CN108846833A - A method of hard disk failure is diagnosed based on TensorFlow image recognition - Google Patents
A method of hard disk failure is diagnosed based on TensorFlow image recognition Download PDFInfo
- Publication number
- CN108846833A CN108846833A CN201810542914.9A CN201810542914A CN108846833A CN 108846833 A CN108846833 A CN 108846833A CN 201810542914 A CN201810542914 A CN 201810542914A CN 108846833 A CN108846833 A CN 108846833A
- Authority
- CN
- China
- Prior art keywords
- hard disk
- image
- angle
- server
- tensorflow
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000012937 correction Methods 0.000 claims abstract description 22
- 238000003745 diagnosis Methods 0.000 claims abstract description 20
- 238000012545 processing Methods 0.000 claims abstract description 9
- 238000012952 Resampling Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims 1
- 238000012423 maintenance Methods 0.000 description 8
- 238000012544 monitoring process Methods 0.000 description 7
- 239000011159 matrix material Substances 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000013475 authorization Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/0008—Industrial image inspection checking presence/absence
-
- G06T5/80—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
- G06T7/337—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods involving reference images or patches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/35—Determination of transform parameters for the alignment of images, i.e. image registration using statistical methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30148—Semiconductor; IC; Wafer
Abstract
The invention discloses a kind of methods based on TensorFlow image recognition diagnosis hard disk failure, include the following steps：Terminal obtains the multi-angle image of presence states to server hard disc；It is transmitted to server end and carries out relative angle normalized, recycle geometric correction method to be registrated multi-angle image, obtain revised hard disk images；Revised hard disk images information is handled by TensorFlow；Server is stored with hard disk initialization normal condition image data；Judgement is compared with initialization normal condition image data in hard disk presence states image data；The presence states information of hard disk is fed back into terminal.The present invention utilizes multi-angle hard disk presence states image, is registrated in conjunction with geometric correction processing, is fused to accurate hard disk images, makes hard disk failure diagnosis process quickly, accurately in conjunction with TensorFlow Intelligent treatment, realizes full-automatic operation.
Description
Technical field
The present invention relates to the monitoring methods of a kind of server and component, belong to field of computer technology, and especially one kind is based on
The method of TensorFlow image recognition diagnosis hard disk failure.
Background technique
Server S erver gradually substitutes conventional small machine at present, runs and carry core application, therefore, for
The reliability and stability of server require just higher and higher and important.
Hard disk is that server is being serviced for storing the important component of Various types of data in order to characterize the working condition of hard disk
Device is equipped with many indicator lights, corresponds to the states such as hard disk startup, reading data or storage, operation and failure.
Currently, field maintenance person relies on experience to the monitoring of disk state and fault diagnosis more, in the insufficient feelings of experience
It under condition, inevitably goes wrong, also increases training and learning cost, operation and maintenance efficiency is lower.In addition, with information technology
Development and informationization degree raising, the computer rooms scale such as server is more and more huger, and related hard disk type is also more, to tie up
Shield personnel increase maintenance difficulties and complexity, and reality statistics is also found often due to monitoring and being safeguarded not to server apparatus
In week, the problem of leading to equipment wide-area failures, therefore, it is necessary to be realized in time and efficiently using effective scheme to hard disk failure monitoring
Identification, to avoid the safety of collapse and the raising data storage of server system.
As Chinese patent (Authorization Notice No. CN103940620B) discloses, " a kind of vehicle failure based on image recognition is sentenced
Not with the method and system of prompt ", the vehicle failure prompt generation that this method prompt vehicle OBD system using mobile intelligent terminal
Code image photographic extracts the failure symbol database pair in characteristics of image and remote server by image procossing, image recognition
Than, inquiry after, corresponding automotive handbook respective page and expert's treatment advice are shown in mobile intelligent terminal.
For another example Chinese patent (Authorization Notice No. CN104504713B) discloses a kind of " EMU operating status picture control
Failure automatic identifying method ", the EMU that current EMU operation troubles image detecting system TEDS equipment is acquired in the recent period
The monitoring for other TEDS equipment acquisition that history monitoring image passes through recently as the EMU on time history figure, same route
The EMU monitoring image that image is acquired as space history figure, current TEDS equipment as current figure, by time history figure,
Space history figure carries out image registration, acquisition time history registration figure and space history registration figure with current figure respectively, by the time
History registration figure is weighted and averaged as historical standard figure, and space history registration figure is weighted as fault right weight
Current figure is changed detection with historical standard figure and obtains feature difference matrix, utilizes feature difference matrix and failure by matrix
Fault flag matrix is calculated in weight matrix.The present invention have can effectively improve EMU operation troubles automatic identification rate,
The advantages of reducing False Rate.
But the above method and system mostly use the image of right place to be identified and handled, and can not carry out integrated condition
Judgement is also easy to produce error, and treatment effeciency reduces.
Summary of the invention
The present invention provides a kind of method based on TensorFlow image recognition diagnosis hard disk failure, for solving existing skill
The problems in art.
The present invention is achieved by the following technical programs：
A method of hard disk failure is diagnosed based on TensorFlow image recognition, is included the following steps：
S1. terminal carries out the shooting of multi-angle to server hard disc to obtain the multi-angle image of presence states；
S2. then, multi-angle image is transmitted to server end by terminal；
S3. after server receives multi-angle image, relative angle normalized is carried out, recycles geometric correction method
Multi-angle image is registrated, revised hard disk images are obtained, relative angle normalizes formula：
yk=akxk+bk (1)
In formula, ak、bkBe angle be K when normalized parameter, xk、ykIt respectively corrects and normalizes front and back under image k angle
Pixel gray value,Respectively pixel average value of the correction image and reference picture under K angle,
The respectively covariance of correction image and covariance and reference picture itself under reference picture k angle；
S4. server handles revised hard disk images information by TensorFlow and obtains hard disk presence states image
Data；
S5. server is stored with hard disk initialization normal condition image data；
S6. judgement is compared with initialization normal condition image data in hard disk presence states image data by server,
To obtain the presence states information of hard disk；
S7. the presence states information of hard disk is fed back into terminal.
A kind of method based on TensorFlow image recognition diagnosis hard disk failure as described above, will take according to equal proportion
The indicator light data of business device are initialised to the end Server by way of mirror image.
A kind of method based on TensorFlow image recognition diagnosis hard disk failure as described above, feeds back to the hard of terminal
The presence states information of disk, including hard disk failure point, failure remarks and failure cause data.
A kind of method based on TensorFlow image recognition diagnosis hard disk failure as described above, the step S3's is several
What bearing calibration registration is to need using on the basis of the high-precision hard disk images for having been subjected to ortho-rectification processing to multi-angle image
The image of correction carries out geometrical registration, and every scape image chooses equally distributed hard disk images characteristic point 20 or more, characteristic point point
Cloth must be uniform and guarantees that hard disk display lamp zone boundary is rectangular, carries out geometric correction, bilinear interpolation using quadratic polynomial
Method carries out the resampling of brightness value, makes correction accuracy control within 0.3 pixel.
Compared with prior art, it is an advantage of the invention that：
1, the present invention carries out fault message diagnosis using hard disk presence states multi-angle image, can preferably improve image letter
The correspondence precision of breath and disk state information reduces server pipe to advantageously improve the efficiency and accuracy of fault diagnosis
Reason and maintenance at.
2, the present invention is based on TensorFlow to improve the treatment effeciency to hard disk presence states image, and is realizing intelligence
It is accurate while changing processing to establish fault diagnosis corresponding informance, under the increasing occasion of existing computer room scale, energy
The learning cost of maintenance personnel is greatly reduced, is conducive to protect the equipment safeties such as hard disk devices, server, computer room.
3, the present invention utilize multi-angle hard disk presence states image, in conjunction with geometric correction processing be registrated, be fused to compared with
Accurate hard disk images can be avoided effectively because image angle is single and environment electromagnetics interfere the error generated to image, knot
Closing TensorFlow Intelligent treatment makes hard disk failure diagnosis process quickly, accurately, realizes full-automatic operation.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described.
Fig. 1 is flow chart of the invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.
As shown in Figure 1, a kind of method based on TensorFlow image recognition diagnosis hard disk failure of the present embodiment, including with
Lower step：
S1. terminal carries out the shooting of multi-angle to server hard disc to obtain the multi-angle image of presence states；
S2. then, multi-angle image is transmitted to server end by terminal；
S3. after server receives multi-angle image, relative angle normalized is carried out, recycles geometric correction method
Multi-angle image is registrated, revised hard disk images are obtained, relative angle normalizes formula：
yk=akxk+bk (1)
In formula, ak、bkBe angle be K when normalized parameter, xk、ykIt respectively corrects and normalizes front and back under image k angle
Pixel gray value,Respectively pixel average value of the correction image and reference picture under K angle,
The respectively covariance of correction image and covariance and reference picture itself under reference picture k angle；
S4. server handles revised hard disk images information by TensorFlow and obtains hard disk presence states image
Data；
S5. server is stored with hard disk initialization normal condition image data；
S6. judgement is compared with initialization normal condition image data in hard disk presence states image data by server,
To obtain the presence states information of hard disk；
S7. the presence states information of hard disk is fed back into terminal.
As shown in Figure 1, terminal can be using the hand-hold communication appliance wirelessly or non-wirelessly communicated, such as mobile phone, tablet computer
Or the computer etc. of camera is installed, APP program can also be installed on hand-hold communication appliance, to improve hard disk images transmission
With the efficiency of processing, it is easy to use, the concrete processing procedure of the present embodiment is as follows：
In step S101, taken pictures the server hard disc display lamp photo of a variety of different angles by terminal APP；
In step S102, the hard disk presence states image taken is submitted to the end Server by terminal APP；
In step S103, after server receives multi-angle image, relative angle normalized is carried out, recycles geometry
Bearing calibration is registrated multi-angle image, obtains revised hard disk images, and relative angle normalizes formula：
yk=akxk+bk (1)
In formula, ak、bkBe angle be K when normalized parameter, xk、ykIt respectively corrects and normalizes front and back under image k angle
Pixel gray value,Respectively pixel average value of the correction image and reference picture under K angle,
The respectively covariance of correction image and covariance and reference picture itself under reference picture k angle；
In step S104, Server is obtained treated hard disk images are sent to TensorFlow at end, and TensorFlow will
Revised hard disk images information processing at unified standard data；
In step S105, the indicator light data of server are initialised to Server by way of mirror image according to equal proportion
End；
In step S106, Server handles hard disk images data according to TensorFlow and normal condition image data carries out
It compares, judges the state of hard disk.
In step S107, processing result is fed back to terminal by Server, and by hard disk failure information in multi-angle hard disk figure
As upper carry out remarks, be described in detail failure there may be the reason of, facilitate operation maintenance personnel to analyze.
Further, the indicator light data of server are initialised to by the present embodiment according to equal proportion by way of mirror image
The end Server.Feed back to the presence states information of the hard disk of terminal, including hard disk failure point, failure remarks and failure cause number
According to.It is base that the geometric correction method registration of the step S3, which is using the high-precision hard disk images for having been subjected to ortho-rectification processing,
Standard carries out geometrical registration to the image that multi-angle image need to correct, and every scape image chooses equally distributed hard disk images characteristic point
20 or more, characteristic point distribution must be uniform and guarantees that hard disk display lamp zone boundary is rectangular, is carried out using quadratic polynomial several
What is corrected, and bilinear interpolation method carries out the resampling of brightness value, makes correction accuracy control within 0.3 pixel.
The present invention carries out fault message diagnosis using hard disk presence states multi-angle image, can preferably improve image information
Server admin is reduced to advantageously improve the efficiency and accuracy of fault diagnosis with the correspondence precision of disk state information
With maintenance at.The treatment effeciency to hard disk presence states image is improved based on TensorFlow, and is being realized at intelligence
It is accurate while reason to establish fault diagnosis corresponding informance, it, can be substantially under the increasing occasion of existing computer room scale
The learning cost for reducing maintenance personnel is conducive to protect the equipment safeties such as hard disk devices, server, computer room.
The technology contents of the not detailed description of the present invention are well-known technique.
Claims (4)
1. a kind of method based on TensorFlow image recognition diagnosis hard disk failure, which is characterized in that include the following steps：
S1. terminal carries out the shooting of multi-angle to server hard disc to obtain the multi-angle image of presence states；
S2. then, multi-angle image is transmitted to server end by terminal；
S3. after server receives multi-angle image, relative angle normalized is carried out, recycles geometric correction method to more
Angular image is registrated, and revised hard disk images are obtained, and relative angle normalizes formula：
yk=akxk+bk (1)
In formula, ak、bkBe angle be K when normalized parameter, xk、ykRespectively correct the picture of normalization front and back under image k angle
First gray value,Respectively pixel average value of the correction image and reference picture under K angle,Respectively
For the covariance of covariance and reference picture itself under correction image and reference picture k angle；
S4. server handles revised hard disk images information by TensorFlow and obtains hard disk presence states picture number
According to；
S5. server is stored with hard disk initialization normal condition image data；
S6. judgement is compared with initialization normal condition image data in hard disk presence states image data by server, to obtain
Take the presence states information of hard disk；
S7. the presence states information of hard disk is fed back into terminal.
2. a kind of method based on TensorFlow image recognition diagnosis hard disk failure according to claim 1, feature
It is, the indicator light data of server is initialised to the end Server by way of mirror image according to equal proportion.
3. a kind of method based on TensorFlow image recognition diagnosis hard disk failure according to claim 1, feature
It is, feeds back to the presence states information of the hard disk of terminal, including hard disk failure point, failure remarks and failure cause data.
4. a kind of method based on TensorFlow image recognition diagnosis hard disk failure according to claim 1, feature
It is, the geometric correction method registration of the step S3 is to be using the high-precision hard disk images for having been subjected to ortho-rectification processing
Benchmark carries out geometrical registration to the image that multi-angle image need to correct, and every scape image chooses equally distributed hard disk images feature
Point 20 or more, characteristic point distribution must be uniform and guarantee that hard disk display lamp zone boundary is rectangular, carried out using quadratic polynomial
Geometric correction, bilinear interpolation method carry out the resampling of brightness value, make correction accuracy control within 0.3 pixel.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810542914.9A CN108846833A (en) | 2018-05-30 | 2018-05-30 | A method of hard disk failure is diagnosed based on TensorFlow image recognition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810542914.9A CN108846833A (en) | 2018-05-30 | 2018-05-30 | A method of hard disk failure is diagnosed based on TensorFlow image recognition |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108846833A true CN108846833A (en) | 2018-11-20 |
Family
ID=64209933
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810542914.9A Pending CN108846833A (en) | 2018-05-30 | 2018-05-30 | A method of hard disk failure is diagnosed based on TensorFlow image recognition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108846833A (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101398886A (en) * | 2008-03-17 | 2009-04-01 | 杭州大清智能技术开发有限公司 | Rapid three-dimensional face identification method based on bi-eye passiveness stereo vision |
| US20120120069A1 (en) * | 2009-05-18 | 2012-05-17 | Kodaira Associates Inc. | Image information output method |
| CN105095031A (en) * | 2015-06-30 | 2015-11-25 | 北京奇虎科技有限公司 | Method and device for hard disk diagnosis |
| CN105260279A (en) * | 2015-11-04 | 2016-01-20 | 四川效率源信息安全技术股份有限公司 | Method and device of dynamically diagnosing hard disk failure based on S.M.A.R.T (Self-Monitoring Analysis and Reporting Technology) data |
| US20160093214A1 (en) * | 2014-09-30 | 2016-03-31 | Xerox Corporation | Vision-based on-street parked vehicle detection via normalized-view classifiers and temporal filtering |
| CN106097714A (en) * | 2016-07-27 | 2016-11-09 | 北京信路威科技股份有限公司 | A kind of through street event detection device based on unmanned plane and detection method |
| CN106093050A (en) * | 2016-05-27 | 2016-11-09 | 国网天津市电力公司 | A kind of for intelligent electric energy meter PCB encapsulation, the automatic checkout system of components and parts |
| CN107220937A (en) * | 2017-05-25 | 2017-09-29 | 云南电网有限责任公司电力科学研究院 | A kind of electrical equipment detection infrared panorama image processing method and platform |
| CN107784635A (en) * | 2017-09-28 | 2018-03-09 | 北京中电汇通科技有限公司 | A kind of server panel automatic identifying method and system |
-
2018
- 2018-05-30 CN CN201810542914.9A patent/CN108846833A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101398886A (en) * | 2008-03-17 | 2009-04-01 | 杭州大清智能技术开发有限公司 | Rapid three-dimensional face identification method based on bi-eye passiveness stereo vision |
| US20120120069A1 (en) * | 2009-05-18 | 2012-05-17 | Kodaira Associates Inc. | Image information output method |
| US20160093214A1 (en) * | 2014-09-30 | 2016-03-31 | Xerox Corporation | Vision-based on-street parked vehicle detection via normalized-view classifiers and temporal filtering |
| CN105095031A (en) * | 2015-06-30 | 2015-11-25 | 北京奇虎科技有限公司 | Method and device for hard disk diagnosis |
| CN105260279A (en) * | 2015-11-04 | 2016-01-20 | 四川效率源信息安全技术股份有限公司 | Method and device of dynamically diagnosing hard disk failure based on S.M.A.R.T (Self-Monitoring Analysis and Reporting Technology) data |
| CN106093050A (en) * | 2016-05-27 | 2016-11-09 | 国网天津市电力公司 | A kind of for intelligent electric energy meter PCB encapsulation, the automatic checkout system of components and parts |
| CN106097714A (en) * | 2016-07-27 | 2016-11-09 | 北京信路威科技股份有限公司 | A kind of through street event detection device based on unmanned plane and detection method |
| CN107220937A (en) * | 2017-05-25 | 2017-09-29 | 云南电网有限责任公司电力科学研究院 | A kind of electrical equipment detection infrared panorama image processing method and platform |
| CN107784635A (en) * | 2017-09-28 | 2018-03-09 | 北京中电汇通科技有限公司 | A kind of server panel automatic identifying method and system |
Non-Patent Citations (2)
| Title |
|---|
| 胡亚斌 等，: "基于多期遥感影像的砂质岸线提取方法——以海阳沙滩为例", 《海洋开发与管理》 * |
| 范文义 等，: "多时相遥感影像相对辐射校正方法对比", 《遥感信息》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105765963B (en) | Apparatus for camera calibration | |
| CN110834327A (en) | Robot control method and device | |
| CN106895791B (en) | Board deformation monitoring and early warning system | |
| CN112163566A (en) | Video image monitoring method and device | |
| JP2011002878A (en) | Meter reading system and meter reading method | |
| CN115018854B (en) | Major hazard source monitoring and early warning system and method thereof | |
| CN104915633A (en) | Image processing device, image processing method, and image processing program | |
| CN111803022A (en) | Vision detection method, detection device, terminal equipment and readable storage medium | |
| CN112640415B (en) | Meter reading device, meter reading system and storage medium | |
| CN116337412A (en) | Screen detection method, device and storage medium | |
| CN110232676B (en) | Method, device, equipment and system for detecting installation state of aircraft cable bracket | |
| CN108846833A (en) | A method of hard disk failure is diagnosed based on TensorFlow image recognition | |
| CN104238980A (en) | Information display apparatus and information display method | |
| US11526976B2 (en) | Using augmented reality to assist in device installation | |
| CN112529548A (en) | Online education platform cloud resource management system | |
| CN111803023A (en) | Vision value correction method, correction device, terminal equipment and storage medium | |
| CN106710024A (en) | Method and device for unattended automatic inspection and alarm | |
| CN114708821B (en) | Intelligent LED display screen system based on multi-sensor data fusion | |
| US20210065468A1 (en) | Systems and methods for artificial intelligence in a vehicle service environment | |
| CN113759765B (en) | FFU intelligent group control system and maintenance method thereof | |
| CN110084076A (en) | Evidence display methods and device | |
| EP4187454A1 (en) | Failure identification and handling method, and system | |
| CN111723754A (en) | Left and right eye identification method, identification device, terminal equipment and storage medium | |
| CN109284691A (en) | A kind of face identification method and system based on user terminal | |
| CN110164118A (en) | Monitoring method, terminal, system and the server of metering 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: 20181120 |
|
| RJ01 | Rejection of invention patent application after publication |