CN112633900A - Industrial Internet of things data verification method based on machine learning - Google Patents

Industrial Internet of things data verification method based on machine learning Download PDF

Info

Publication number
CN112633900A
CN112633900A CN202011481568.1A CN202011481568A CN112633900A CN 112633900 A CN112633900 A CN 112633900A CN 202011481568 A CN202011481568 A CN 202011481568A CN 112633900 A CN112633900 A CN 112633900A
Authority
CN
China
Prior art keywords
data
training
machine learning
verified
things
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.)
Pending
Application number
CN202011481568.1A
Other languages
Chinese (zh)
Inventor
陈曦
张晓枫
王胜
吕志
王慎
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.)
Beijing Guodiantong Network Technology Co Ltd
Original Assignee
Beijing Guodiantong Network Technology Co Ltd
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 Beijing Guodiantong Network Technology Co Ltd filed Critical Beijing Guodiantong Network Technology Co Ltd
Priority to CN202011481568.1A priority Critical patent/CN112633900A/en
Publication of CN112633900A publication Critical patent/CN112633900A/en
Pending legal-status Critical Current

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
    • G06Q30/00—Commerce
    • G06Q30/018—Certifying business or products
    • G06Q30/0185—Product, service or business identity fraud
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Abstract

The invention belongs to the technical field of industrial Internet of things, and discloses an industrial Internet of things data verification method based on machine learning, which comprises the following steps: determining data items, for each data item: respectively obtaining m training samples to form a training set; establishing an initial learning model; training the initial learning model by adopting a training set to obtain a trained strong classifier; and acquiring data to be verified, using the data to be verified as the input of the trained strong classifier, performing true and false classification on the data to be verified, outputting the classification result of the data to be verified, and completing the verification of the data. The invention obtains the data verification model by machine learning training, thereby identifying the authenticity of the data to be verified, directly sending abnormal data to the operation and maintenance department for analysis and processing, preventing the maliciously tampered data from reaching the category management center, and ensuring the authenticity of the data in the subsequent links.

Description

Industrial Internet of things data verification method based on machine learning
Technical Field
The invention relates to the technical field of industrial Internet of things, in particular to an industrial Internet of things data verification method based on machine learning.
Background
The intelligent Internet of things platform for the electrical equipment can strengthen the real-time interaction coordination capability of the electrical equipment, construct service ecology with complementary advantages and mutual benefits and win-win, is driven by internal and external requirements, takes a supply chain of the electrical equipment as a main line, and leads core interest requirements of cooperation partners such as a supply side and a matching service third party and the like by a demand side to form a social coordination production mode and an organization mode.
The intelligent internet of things data gateway is used as an extension of an EIP (electronic information platform) system on a supplier side and bears a bridge for connecting a factory data acquisition system and a class management center of the supplier, the factory data acquisition system is connected with the intelligent internet of things data gateway through an internet of things data docking channel, and the intelligent internet of things data gateway is connected with the class management center through an internet of things data forwarding channel, so that raw material inspection data, production process and process inspection data, product test data, key process video data and the like acquired by the factory side are uploaded to the class management center of the internet of things through the intelligent internet of things data gateway.
In the process, the authenticity of the data of the internet of things collected from the supplier side is very important for the operation of the EIP platform and the application of each subsequent link. Therefore, a perfect data authenticity guarantee mechanism needs to be established, and in the prior art, the reality and the effectiveness of the internet of things data provided by a supplier are ensured by matching with corresponding management measures through various means such as data active capture, data fidelity and the like. Even in this case, it is still not guaranteed that data is maliciously tampered in the transmission process, and therefore, data in each link needs to be verified, and authenticity of data uploaded to the category management center is guaranteed.
Disclosure of Invention
Aiming at the problems in the prior art, the invention aims to provide a machine learning-based industrial internet of things data verification method, which is characterized in that a data verification model is obtained by utilizing machine learning training, so that authenticity of data to be verified is distinguished, abnormal data is directly sent to an operation and maintenance department for analysis and processing, malicious tampered data is prevented from reaching a category management center, and authenticity of data in a subsequent link is ensured.
In order to achieve the purpose, the invention is realized by adopting the following technical scheme.
The industrial Internet of things data verification method based on machine learning comprises the following steps:
step 1, determining data items, and for each data item: respectively obtaining m training samples to form a training set; establishing an initial learning model;
wherein the training samples are historical product inspection performance data, equipment parameters or video data provided by a supplier; m is greater than 3000;
step 2, training the initial learning model by adopting a training set to obtain a trained strong classifier;
and 3, acquiring data to be verified, using the data to be verified as the input of the trained strong classifier, performing true-false classification on the data to be verified, outputting a classification result of the data to be verified, and completing verification of the data.
Further, the data items include raw material inspection data, production process inspection data, product test data, and key process video data.
Further, the initial learning model is composed of K weak learners, and each weak learner is a convolutional neural network.
Further, the training of the initial learning model by using the training set specifically includes:
2.1 initializing training sample weight to D1；
Figure BDA0002838269000000021
2.2, using a weight DkInputting the training sample into the Kth weak learner to carry out data training to obtain the Kth weak classifier Gk(x)；
2.3, calculating the Kth weak classifier Gk(x) Classification error rate of (2):
Figure BDA0002838269000000031
wherein (x)i，yi) Is the ith training sample; i (G)k(xi)≠yi) Error rate of the ith training sample on the Kth weak classifier;
2.4, calculating the coefficients of the weak classifier:
Figure BDA0002838269000000032
2.5, updating the weight distribution of the sample set based on the classification error:
Figure BDA0002838269000000033
Figure BDA0002838269000000034
wherein Z iskIs a normalization factor;
2.6, repeating the steps 2.2-2.5 until the K weak learner finishes the data training;
2.7, constructing the trained strong classifier as follows:
Figure BDA0002838269000000035
furthermore, the updating of the weight distribution of the sample set is specifically to adopt a forward step-by-step learning algorithm and utilize the result f of the previous strong classifierk-1(x) And the current weak learner G (x) to update the model of the current strong classifier; the loss function is:
Figure BDA0002838269000000036
where α is the coefficient of the current weak learner.
Further, the data to be verified is a data segment obtained through data capture or fidelity data after data fidelity, and the data segment and the fidelity data respectively contain product performance detection data or equipment parameters provided by a supplier.
Further, configuring data items, frequency and ways of data sampling through an interactive interface; and training a strong classifier correspondingly for each data item.
Further, in step 3, when the output result is true, the output result is interpreted as real data, and the real data is directly transmitted to the class management center;
when the output result is false, the data is abnormal data, the abnormal data is transmitted to an operation and maintenance group for processing, and the processing result is fed back to the gateway and the supplier; the supplier further screens the abnormal data, and the gateway does not further transmit the abnormal data.
Compared with the prior art, the invention has the beneficial effects that:
the method adopts a machine learning algorithm to learn various data provided by suppliers in the industrial Internet of things, and correspondingly obtains a strong classifier of each data; and in the subsequent process of capturing data and data fidelity by the gateway, the captured data or the data after fidelity is subjected to authenticity identification through the strong classifier, and the data with a false judgment result is pushed to an operation and maintenance group for analysis and processing. The invention introduces machine learning into the industrial Internet of things to identify the authenticity of the data captured by the gateway or the data after fidelity, thereby ensuring the correctness of the data transmitted to the class management center and preventing the maliciously tampered data from entering the class management center.
Drawings
The invention is described in further detail below with reference to the figures and specific embodiments.
Fig. 1 is a flowchart of a machine learning-based industrial internet of things data verification method provided by an embodiment of the invention;
FIG. 2 is a flow chart of machine learning according to an embodiment of the present invention.
Detailed Description
Embodiments of the present invention will be described in detail below with reference to examples, but it will be understood by those skilled in the art that the following examples are only illustrative of the present invention and should not be construed as limiting the scope of the present invention.
Referring to fig. 1, the invention provides a machine learning-based industrial internet of things data verification method, which includes the following steps:
step 1, determining data items, and for each data item: respectively obtaining m training samples to form a training set; establishing an initial learning model;
specifically, the data items include raw material inspection data, production process inspection data, product test data, and key process video data. Each data item is more important production process data on the supplier side; each data item needs to train a strong classifier to perform authenticity identification on gateway captured data or data after fidelity. The training sample of each data item is historical product inspection performance data, equipment parameters or video data provided by a supplier; in this embodiment, the historical inspection performance data of 30000 products is selected as a training sample to form a training sample set S { (x)1,y1),…,(xm,ym)}.
The initial learning model consists of K weak learners, and each weak learner is a convolutional neural network.
Step 2, training the initial learning model by adopting a training set to obtain a trained strong classifier;
2.1 initializing training sample weight to D1；
Figure BDA0002838269000000051
2.2, 2 … K for K ═ 1;
by having a weight DkTraining sampleInputting the K weak learner to perform data training to obtain the K weak classifier Gk(x)；
2.3, calculating the Kth weak classifier Gk(x) Classification error rate of (2):
Figure BDA0002838269000000061
wherein (x)i，yi) Is the ith training sample; i (G)k(xi)≠yi) Error rate of the ith training sample on the Kth weak classifier;
2.4, calculating the coefficients of the weak classifier:
Figure BDA0002838269000000062
2.5, updating the weight distribution of the sample set based on the classification error:
Figure BDA0002838269000000063
Figure BDA0002838269000000064
wherein Z iskIs a normalization factor;
2.6, repeating the steps 2.2-2.5 until the K weak learner finishes the data training;
2.7, constructing the trained strong classifier as follows:
Figure BDA0002838269000000065
further, updating the weight distribution of the sample set, specifically adopting a forward step-by-step learning algorithm and utilizing the result f of the previous strong classifierk-1(x) And the current weak learner G (x) to update the current strong classifierThe model of (2); the loss function is:
Figure BDA0002838269000000066
where α is the coefficient of the current weak learner.
And 3, acquiring data to be verified, using the data to be verified as the input of the trained strong classifier, performing true-false classification on the data to be verified, outputting a classification result of the data to be verified, and completing verification of the data.
The data to be verified in this embodiment is a data segment obtained by data capture or fidelity data after data fidelity, and the data segment and the fidelity data respectively contain product performance detection data, device parameters or video data provided by a supplier.
In the embodiment, the data items, the frequency and the way of data sampling are configured through an interactive interface; and training a strong classifier correspondingly for each data item.
And when the output result is true, the real data is indicated, and the real data is directly transmitted to the class management center for subsequent management.
When the output result is false, the data is abnormal data, the abnormal data is transmitted to an operation and maintenance group for analysis and processing, and the processing result is fed back to the gateway and the supplier; the supplier further screens the abnormal data, and the gateway does not further transmit the abnormal data.
The authenticity of data received by the product management center is ensured through the operation.
In fact, each data item corresponds to a standard value range, and after configuration is successful, data fidelity and verification service for the data item can be started, and the service can be manually stopped at a later stage. After the application service is started, on one hand, the gateway performs data capture and calculation according to the configured frequency and requirements, and updates the standard value range of the data item; on the other hand, when the gateway relates to the related data item information in the process of capturing the data, the gateway compares the related data item information with the corresponding standard value range to judge whether the related data item information exceeds the fluctuation range.
And if the abnormal result data exceeds the value range, pushing the abnormal result data to an operation and maintenance group for processing, and tracking a processing result. The analysis of the processing result is an important basis and support for subsequent continuous upgrading and fidelity model algorithm optimization.
In the invention, the authenticity of the data of the Internet of things collected from the supplier side is very important for the operation of the EIP platform and the application of each subsequent link. Therefore, a perfect data authenticity guarantee mechanism needs to be established, and the data of the production and manufacturing, the delivery test, the video monitoring and other things provided by a supplier are ensured to be real and effective by matching with corresponding management measures through various means such as data active capture, machine learning verification, network connection technology application, relevance verification of equipment parameters, a data fidelity verification model and the like.
Although the present invention has been described in detail in this specification with reference to specific embodiments and illustrative embodiments, it will be apparent to those skilled in the art that modifications and improvements can be made thereto based on the present invention. Accordingly, such modifications and improvements are intended to be within the scope of the invention as claimed.

Claims (8)

1. The industrial Internet of things data verification method based on machine learning is characterized by comprising the following steps:
step 1, determining data items, and for each data item: respectively obtaining m training samples to form a training set; establishing an initial learning model;
wherein the training samples are historical product inspection performance data, equipment parameters or video data provided by a supplier; m is more than 3000;
step 2, training the initial learning model by adopting a training set to obtain a trained strong classifier;
and 3, acquiring data to be verified, using the data to be verified as the input of the trained strong classifier, performing true-false classification on the data to be verified, outputting a classification result of the data to be verified, and completing verification of the data.
2. The machine learning-based industrial internet of things data validation method of claim 1, wherein the data items comprise raw material inspection data, production process inspection data, product test data and key process video data.
3. The machine learning-based industrial internet of things data verification method according to claim 1, wherein the initial learning model is composed of K weak learners, and each weak learner is a convolutional neural network.
4. The machine learning-based industrial internet of things data verification method according to claim 3, wherein the initial learning model is trained by using a training set, specifically:
2.1 initializing training sample weight to D1；
D1＝(w11，W12…W1m)；
Figure FDA0002838268990000011
2.2, using a weight DkInputting the training sample into the Kth weak learner to carry out data training to obtain the Kth weak classifier Gk(x)；
2.3, calculating the Kth weak classifier Gk(x) Classification error rate of (2):
Figure FDA0002838268990000021
wherein (x)i，yi) Is the ith training sample; i (G)k(xi)≠yi) Error rate of the ith training sample on the Kth weak classifier;
2.4, calculating the coefficients of the weak classifier:
Figure FDA0002838268990000022
2.5, updating the weight distribution of the sample set based on the classification error:
Figure FDA0002838268990000023
Figure FDA0002838268990000024
wherein Z iskIs a normalization factor;
2.6, repeating the steps 2.2-2.5 until the K weak learner finishes the data training;
2.7, constructing the trained strong classifier as follows:
Figure FDA0002838268990000025
5. the industrial Internet of things data verification method based on machine learning as claimed in claim 4, wherein the weight distribution of the sample set is updated by using the result f of the previous strong classifier by adopting a forward step-by-step learning algorithmk-1(x) And the current weak learner G (x) to update the model of the current strong classifier; the loss function is:
Figure FDA0002838268990000026
where α is the coefficient of the current weak learner.
6. The machine learning-based industrial internet of things data verification method according to claim 1, wherein the data to be verified is a data segment obtained through data capture or fidelity data after data fidelity, and the data segment and the fidelity data respectively contain product performance detection data or equipment parameters provided by a supplier.
7. The machine learning-based industrial internet of things data verification method according to claim 1, characterized in that data items, frequency and ways of data sampling are configured through an interactive interface; and training a strong classifier correspondingly for each data item.
8. The machine learning-based industrial internet of things data verification method according to claim 1, wherein in step 3, when the output result is true, the output result is interpreted as true data, and the true data is directly transmitted to a class management center;
when the output result is false, the data is abnormal data, the abnormal data is transmitted to an operation and maintenance group for processing, and the processing result is fed back to the gateway and the supplier; the supplier further screens the abnormal data, and the gateway does not further transmit the abnormal data.
CN202011481568.1A 2020-12-16 2020-12-16 Industrial Internet of things data verification method based on machine learning Pending CN112633900A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011481568.1A CN112633900A (en) 2020-12-16 2020-12-16 Industrial Internet of things data verification method based on machine learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011481568.1A CN112633900A (en) 2020-12-16 2020-12-16 Industrial Internet of things data verification method based on machine learning

Publications (1)

Publication Number Publication Date
CN112633900A true CN112633900A (en) 2021-04-09

Family

ID=75313294

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011481568.1A Pending CN112633900A (en) 2020-12-16 2020-12-16 Industrial Internet of things data verification method based on machine learning

Country Status (1)

Country Link
CN (1) CN112633900A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113159448A (en) * 2021-05-12 2021-07-23 烟台应辉智能科技有限公司 Automatic analysis and discrimination method based on environmental protection big data

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103559508A (en) * 2013-11-05 2014-02-05 福建省视通光电网络有限公司 Video vehicle detection method based on continuous Adaboost
CN108596268A (en) * 2018-05-03 2018-09-28 湖南大学 A kind of data classification method
CN109726767A (en) * 2019-01-13 2019-05-07 胡燕祝 A kind of perceptron network data classification method based on AdaBoost algorithm
WO2020151173A1 (en) * 2019-01-25 2020-07-30 深信服科技股份有限公司 Webpage tampering detection method and related apparatus
CN111860658A (en) * 2020-07-24 2020-10-30 华北电力大学（保定） Transformer fault diagnosis method based on cost sensitivity and integrated learning

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103559508A (en) * 2013-11-05 2014-02-05 福建省视通光电网络有限公司 Video vehicle detection method based on continuous Adaboost
CN108596268A (en) * 2018-05-03 2018-09-28 湖南大学 A kind of data classification method
CN109726767A (en) * 2019-01-13 2019-05-07 胡燕祝 A kind of perceptron network data classification method based on AdaBoost algorithm
WO2020151173A1 (en) * 2019-01-25 2020-07-30 深信服科技股份有限公司 Webpage tampering detection method and related apparatus
CN111860658A (en) * 2020-07-24 2020-10-30 华北电力大学（保定） Transformer fault diagnosis method based on cost sensitivity and integrated learning

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113159448A (en) * 2021-05-12 2021-07-23 烟台应辉智能科技有限公司 Automatic analysis and discrimination method based on environmental protection big data

Similar Documents

Publication Publication Date Title
CN111157682A (en) Air quality monitoring and predicting system and method
CN111431986A (en) Industrial intelligent quality inspection system based on 5G and AI cloud edge cooperation
US11615166B2 (en) System and method for classifying image data
CN110138786A (en) Web method for detecting abnormality and system based on SMOTETomek and LightGBM
CN110298662A (en) Transaction repeats the automated detection method and device submitted
CN112633900A (en) Industrial Internet of things data verification method based on machine learning
CN114495983A (en) Equipment failure voiceprint monitoring system based on cloud edge collaboration
CN113313280A (en) Cloud platform inspection method, electronic equipment and nonvolatile storage medium
CN108538290A (en) A kind of intelligent home furnishing control method based on audio signal detection
CN114895634A (en) Product production line automatic control system based on machine vision
CN112561097B (en) Bearing monitoring method and system based on cloud and mist edge cooperation
CN106446791A (en) Smart city public monitoring system
CN111817935B (en) Internet intelligent home data processing method and system
CN116560322B (en) Production control method and system for assembly production line
CN116756518A (en) Cloud edge cooperative power defect identification method and related device based on federal learning
CN115659214A (en) Energy industry data credible evaluation method based on PaaS platform
CN116308679A (en) Supply chain abnormal order processing method, device, equipment and storage medium
CN110320802A (en) Complication system signal sequence recognition methods based on data visualization
CN114757234A (en) Efficient big data intelligent regulation and control system
CN203120069U (en) Video quality detecting device and system
CN116094778B (en) Real-time monitoring system based on Internet
CN113596229B (en) Digital home smart phone based on cloud architecture
CN114722061B (en) Data processing method and device, equipment and computer readable storage medium
CN115361256B (en) Edge computing intelligent gateway oriented to intelligent security monitoring field and implementation method
CN111062468B (en) Training method and system for generating network, and image generation method and device

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
RJ01 Rejection of invention patent application after publication

Application publication date: 20210409

RJ01 Rejection of invention patent application after publication