2,…,mi,N,1),w＝(w1,w2,…,wN,n)。
And 5-6, solving the formula in the step 5-5 by using maximum likelihood estimation, and for a given fault library, firstly setting:
P(yi＝1|xi)＝π(xi)
obtaining:
P(yi＝0|xi)＝1-π(xi)
the likelihood function is:
Figure BDA0003118834170000061
where N is the total number of samples.
The log-likelihood function L (w) is:
Figure BDA0003118834170000062
obtaining a bivariate logarithm probability regression model w*：
w*＝argwmin L(w)；
argwCalculating a sign for one of mathematics;
step 5-7, order
Figure BDA0003118834170000063
Obtaining const bivariate logarithm probability regression model f through steps 5-61,f2,…,fconst，fconstRepresenting the const second log probability regression model, for the k fault class, passing the jth second log probability regression model fjThe result is calculated as
Figure BDA0003118834170000064
J is 1. ltoreq. const, and
Figure BDA0003118834170000065
binary error correction output codes for K failure classes are obtained, as shown in table 1 below.
TABLE 1
Figure BDA0003118834170000071
Step 5-8, obtaining the error correction output code of each category through step 5-7, wherein the error correction output code of the category k is
Figure BDA0003118834170000072
Wherein
Figure BDA0003118834170000073
Indicating that for the kth failure class, the second-term log-probability regression model f is passedconstA result of the calculation; the error correction output code for any sample is
Figure BDA0003118834170000074
Wherein
Figure BDA0003118834170000075
Indicating that for the temp fault class, the second term log probability regression model f is passedconstThe calculated result, error correction output code ec of any sampletempThe distance dist (k, temp) from the kth class of error correction output codes is:
Figure BDA0003118834170000076
wherein
Figure BDA0003118834170000077
Show that for the kth fault class, through the s second term log probability regression model fsA result of the calculation;
Figure BDA0003118834170000078
indicating that for the temp fault class, through the s second log probability regression model fsA result of the calculation; step 5-9, obtaining a multi-term logarithm probability regression model y*：
y*＝argymin{dist(y,temp)}；
Wherein y is all categories, y is more than or equal to 1 and less than or equal to K, temp is input data to be classified, dist (y, temp) represents the distance between the error correction output code of the temp category and the error correction output code of the y category;
and 5-10, replacing the current fault detection model G by the new fault detection model to detect new problems of the user.
Compared with the prior art, the invention has the beneficial effects that:
(1) by the method, when the industrial internet equipment has a fault, the fault information can be transmitted to the remote industrial internet platform in real time, and a fault detection model in the industrial internet platform can detect the fault in real time based on the fault information;
(2) by the method, a user can conveniently and quickly acquire a fault solution based on the two-dimension code and the error code provided in the remote industrial internet;
(3) by the method, a user can feed back whether the fault solution provided by the industrial internet platform is effective or not in real time, and the built-in remote automatic modeling and model tuning mechanism aiming at the industrial internet equipment can automatically optimize according to the updated parameters so as to create a new model.
Drawings
The foregoing and/or other advantages of the invention will become further apparent from the following detailed description of the invention when taken in conjunction with the accompanying drawings.
Fig. 1 is a flow chart of the inventive device fault detection.
FIG. 2 is a basic block diagram of the method of the present invention.
Detailed Description
The invention discloses a method for supporting remote detection and processing of industrial Internet equipment, and a flow chart of the method is shown in figures 1 and 2, and the method comprises the following steps:
step 1, constructing a fault library, a method library and a fault detection model of the equipment based on equipment delivery information and equipment test data of the industrial internet equipment. The fault library is the fault description information of the equipment and the fault category to which the fault belongs. The method library is a set of fault solutions, each fault solution is composed of two parts, one part is a fault code, and the other part is a specific description of a problem solution corresponding to the fault code, and is generally multimedia content;
step 2, after the industrial internet equipment breaks down, the fault description information is sent to an industrial internet platform;
step 3, after the industrial internet platform obtains the fault description information, obtaining a fault solution in the method library through a fault detection model, and feeding back the fault solution to the industrial internet equipment in the form of a two-dimensional code and a fault code;
step 4, a user can obtain a fault solution by scanning the two-dimensional code or inputting the fault code;
and 5, processing the fault of the industrial Internet equipment by the user through a fault solution. If the fault is successfully solved, the process is ended; if the fault solution cannot solve the fault, the user can feed the fault information back to the industrial interconnection network platform, and the platform starts the optimization of the fault detection model after collecting enough information.
2. The method for remotely and intelligently detecting and processing the industrial internet equipment as claimed in claim 1, wherein the equipment failure information and processing method obtained in the step 1 can be described as follows:
(1) the fault library M contains a plurality of kinds of fault description information and fault types to which the faults belong, where M is { M }1,m2,m3… }, where the ith fault description information mi＝{mi,1,mi,2,…,mi,n,yiIt can be seen that one piece of fault description information is composed of n +1 parameters, the first n parameters are description information of n dimensions of the current industrial internet equipment, and y isiIs the category to which the fault belongs;
(2) the method library W contains solutions corresponding to a plurality of faults, wherein W is { W ═ W1,w2,…,wlH, wherein the ith failure solution wi＝{eri,mediIt can be seen that one solution consists of two parts, eriIndicating a fault code which is a string of characters; mediRepresenting a piece of multimedia content, which content demonstrates eriThe specific solution to the described failure; class of fault eriAnd the category y to which the fault belongsiThere is a correspondence relationship that the fault detection model G can pass through the calculated yiFind out the corresponding fault eriAnd locating the corresponding method in the method library.
(3) The fault detection model is denoted herein by G. The fault detection model can be based on the fault description information miAnd calculating the fault category to which the current fault belongs and a solution corresponding to the fault.
It should be noted that, for an industrial internet device, the information is what the device has when it leaves the factory, and the manufacturer will perform relevant tests when the device leaves the factory and obtain the corresponding device data.
3. The method as claimed in claim 2, wherein in step 2, the industrial internet device is generally equipped with a sensor to collect information of the device in real time. When equipment fails, the sensor collects internal equipment information to form a fault vector mtempAnd transmitting the data to the industrial Internet platform at the nearest position. The industrial internet platform stores a fault library M, a method library W and a fault detection model G.
4. The method as claimed in claim 3, wherein in step 3, the fault detection model G can calculate a solution corresponding to the fault through a fault vector transmitted to the industrial internet platform. The industrial internet platform collects the information, generates two-dimension code information for describing multimedia contents of the fault solution method, and simultaneously sends the two-dimension code information and the fault code to the industrial internet equipment.
5. The method as claimed in claim 4, wherein in step 4, since the two-dimensional code includes multimedia content, the user can scan the two-dimensional code to obtain the description content of the failure solution, when the content includes video and text content, the video and text are returned to the user, otherwise, only the text content is returned. Meanwhile, the present invention considers some industrial internet devices having limited display contents (only fault codes can be displayed), and thus, the present invention also allows a user to obtain a fault solution by inputting a fault code.
6. The method as claimed in claim 5, wherein in step 5, the fault library and the method library built in the remote industrial internet platform are obtained by a manufacturer through testing before the equipment is delivered from the manufacturer, but a user may encounter some problems that do not occur when the manufacturer tests during actual use. Therefore, the trouble code and trouble solution transmitted from the remote industrial internet platform may not always solve the problem occurred by the user, and in particular, may be classified as follows:
(1) the user obtains a fault solution by inputting the fault code or scanning the two-dimensional code, and the problem of the industrial internet equipment is successfully solved through the fault solution. When the equipment actually works, part of faults can not be solved by users (for example, parts need to be replaced), and manufacturers must be informed to solve the faults in a manner of dispatching professional personnel. Among all solutions obtained by scanning the two-dimensional code or inputting the fault code, the present invention considers that the current fault solution is effective if the fault occurred by the user is successfully solved. After the solution pushed by the industrial internet platform successfully solves the fault existing in the industrial internet equipment, the fault detection and processing flow is ended;
(2) the problem that the equipment appears cannot be solved by a fault solution scheme obtained by inputting a fault code or scanning a two-dimensional code by a user. The two-dimensional code processing pushed to the user by the industrial internet platform comprises a fault solution, and the two-dimensional code processing system further comprises a window which is convenient for the user to feed back, and when the user cannot process a fault through the pushed fault solution, the information can be fed back to the remote industrial internet platform through the feedback window. After the remote industrial internet platform obtains user feedback, a manufacturer is firstly informed to send a professional to go to the place where the industrial internet equipment is located to overhaul the equipment, and the fault vector m is markedtemp. After the professional staff detects and eliminates the fault, the staff willAnd submitting the solution of the fault to an industrial Internet control platform through a background. The industrial Internet platform generates a fault code er for the schemetempAnd multimedia content medtemp. Then, the industrial internet platform processes the data according to the following steps:
(2-1) the industrial Internet control center can transmit the fault vector mtempFault code devicetempAnd multimedia content medtempAdding a record into the temporary sample library, wherein the record has a pattern of mtemp,ertemp,medtemp,counter]. Wherein, counterThe number of records corresponding to the current fault code is recorded. Since the error is a failure that does not exist in the factory test, the failure code should also be a new string;
(2-2) the industrial Internet platform can continuously collect the use condition of the user and record the fault which can not be solved when the number of the existing records is countederWhen the value of gamma is larger than the threshold value gamma, the value of gamma is 100, and the industrial internet platform starts the optimization process of the detection model G;
(2-3) the industrial Internet platform adds the records meeting the requirements in the temporary sample library into the fault library to form { m }temp,1,mtemp，2,…,mtemp,n,ytempRecording, then the industrial internet platform will start the training process, where mtemp,nFor this record, the description information of the nth dimension of the industrial Internet equipment, ytempAnd recording the category of the industrial Internet equipment fault in the record. It should be noted that y istempAnd ertempIs in one-to-one correspondence with ytempGenerally the existing maximum yiValue plus 1, i.e. ytemp＝maxyi+1；
(2-4) for convenience of consideration, the present invention uses a log probability regression model. Meanwhile, fault detection in an industrial internet environment generally refers to a multi-classification problem, and therefore, the invention adopts Error Correcting Output Codes (ECOCs) to solve the multi-classification problem. After the data in the temporary sample library is added into the fault library, assuming that the current common K types exist, the data will beThe class K faults are divided into two classes, wherein
Figure BDA0003118834170000111
Sample as a good example, the rest
Figure BDA0003118834170000112
The sample is a counter example. The error correction output code is a common splitting technology in a 'Many-to-Many' (Many vs. Many, MvM) splitting strategy, and is a common technology for solving the problem of multi-classification in machine learning.
(2-5) finally, it is possible to output using the error correction output code
Figure BDA0003118834170000113
And each partition needs to be constructed with two classification models. The conditional probability distribution of the binomial logistic regression model is:
Figure BDA0003118834170000114
Figure BDA0003118834170000115
the following steps are mainly to find the parameter w, where xi＝(mi,1,mi,2,…,mi,n1) is equivalent to mi,w＝(w1,w2,…,wmAnd n) is a model parameter.
(2-6) the present invention solves the above problem using maximum likelihood estimation, for a given fault bank, assuming first:
P(yi＝1|xi)＝π(xi)
then it can be obtained:
P(yi＝0|xi)＝1-π(xi)
the likelihood function is:
Figure BDA0003118834170000116
the log-likelihood function is:
Figure BDA0003118834170000117
Figure BDA0003118834170000121
a two-term log-probability regression model can be obtained as:
w*＝argwmin L(w)
(2-7) order
Figure BDA0003118834170000122
Const bivariate log probability regression model f can be obtained through the steps (2-6)1,f2,…,fconst. Log probability model f for the kth fault classjThe result is calculated as
Figure BDA0003118834170000123
Here, j is 1. ltoreq. const, and
Figure BDA0003118834170000124
binary error correction output codes for K failure classes can be obtained as shown in table 1 below.
TABLE 1
Figure BDA0003118834170000125
(2-8) the error correction output code for each class is obtained from Table 1, and the error correction output code for class k is
Figure BDA0003118834170000126
The error correction output code for any sample is
Figure BDA0003118834170000127
Distance between themThe separation is as follows:
Figure BDA0003118834170000128
(2-9) by the above formula, a multinomial Log-probability regression model can be obtained
y*＝argymin{dist(y,temp)}
Wherein y is all categories, y is more than or equal to 1 and less than or equal to K, and temp is input data to be classified.
(2-10) the new fault detection model replaces the current fault detection model G, and new problems of the user are detected.
Examples
The embodiment uses a type a industrial internet device as an example. The specific implementation flow is shown in fig. 2.
When the equipment leaves the factory, the fault library M is { M ═ M1,m2,m31, wherein the 1 st failure description information m1＝{m1,1,m1,2,m1,3,y1}, 2 nd fault description information m2＝{m2,1,m2,2,m2,2,y2}, 3 rd failure description information m3＝{m3,1,m3,2,m3,3,y3}; method library W ═ { W1,w2,w3H, with 1 st failure solution w1＝{er1,med1}, 2 nd failure solution w2＝{er2,med2}, item 3 failure resolution w3＝{er3,med3}. Class of fault er1And the category y to which the fault belongs1Corresponding to the type of fault er2And the category y to which the fault belongs2Corresponding to the type of fault er3And the category y to which the fault belongs3And (7) corresponding.
As shown in fig. 2, when a type a industrial internet device fails, the sensor collects internal device information to form a failure vector mtempAnd the parallel connection network is uploaded to an industrial internet platform. The industrial internet platform meetingJudging the type of the fault according to the fault detection model G, if the fault is a hard fault, storing information, and automatically informing a professional maintainer to maintain the system; if the fault type is a soft fault, the industrial internet platform generates two-dimension code information for multimedia content describing a fault solution, and simultaneously sends the two-dimension code information and the fault code to the industrial internet equipment.
The user can obtain the corresponding solution and the video by scanning the two-dimensional code or inputting the fault code. If the scheme successfully solves the fault of the user, the fault detection and processing flow is finished; if the failure solution does not solve the problem with the device. The user can feed back the information to the remote industrial internet platform. After the remote industrial internet platform obtains user feedback, a manufacturer is firstly informed to send a professional to go to the place where the industrial internet equipment is located to overhaul the equipment, and the fault vector m is markedtemp. After the professional detects and eliminates the fault, the professional submits the solution of the fault to the industrial Internet control platform through the background. The industrial Internet platform generates a fault code er for the schemetempAnd multimedia content medtemp. Then, the industrial internet platform processes the data according to the following steps:
(1) the industrial internet platform can transmit the fault vector mtempFault code devicetempAnd multimedia content medtempAdding a record into the temporary sample library, wherein the record has a pattern of mtemp,ertemp,medtemp,counter]. Wherein, counterAnd adding 1 to the original number of records corresponding to the current fault code.
(2) The industrial Internet platform can continuously collect the use condition of the user, record the unsolvable faults, and count the number of the existing recordserWhen the value is larger than the threshold value gamma, the industrial internet platform starts the optimization process of the detection model G; the industrial Internet platform adds the records meeting the requirements in the temporary sample library into a fault library to form mtemp,1,mtemp,2,…,mtemp,20,ytempRecord of where ytempAnd ertempIs in a one-to-one correspondence.
(3) The industrial internet platform starts a training process, a new fault detection model obtained through the logarithm probability regression model replaces the current fault detection model G, and the new model is used for processing after the next fault occurs.
As can be seen from the foregoing technical solutions, an embodiment of the present invention provides a remote intelligent detection and processing method for supporting an industrial internet device, including: step 1, constructing a fault library, a method library and a fault detection model of the equipment based on equipment delivery information and equipment test data of the industrial internet equipment. The fault library is the fault description information of the equipment and the fault category to which the fault belongs. The method library is a set of fault solving methods, each fault solving method is composed of two parts, one part is an error code, and the other part is a concrete description for solving the problem corresponding to the error code and is generally multimedia content; step 2, after the industrial internet equipment fails, the fault description information is sent to the industrial internet platform; step 3, after the industrial internet platform obtains the fault description information, obtaining a fault solution in the method library through a fault detection model, and feeding back the fault solution to the industrial internet equipment in the form of a two-dimensional code and a fault code; step 4, a user can obtain a fault solution by scanning the two-dimensional code or inputting the fault code; and 5, processing the fault of the industrial Internet equipment by the user through the fault solution, feeding the fault information back to the industrial Internet platform by the user when the fault solution cannot solve the fault, and starting the optimization of the fault detection model after the platform collects enough information.
In specific implementation, the present invention further provides a computer storage medium, where the computer storage medium may store a program, and when the program is executed, the program may include some or all of the steps in each embodiment of the remote intelligent detection and processing method for supporting the industrial internet device provided by the present invention. The storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM) or a Random Access Memory (RAM).
The invention provides a remote intelligent detection and processing method supporting industrial internet equipment, and a plurality of methods and ways for implementing the technical scheme are provided, the above description is only a preferred embodiment of the invention, and it should be noted that, for those skilled in the art, a plurality of improvements and modifications can be made without departing from the principle of the invention, and these improvements and modifications should also be regarded as the protection scope of the invention. All the components not specified in the present embodiment can be realized by the prior art.

Claims (5)

1. A remote intelligent detection and processing method supporting industrial Internet equipment is characterized by comprising the following steps:
step 1, constructing a fault library M, a method library W and a fault detection model G of equipment based on equipment delivery information and equipment test data of industrial internet equipment;
step 2, after the industrial internet equipment breaks down, the fault description information is sent to an industrial internet platform;
step 3, after the industrial internet platform obtains the fault description information, obtaining a fault solution in the method library through a fault detection model, and feeding back the fault solution to the industrial internet equipment in the form of a two-dimensional code and a fault code;
step 4, the user obtains a fault solution by scanning the two-dimensional code or inputting the fault code;
and 5, processing the fault of the industrial Internet equipment by the user through a fault solution.
2. The method according to claim 1, wherein in step 1, the fault library M contains fault description information and a fault class to which the fault belongs, and M ═ is (M ═ is1，m2，m3… }, where the ith fault description information mi＝{mi，1，mi，2，…，mi，n，yiOne piece of fault description information consists of n +1 parameters, and the first n parameters are description information of n dimensions of the current industrial internet equipment, namely mi，nDescription information of nth dimension for industrial internet device, yiIs the category to which the fault belongs;
the method library W contains solutions corresponding to faults, W ═ W1，w2，…，wlH, wherein the ith failure solution wi＝{eri，medi}，eriIndicating a fault code which is a string of characters; mediRepresenting a piece of multimedia content, said multimedia content presenting an eriThe specific solution to the described failure; fault code deviceiAnd the category y to which the fault belongsiThe corresponding relation exists, and the fault detection model G can describe the information m according to the faultiCalculating the fault category y to which the current fault belongsiFind out the corresponding fault eriAnd locating the corresponding fault solution in the method library.
3. The method according to claim 2, wherein in step 2, the industrial internet equipment is equipped with a sensor to collect information of the equipment in real time; when equipment fails, the sensor collects internal equipment information to form a fault vector mtempTransmitting the information to an industrial internet platform at the nearest position; the industrial internet platform stores a fault library M, a method library W and a fault detection model G.
4. The method of claim 3, wherein step 4 comprises: the two-dimension code comprises multimedia content, a user can obtain description content of a fault solution by scanning the two-dimension code, when the description content comprises video and text content, the video and the text are returned to the user, and if the description content only comprises the text content, the text content is only returned; if the industrial internet device can only display the fault code, the user obtains a fault solution by inputting the fault code.
5. The method according to claim 4, wherein in step 5, the user processes the fault of the industrial internet device through the fault solution, including the following two cases:
in the first situation, a user obtains a fault solution by inputting a fault code or scanning a two-dimensional code, successfully solves the problem of the industrial internet equipment through the fault solution, and ends the fault detection and processing flow;
in the second situation, the problem that the equipment appears cannot be solved through the fault solution scheme obtained by inputting the fault code or scanning the two-dimensional code by the user, the user feeds back to the remote industrial internet platform through the feedback window, and after the remote industrial internet platform receives the user feedback, the user can firstly inform the manufacturer to send a professional to go to the place where the industrial internet equipment is located to overhaul the equipment, and mark the fault vector mtemp(ii) a After the professional personnel detects and eliminates the fault, the fault solution is submitted to an industrial internet control platform through a background; the industrial internet platform can generate fault codes er for the solutions of the faultstempAnd multimedia content medtempThen, the industrial internet platform can process according to the following steps:
step 5-1, the industrial internet platform can convert the fault vector mtempFault code devicetempAnd multimedia content medtempAdding a formed record into a temporary sample library, wherein the record has a record style of mtemp，ertemp，medtemp，counter]Therein, counterThe number of records corresponding to the current fault code is recorded;
step 5-2, the industrial Internet platform can continuously collect the use condition of the user and record the fault which can not be solved, when the number of the existing records is countederWhen the value is larger than the threshold value gamma, the industrial internet platform starts the optimization process of the fault detection model G;
and 5-3, adding the records meeting the requirements in the temporary sample library into a fault library by the industrial internet platform to form a record { m }temp，1，mtemp，2，…，mtemp，n，ytempThen the industrial internet platform starts the optimization process of the fault detection model G, wherein mtemp，nFor recording the description information of the nth dimension of the industrial Internet appliance in question, ytempRecording the category of the industrial internet equipment fault in the record;
and 5-4, after the data in the temporary sample library is added into the fault library, setting the current common K-type faults, and dividing the K-type faults into two types, wherein the K-type faults are
Figure FDA0003118834160000021
Sample as a good example, the rest
Figure FDA0003118834160000022
The individual example is a counterexample;
step 5-5, utilizing error correction output code to make final output
Figure FDA0003118834160000023
For each partition, a two-classification model needs to be constructed, and the conditional probability distribution of the binomial logistic regression model is as follows:
Figure FDA0003118834160000031
Figure FDA0003118834160000032
wherein, P (y)i＝1|xi) When the sample is xiA probability that the category is a first category;
P(yi＝0|xi) When the sample is xiThen, the class is the probability of the second class;
xi＝(mi，1，mi，2，…，mi，n1) is equivalent to miI.e. the ith fault description information, yiFor the class to which the fault belongs, w ═ w1，w2，…，wNB) is a model parameter, wNThe model parameter of the Nth sample is N, the total number of samples is N, and b is a constant parameter;
and 5-6, solving the formula in the step 5-5 by using maximum likelihood estimation, and for a given fault library, firstly setting:
P(yi＝1|xi)＝π(xi)
obtaining:
P(yi＝0|xi)＝1-π(xi)
the likelihood function is:
Figure FDA0003118834160000033
the log-likelihood function L (w) is:
Figure FDA0003118834160000034
obtaining a bivariate logarithm probability regression model w*：
w*＝argw min L(w)；
Step 5-7, order
Figure FDA0003118834160000035
Obtaining const bivariate logarithm probability regression model f through steps 5-61，f2，…，fconst，fconstRepresenting the const second log-probability regression model, for the k fault class, passing the jth second log-probability regression model fjThe result is calculated as
Figure FDA0003118834160000036
J is 1. ltoreq. const, and
Figure FDA0003118834160000037
obtaining binary error correction output codes of K fault categories;
step 5-8, obtaining the error correction output code of each category through step 5-7, wherein the error correction output code of the category k is
Figure FDA0003118834160000041
Wherein
Figure FDA0003118834160000042
Indicating that for the kth failure class, the second-term log-probability regression model f is passedconstA result of the calculation; the error correction output code for any sample is
Figure FDA0003118834160000043
Wherein
Figure FDA0003118834160000044
Indicating that for the temp fault class, the second term log probability regression model f is passedconstThe calculated result, error correction output code ec of any sampletempThe distance dist (k, temp) from the kth class of error correction output codes is:
Figure FDA0003118834160000045
wherein
Figure FDA0003118834160000046
Indicating that for the kth fault class, through the s second term log-probability regression model fsA result of the calculation;
Figure FDA0003118834160000047
indicating that for the temp fault class, through the s second log probability regression model fsA result of the calculation;
step 5-9, obtaining a multi-term logarithm probability regression model y*：
y*＝argy min{dist(y，temp)}；
Wherein y is all categories, y is more than or equal to 1 and less than or equal to K, temp is input data to be classified, dist (y, temp) represents the distance between the error correction output code of the temp category and the error correction output code of the y category;
and 5-10, replacing the current fault detection model G by the new fault detection model to detect new problems of the user.
CN202110670108.1A 2021-06-17 2021-06-17 Remote intelligent detection and processing method supporting industrial Internet equipment Active CN113469833B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110670108.1A CN113469833B (en) 2021-06-17 2021-06-17 Remote intelligent detection and processing method supporting industrial Internet equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110670108.1A CN113469833B (en) 2021-06-17 2021-06-17 Remote intelligent detection and processing method supporting industrial Internet equipment

Publications (2)

Publication Number Publication Date
CN113469833A true CN113469833A (en) 2021-10-01
CN113469833B CN113469833B (en) 2023-08-29

Family

ID=77870183

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110670108.1A Active CN113469833B (en) 2021-06-17 2021-06-17 Remote intelligent detection and processing method supporting industrial Internet equipment

Country Status (1)

Country Link
CN (1) CN113469833B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CA2455457A1 (en) * 2003-05-12 2004-11-12 Simmonds Precision Products, Inc. Wire fault detection
CN106054858A (en) * 2016-05-27 2016-10-26 大连楼兰科技股份有限公司 Decision tree classification and fault code classification-based vehicle remote diagnosis and spare part retrieval method
CN106095785A (en) * 2016-05-27 2016-11-09 大连楼兰科技股份有限公司 DTC based on decision tree classification diagnosis vehicle work item and spare part search method
CN112421454A (en) * 2020-09-23 2021-02-26 国家电网有限公司 Substation equipment diagnosis and maintenance platform based on Internet of things

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CA2455457A1 (en) * 2003-05-12 2004-11-12 Simmonds Precision Products, Inc. Wire fault detection
CN106054858A (en) * 2016-05-27 2016-10-26 大连楼兰科技股份有限公司 Decision tree classification and fault code classification-based vehicle remote diagnosis and spare part retrieval method
CN106095785A (en) * 2016-05-27 2016-11-09 大连楼兰科技股份有限公司 DTC based on decision tree classification diagnosis vehicle work item and spare part search method
CN112421454A (en) * 2020-09-23 2021-02-26 国家电网有限公司 Substation equipment diagnosis and maintenance platform based on Internet of things

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
HONGLI NIE等: "Data security over wireless transmission for enterprise multimedia security with fountain codes", 《MULTIMEDIA TOOLS AND APPLICATIONS》, vol. 79, pages 10781, XP037110149, DOI: 10.1007/s11042-019-08479-z *
雷蕾等: "ECOC多类分类研究综述", 《电子学报》, vol. 42, no. 9, pages 1794 - 1800 *

Also Published As

Publication number Publication date
CN113469833B (en) 2023-08-29

Similar Documents

Publication Publication Date Title
CN108470022B (en) Intelligent work order quality inspection method based on operation and maintenance management
CN110689148A (en) Metering equipment fault detection method based on AR technology
CN115719283A (en) Intelligent accounting management system
CN117234785B (en) Centralized control platform error analysis system based on artificial intelligence self-query
CN110009260B (en) Method and device for processing shift-switching information of sewage treatment plant
CN113469833A (en) Remote intelligent detection and processing method supporting industrial internet equipment
CN116452212B (en) Intelligent customer service commodity knowledge base information management method and system
CN110766248B (en) Workshop artificial factor reliability assessment method based on SHEL and interval intuitionistic fuzzy assessment
CN114625901B (en) Multi-algorithm integration method and device
CN112801537B (en) Non-stop accident analysis method, system, medium and equipment for power generation enterprise
CN104238419B (en) Hand-held energy-saving monitoring intelligent work system and method for work
CN113076217A (en) Disk fault prediction method based on domestic platform
CN116993307B (en) Collaborative office method and system with artificial intelligence learning capability
CN117649105B (en) Substation work ticket intelligent ticket filling method based on RPA flow automation
CN117787680A (en) Business process mining method and equipment based on management system
CN117036334A (en) Industrial anomaly detection and positioning method based on global attention flow model
CN117236642A (en) Two-ticket spot inspection method, two-ticket spot inspection device, computer equipment and storage medium
JP2024037666A (en) Vulnerability mining method and system for competitive industrial control systems
CN117744714A (en) Callback-based deep neural network error positioning method
CN117370284A (en) Log error data tracking method and system
CN114862148A (en) Information supervision project big data quality detection system and method
CN115527228A (en) Classification method, device, equipment and medium based on self-feedback service platform
CN115455824A (en) Process simulation method and system based on cross attribute influence
CN117649105A (en) Substation work ticket intelligent ticket filling method and system based on RPA process automation
CN113010709A (en) Film and television scenario platform chance-passing quantitative evaluation system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant