CN105740822A - Mechanical fault diagnosis optimization method and system - Google Patents
Mechanical fault diagnosis optimization method and system Download PDFInfo
- Publication number
- CN105740822A CN105740822A CN201610069441.6A CN201610069441A CN105740822A CN 105740822 A CN105740822 A CN 105740822A CN 201610069441 A CN201610069441 A CN 201610069441A CN 105740822 A CN105740822 A CN 105740822A
- Authority
- CN
- China
- Prior art keywords
- diagnosis
- sample
- fault
- diagnostic
- diagnostic sample
- 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
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Abstract
The invention discloses a mechanical fault diagnosis optimization method and system. The method comprises a preparation step and a diagnosis step. In the preparation step, a diagnosis sample library comprising all known diagnosis samples at present is constructed, a plurality of unknown diagnosis samples are constructed, and according to the known diagnosis samples and the unknown diagnosis samples, a fault diagnosis algorithm is trained and generated; and in the diagnosis step, a diagnosis sample representing a current running state of machinery is obtained from a site by applying the trained fault diagnosis algorithm, if a diagnosis result is that the running state is unknown, the running state of the machinery is determined by utilizing a technology of expert analysis, field analysis or the like, the diagnosis sample is updated as a known diagnosis sample, and after an unknown diagnosis sample similar to the recently updated known diagnosis sample is deleted, the fault diagnosis algorithm is re-trained. According to the method and system, the diagnosis sample library and the fault diagnosis algorithm can be dynamically perfected, so that the accuracy of fault diagnosis is enhanced, the running quality of machinery equipment is improved, and the production efficiency is raised.
Description
Technical field
The present invention relates to mechanical fault detection field, particularly to a kind of mechanical fault diagnosis optimization method and system.
Background technology
Along with developing rapidly of modern industry and science and technology, the structure of modern comfort becomes increasingly complex, function is more and more perfect, automaticity is more and more higher, not only interrelated between the different piece of same equipment, closely bring disaster upon conjunction, and between distinct device, there is also relation closely, form an entirety in process of production.Therefore, place's fault is possible to cause chain reaction, causes that the even whole production process of whole equipment is not normally functioning, so mechanical fault diagnosis has great importance.
Fault diagnosis is inspired from medical test and diagnosis, and the emerging comprehensive cross discipline got up along with the development of the modern science and technology such as system engineering, theory of information, cybernetics, electronic technology, computer technology, information processing, artificial intelligence, its research contents relate to failure mechanism, Sensors & Testing Technology, signal analysis and data process, automatically control, the technical field such as system identification, specialist system and computer software and hardware.
At present due to reason that restriction and the fault sample of technology lack relatively, the problem that in actual production process, the fault diagnosis technology of application always exists wrong diagnosis, this is unacceptable in indivedual fields, set up a perfect fault sample storehouse for this, design the problem that the high fault diagnosis algorithm of accuracy rate of diagnosis just becomes in the urgent need to address.
Summary of the invention
It is an object of the invention to: low for current mechanical breakdown recognition accuracy, the incomplete technical problem of fault sample, it is provided that a kind of mechanical fault diagnosis optimization method and system.
In order to realize foregoing invention purpose, the technical solution used in the present invention is: a kind of mechanical fault diagnosis optimization method, including preparation process and diagnosis algorithm, described preparation process includes: first build the diagnostic sample storehouse comprising all known diagnosis samples at present, then according to described known diagnosis sample, build some unknown diagnostic sample, and described unknown diagnostic sample is saved in described diagnostic sample storehouse；Wherein, the running status of machinery is divided into normal condition, incipient fault state and malfunction, described known diagnosis sample is respectively used to characterize the known normal condition of machinery, known incipient fault state and known fault conditions, and described unknown diagnostic sample is for characterizing the unknown running status of machinery；
Finally, according to described known diagnosis sample all of in described diagnostic sample storehouse and described unknown diagnostic sample, train and generate fault diagnosis algorithm, and the diagnostic result of described fault diagnosis algorithm is corresponding with described known normal condition, described known incipient fault state, described known fault conditions and described unknown running status respectively；
Described diagnosis algorithm includes:
The first step: the diagnostic sample characterizing machinery current operating conditions using the diagnosis of described fault diagnosis algorithm to obtain from scene, finally exports diagnostic result；
Second step: if the described diagnostic result in the first step is described unknown running status, it is determined that the Mechanical Running Condition that described diagnostic sample characterizes, and the described diagnostic sample having determined that the Mechanical Running Condition characterized is updated to described known diagnosis sample；
3rd step: after deleting the described unknown diagnostic sample similar to the described known diagnosis sample of recent renewal, according to described known diagnosis sample all of in described diagnostic sample storehouse and described unknown diagnostic sample, re-training also generates fault diagnosis algorithm；
4th step: current described fault diagnosis algorithm is replaced with the described fault diagnosis algorithm that in the 3rd step, re-training generates.
According to a kind of specific embodiment, in the described first step, to the described known diagnosis sample characterizing described known normal condition, carry out variation process, to build described unknown diagnostic sample；Wherein,
Described variation processes: will characterize the described known diagnosis sample of described known fault conditions, and the wavelet packet random superposition obtained after wavelet analysis is to the described known diagnosis sample characterizing described known normal condition；Or characterizing the described known diagnosis sample of described known fault conditions, the intrinsic mode function obtained after EMD decomposes is added on the described known diagnosis sample characterizing described known normal condition.
According to a kind of specific embodiment, in described second step, by fault diagnosis expert system or expert, described diagnostic sample is analyzed, or by field service personnel, the plant equipment diagnosed as unknown running status is overhauled, to determine the Mechanical Running Condition that described diagnostic sample characterizes, and the described diagnostic sample having determined that the Mechanical Running Condition characterized is updated to described known diagnosis sample.
According to a kind of specific embodiment, in described 3rd step, by extracting the fault signature of the described known diagnosis sample of recent renewal and the described unknown diagnostic sample in described diagnostic sample storehouse, and the frequency band energy ratio of relatively more described unknown diagnostic sample and the described fault signature of the described known diagnosis sample of recent renewal, delete the frequency band energy of described fault signature than the described unknown diagnostic sample beyond set point.
According to a kind of specific embodiment, training generates described fault diagnosis algorithm and includes:
Extract described known diagnosis sample and the fault signature of described unknown diagnostic sample, and the input with described fault signature for described fault diagnosis algorithm, train described known normal condition or described known incipient fault state or known fault conditions that the diagnostic result of described fault diagnosis algorithm characterizes with described known diagnosis sample respectively, or the unknown running status that described unknown diagnostic sample characterizes is corresponding；
After training generates described fault diagnosis algorithm, by verifying that the accuracy rate of diagnosis of the described fault diagnosis algorithm generated is trained in Sample Storehouse checking, if accuracy rate of diagnosis is not less than design load, then train successfully, otherwise adjust described fault diagnosis algorithm, and be again trained, until training successfully.
According to a kind of specific embodiment, in described diagnosis algorithm, use the diagnostic sample characterizing machinery current operating conditions that the diagnosis of described fault diagnosis algorithm obtains from scene, after the described diagnostic result of final output, if the mechanical current operating conditions that described diagnostic result and described diagnostic sample characterize does not correspond, then redefine the Mechanical Running Condition that described diagnostic sample characterizes, and the described diagnostic sample having determined that the Mechanical Running Condition characterized is updated to described known diagnosis sample.
According to a kind of specific embodiment, including in-situ processing unit and server, wherein,
Described in-situ processing unit, for obtaining the diagnostic sample characterizing machinery current operating conditions from scene and obtaining fault diagnosis algorithm from described server, and use described fault diagnosis algorithm to diagnose described diagnostic sample, if the diagnostic result of described diagnostic sample is unknown running status, then the described diagnostic sample of this fault diagnosis is sent to described server；
Described server, for in preparation process, build the diagnostic sample storehouse comprising all known diagnosis samples at present and build some unknown diagnostic sample, and described known diagnosis sample is respectively used to characterize the known normal condition of machinery, known incipient fault state and known fault conditions, and described unknown diagnostic sample is for characterizing the unknown running status of machinery；With in diagnosis algorithm, when the described diagnostic result that described in-situ processing unit obtains after described diagnostic sample is diagnosed is described unknown running status, determine that the running status of the machinery that described diagnostic sample characterizes is normal condition or incipient fault state or malfunction, and after described diagnostic sample is updated to known diagnosis sample, delete the described unknown diagnostic sample similar to the described known diagnosis sample of recent renewal；And respectively in described preparation process and described diagnosis algorithm, according to described known diagnosis sample all of in described diagnostic sample storehouse and described unknown diagnostic sample, train and generate fault diagnosis algorithm, and the diagnostic result of described fault diagnosis algorithm is corresponding with described known normal condition, described known incipient fault state, described known fault conditions and described unknown running status respectively.
According to a kind of specific embodiment, described in-situ processing unit includes,
Gather subelement, for obtaining the described diagnostic sample characterizing machinery current operating conditions from scene；
Storing sub-units, for preserving the described diagnostic sample for this diagnosis, and preserves the described fault diagnosis algorithm obtained from described server；
Operator unit, for calling described fault diagnosis algorithm in described storing sub-units and the described diagnostic sample for this diagnosis, and uses described fault diagnosis algorithm that described diagnostic sample is diagnosed, and final exports described diagnostic result；
Communicator unit, for obtaining fault diagnosis algorithm from described server, and when the described diagnostic result of this diagnosis is unknown running status, for sending the described diagnostic sample of this diagnosis to described server.
According to a kind of specific embodiment, described server includes,
NE, for setting up and the communication connection of described communicator unit；
Database Unit, for building the diagnostic sample storehouse comprising all known diagnosis samples at present, and storage known diagnosis sample, unknown diagnostic sample and described fault diagnosis algorithm；
Running status confirmation unit, for determining that the mechanical running status that described diagnostic sample characterizes is normal condition or incipient fault state or malfunction, and is updated to known diagnosis sample by described diagnostic sample；
Unknown diagnostic sample construction unit, for the described known diagnosis sample characterizing described known normal condition, carrying out variation process, to build described unknown diagnostic sample；
Unknown diagnostic sample deletes unit, after extracting the described known diagnosis sample of recent renewal and the fault signature of the described unknown diagnostic sample in described diagnostic sample storehouse, compare the frequency band energy ratio of described unknown diagnostic sample and the described fault signature of the described known diagnosis sample of recent renewal, and the frequency band energy of the fault signature described in deletion is than the described unknown diagnostic sample beyond set point；
Fault diagnosis algorithm training unit, for according to described known diagnosis sample all of in described diagnostic sample storehouse and described unknown diagnostic sample, training and generate fault diagnosis algorithm, the diagnostic result of wherein said fault diagnosis algorithm is corresponding with described known normal condition, described known incipient fault state, described known fault conditions and described unknown running status respectively；
Fault diagnosis algorithm authentication unit, for checking Sample Storehouse is carried out fault diagnosis, the accuracy rate of diagnosis of the described fault diagnosis algorithm that checking training generates, if accuracy rate of diagnosis is not less than design load, then train successfully, and described fault diagnosis algorithm is stored in Database Unit, otherwise adjust described fault diagnosis algorithm, and be again trained, until train successfully；
Update information transmitting unit, for training after successfully at described fault diagnosis training unit, the information of described fault diagnosis algorithm will be updated, by described NE, send to described in-situ processing unit, so that current described fault diagnosis algorithm is replaced with the described fault diagnosis algorithm trained after successfully by described in-situ processing unit.
According to a kind of specific embodiment, described in-situ processing subelement also includes sample amendment subelement, for in described diagnosis algorithm, when the mechanical current operating conditions that described diagnostic sample characterizes and described diagnostic result do not correspond, and determine by field service personnel and revise that the Mechanical Running Condition that described diagnostic sample characterizes is after normal condition or incipient fault state or malfunction, described diagnostic sample is sent to described server by described communicator unit, and described diagnostic sample is updated to described known diagnosis sample.
Compared with prior art, beneficial effects of the present invention:
The present invention is capable of the dynamically perfect of diagnostic sample storehouse; diagnostic sample according to the diagnostic sample storehouse after improving simultaneously; improve fault diagnosis algorithm; and the fault diagnosis algorithm after improving is updated to in-situ processing unit; the running quality of elevating mechanism equipment; reduce downtime, improve production efficiency.The present invention often improves a diagnostic sample storehouse simultaneously, then similar to the diagnostic sample updated in Deletion Diagnostics Sample Storehouse unknown diagnostic sample, simplifies fault diagnosis algorithm, improves the accuracy rate of fault diagnosis.
Accompanying drawing explanation
Fig. 1 is the schematic flow sheet of the inventive method；
Fig. 2 is the flow chart of a kind of embodiment of the inventive method diagnosis algorithm；
Fig. 3 is the structural representation of present system；
Fig. 4 is the structure chart of a kind of embodiment of present system；
Fig. 5 is the structure chart of present system in-situ processing unit；
Fig. 6 is the structure chart of a kind of embodiment of present system in-situ processing unit；
Fig. 7 is the structure chart of present system server.
Detailed description of the invention
Below in conjunction with detailed description of the invention, the present invention is described in further detail.But this should not being interpreted as, the scope of the above-mentioned theme of the present invention is only limitted to below example, and all technology realized based on present invention belong to the scope of the present invention.
Schematic flow sheet in conjunction with the inventive method shown in Fig. 1；Wherein, the mechanical fault diagnosis optimization method of the present invention, including preparation process and diagnosis algorithm.
Wherein, preparation process includes: first, builds the diagnostic sample storehouse comprising all known diagnosis samples at present, then according to the known diagnosis sample in diagnostic sample storehouse, builds some unknown diagnostic sample, and the unknown diagnostic sample is saved in diagnostic sample storehouse；Wherein, the running status of machinery is divided into normal condition, incipient fault state and malfunction, known diagnosis sample is respectively used to characterize the known normal condition of machinery, known incipient fault state and known fault conditions, and unknown diagnostic sample is for characterizing the unknown running status of machinery.
Finally, according to known diagnosis sample all of in diagnostic sample storehouse and unknown diagnostic sample, train and generate fault diagnosis algorithm, and the diagnostic result of fault diagnosis algorithm is corresponding with known normal condition, known incipient fault state, known fault conditions and unknown running status respectively.
In an embodiment, the mode building unknown diagnostic sample in preparation process is: to the known diagnosis sample characterizing known normal condition, carry out variation process, to build unknown diagnostic sample.
Concrete, variation processes and is: will characterize the known diagnosis sample of known fault conditions, and the wavelet packet random superposition obtained after wavelet analysis is to the known diagnosis sample characterizing known normal condition；Or characterizing the known diagnosis sample of known fault conditions, the intrinsic mode function obtained after EMD decomposes is added on the known diagnosis sample characterizing known normal condition.
The diagnosis algorithm of the present invention includes:
The first step: the diagnostic sample characterizing machinery current operating conditions using fault diagnosis algorithm diagnosis to obtain from scene, finally exports diagnostic result.
In an embodiment, generally with certain frequency, the diagnostic sample characterizing machinery current operating conditions is obtained from scene, use fault diagnosis algorithm that diagnostic sample is diagnosed, if diagnostic result is known normal condition or known incipient fault state or known fault conditions, then it represents that the diagnostic sample of machinery current operating conditions is normal condition or incipient fault state or known fault conditions, simultaneously, wait when next time obtains diagnostic sample, then use fault diagnosis algorithm to diagnose.If diagnostic result is unknown running status, then enter second step.
Second step: if the described diagnostic result in the first step is described unknown running status, it is determined that the Mechanical Running Condition that described diagnostic sample characterizes, and the described diagnostic sample having determined that the Mechanical Running Condition characterized is updated to described known diagnosis sample.
In an embodiment, in second step, by fault diagnosis expert system or expert, described diagnostic sample is analyzed, or by field service personnel, the plant equipment diagnosed as unknown running status is overhauled, to determine that Mechanical Running Condition that diagnostic sample characterizes is for the one in normal condition or incipient fault state or malfunction, after determining, diagnostic sample is updated to known diagnosis sample, then the Mechanical Running Condition that this diagnostic sample characterizes is defined as the one in known normal condition or known incipient fault state or known fault conditions.Further, described diagnostic sample is updated to described known diagnosis sample after determining by the Mechanical Running Condition characterized when this diagnostic sample.
3rd step: after deleting the unknown diagnostic sample similar to the known diagnosis sample of recent renewal, according to known diagnosis sample all of in diagnostic sample storehouse and unknown diagnostic sample, re-training also generates fault diagnosis algorithm.
When implementing, the mode deleting the unknown diagnostic sample similar to the known diagnosis sample of recent renewal is: the fault signature of known diagnosis sample with the unknown diagnostic sample in diagnostic sample storehouse by extracting recent renewal, and the frequency band energy ratio of the unknown diagnostic sample of comparison and the fault signature of the known diagnosis sample of recent renewal, the frequency band energy of the fault signature of deletion is than the unknown diagnostic sample beyond set point.Due to diagnostic sample storehouse of every renolation, then similar to the diagnostic sample updated in Deletion Diagnostics Sample Storehouse unknown diagnostic sample, thus simplifying fault diagnosis algorithm, improve the accuracy rate of fault diagnosis.
When implementing, training generates fault diagnosis algorithm and includes: extract known diagnosis sample and the fault signature of unknown diagnostic sample, and with input that fault signature is fault diagnosis algorithm, known normal condition that the diagnostic result of training fault diagnosis algorithm characterizes with known diagnosis sample respectively or known incipient fault state or known fault conditions, or the unknown running status that unknown diagnostic sample characterizes is corresponding.
After training generates fault diagnosis algorithm, by verifying that the accuracy rate of diagnosis of the fault diagnosis algorithm generated is trained in Sample Storehouse checking, if accuracy rate of diagnosis is not less than design load, then train successfully, otherwise adjust fault diagnosis algorithm, and be again trained, until training successfully.
To adopt three layers BP neural network failure diagnosis algorithm that diagnostic sample storehouse is trained, according to known diagnosis sample all of in diagnostic sample storehouse and unknown diagnostic sample, generate three layers BP neural network failure diagnosis algorithm, it is verified again through checking Sample Storehouse, and constantly revise BP neutral net hidden layer node quantity, until diagnosis accuracy reaches designing requirement, so that it is guaranteed that the diagnostic accuracy of the BP neural network algorithm of training generation.Certain present invention can also adopt the intelligent algorithms such as the algorithm of support vector machine in the middle of prior art.
In the specific implementation, fault signature extracting method can select short time discrete Fourier transform, wavelet transformation, IHT method, EMD method to extract vibration signals spectrograph, and recycling merges the methods such as energy spectrum, Wavelet Packet Frequency Band Energy, intrinsic mode function energy and builds fault feature vector.And, training generates in the process of fault diagnosis algorithm, some diagnostic sample that the checking Sample Storehouse adopted can be chosen from diagnostic sample storehouse, the fault diagnosis algorithm that training generates is used to diagnose these diagnostic sample, and calculate accuracy rate of diagnosis, carry out the accuracy of the fault diagnosis of the fault diagnosis algorithm that objective evaluation training generates.
4th step: current fault diagnosis algorithm is replaced with the fault diagnosis algorithm that in the 3rd step, re-training generates.Thus improving fault diagnosis algorithm, the running quality of elevating mechanism equipment, reduces downtime, improves production efficiency.
In the present invention, the running status of machinery is divided into normal condition, incipient fault state and malfunction, refer to and mechanical all of normal conditions is summarized as normal condition, mechanical all of incipient fault situation is summarized as incipient fault state, mechanical all of failure situations is summarized as malfunction.Wherein, although incipient fault situation refers to that machinery deviation normal conditions remains to work under this situation, and there is the risk of the situation that breaks down.
Flow chart in conjunction with a kind of embodiment of the inventive method diagnosis algorithm shown in Fig. 2；Wherein, in diagnosis algorithm, use the diagnostic sample characterizing machinery current operating conditions that fault diagnosis algorithm diagnosis obtains from scene, after final output diagnostic result, if the mechanical current operating conditions that diagnostic result and diagnostic sample characterize does not correspond, then redefining the Mechanical Running Condition that diagnostic sample characterizes is normal condition or incipient fault state or malfunction, and after determining, diagnostic sample is updated to known diagnosis sample.Thus widening the source of the diagnostic sample of the present invention, and revising wrong diagnostic sample, improving the accuracy rate of diagnosis of fault diagnosis algorithm, improve the practicality of the present invention.
Based on the same inventive concept with the mechanical fault diagnosis optimization method of the present invention, the present invention also provides for a kind of Diagnosis system of mechanical failure.Structural representation in conjunction with the present system shown in Fig. 3；Wherein, the Diagnosis system of mechanical failure of the present invention includes in-situ processing unit and server.
Wherein, in-situ processing unit for obtaining the diagnostic sample characterizing machinery current operating conditions and obtaining fault diagnosis algorithm from server from scene, and use fault diagnosis algorithm diagnosis diagnostic sample, if the diagnostic result of diagnostic sample is unknown running status, then the diagnostic sample of this fault diagnosis is sent to server.
Server is in preparation process, build the diagnostic sample storehouse comprising all known diagnosis samples at present and build some unknown diagnostic sample, and known diagnosis sample is respectively used to characterize the known normal condition of machinery, known incipient fault state and known fault conditions, and unknown diagnostic sample is for characterizing the unknown running status of machinery.
Server is additionally operable in diagnosis algorithm, when the diagnostic result that in-situ processing unit obtains after diagnostic sample is diagnosed is for unknown running status, determine that the running status of the machinery that diagnostic sample characterizes is normal condition or incipient fault state or malfunction, and after diagnostic sample is updated to known diagnosis sample, delete the unknown diagnostic sample similar to the known diagnosis sample of recent renewal.
And server is respectively in preparation process and diagnosis algorithm, for according to known diagnosis sample all of in diagnostic sample storehouse and unknown diagnostic sample, train and generate fault diagnosis algorithm, and the diagnostic result of fault diagnosis algorithm is corresponding with known normal condition, known incipient fault state, known fault conditions and unknown running status respectively.
Structure chart in conjunction with a kind of embodiment of the present system shown in Fig. 4；Wherein, being made up of a server cluster multiple servers, and plant equipment and in-situ processing unit are numbered, wherein, all plant equipment are all certain particular elements of the equipment of same model or same equipment.Owing to incorporating substantial amounts of plant equipment in system, it occurs the probability of unknown operation conditions to increase, thus accelerating the perfect of diagnostic sample storehouse.
Structure chart in conjunction with the present system in-situ processing unit shown in Fig. 5；Wherein, in-situ processing unit includes gathering subelement, storing sub-units, operator unit and communicator unit.
Wherein, gather subelement for obtaining the diagnostic sample characterizing machinery current operating conditions from scene, and the diagnostic sample of acquisition is saved in storing sub-units.Storing sub-units is for preserving the diagnostic sample for this diagnosis, and preserves the fault diagnosis algorithm obtained from server.Operator unit is for calling fault diagnosis algorithm in storing sub-units and the diagnostic sample for this diagnosis, and uses fault diagnosis algorithm that diagnostic sample is diagnosed, and finally exports diagnostic result.Communicator unit is for obtaining fault diagnosis algorithm from server, and when the diagnostic result of this diagnosis is unknown running status, is used for sending to server the diagnostic sample of this diagnosis.
Structure chart in conjunction with a kind of embodiment of the present system in-situ processing unit shown in Fig. 6；Wherein, in-situ processing subelement also includes sample amendment subelement, for in diagnosis algorithm, when the mechanical current operating conditions that diagnostic sample characterizes and diagnostic result do not correspond, determined and revise the Mechanical Running Condition that diagnostic sample characterizes by field service personnel after, diagnostic sample is sent to server by communicator unit, and diagnostic sample is updated to known diagnosis sample.
Structure chart in conjunction with the server of the present system shown in Fig. 7；Wherein, server includes NE, Database Unit, running status confirmation unit, unknown diagnostic sample construction unit, fault diagnosis algorithm training unit, fault diagnosis algorithm authentication unit, unknown diagnostic sample deletion unit and updates information transmitting unit.
Wherein, NE is used for setting up and the communication connection of communicator unit, and, NE is capable of and the connection of multiple communicator unit simultaneously.Database Unit is for building the diagnostic sample storehouse comprising all known diagnosis samples at present, and storage known diagnosis sample, unknown diagnostic sample and fault diagnosis algorithm.The running status of the machinery that running status confirmation unit characterizes for determining described diagnostic sample, and described diagnostic sample is updated to known diagnosis sample.Unknown diagnostic sample construction unit is for the known diagnosis sample characterizing known normal condition, carrying out variation process, to build unknown diagnostic sample.
Unknown diagnostic sample deletes unit, after extracting the known diagnosis sample of recent renewal and the fault signature of the unknown diagnostic sample in diagnostic sample storehouse, compare the frequency band energy ratio of unknown diagnostic sample and the fault signature of the known diagnosis sample of recent renewal, and the frequency band energy of the fault signature deleted is than the unknown diagnostic sample beyond set point.
Fault diagnosis algorithm training unit, for according to known diagnosis sample all of in diagnostic sample storehouse and unknown diagnostic sample, training and generate fault diagnosis algorithm, wherein the diagnostic result of fault diagnosis algorithm is corresponding with known normal condition, known incipient fault state, known fault conditions and unknown running status respectively.
Fault diagnosis algorithm authentication unit, for verifying the accuracy rate of diagnosis of the fault diagnosis algorithm of Sample Storehouse checking training generation, if accuracy rate of diagnosis is not less than design load, then train successfully, and fault diagnosis algorithm is stored in Database Unit, otherwise adjust fault diagnosis algorithm, and be again trained, until training successfully.
Update information transmitting unit, for training after successfully at fault diagnosis training unit, by updating the information of fault diagnosis algorithm, pass through NE, send to in-situ processing unit, so that current fault diagnosis algorithm is replaced with by in-situ processing unit trains the fault diagnosis algorithm after successfully.
The present invention is capable of the dynamically perfect of diagnostic sample storehouse; diagnostic sample according to the diagnostic sample storehouse after improving simultaneously; improve fault diagnosis algorithm; and the fault diagnosis algorithm after improving is updated to in-situ processing unit; improve the accuracy rate of fault diagnosis; thus the running quality of elevating mechanism equipment, reduce downtime, improve production efficiency.
Above in conjunction with accompanying drawing, the specific embodiment of the present invention is described in detail, but the present invention is not restricted to above-mentioned embodiment, without departing from the spirit and scope situation of claims hereof, those skilled in the art may be made that various amendment or remodeling.
Claims (10)
1. a mechanical fault diagnosis optimization method, it is characterised in that include preparation process and diagnosis algorithm, wherein,
Described preparation process includes: first build the diagnostic sample storehouse comprising all known diagnosis samples at present, then according to described known diagnosis sample, builds some unknown diagnostic sample, and described unknown diagnostic sample is saved in described diagnostic sample storehouse；Wherein, the running status of machinery is divided into normal condition, incipient fault state and malfunction, described known diagnosis sample is respectively used to characterize the known normal condition of machinery, known incipient fault state and known fault conditions, and described unknown diagnostic sample is for characterizing the unknown running status of machinery；
Finally, according to described known diagnosis sample all of in described diagnostic sample storehouse and described unknown diagnostic sample, train and generate fault diagnosis algorithm, and the diagnostic result of described fault diagnosis algorithm is corresponding with described known normal condition, described known incipient fault state, described known fault conditions and described unknown running status respectively；
Described diagnosis algorithm includes:
The first step: the diagnostic sample characterizing machinery current operating conditions using the diagnosis of described fault diagnosis algorithm to obtain from scene, finally exports diagnostic result；
Second step: if the described diagnostic result in the first step is described unknown running status, it is determined that the Mechanical Running Condition that described diagnostic sample characterizes, and the described diagnostic sample having determined that the Mechanical Running Condition characterized is updated to described known diagnosis sample；
3rd step: after deleting the described unknown diagnostic sample similar to the described known diagnosis sample of recent renewal, according to described known diagnosis sample all of in described diagnostic sample storehouse and described unknown diagnostic sample, re-training also generates fault diagnosis algorithm；
4th step: current described fault diagnosis algorithm is replaced with the described fault diagnosis algorithm that in the 3rd step, re-training generates.
2. mechanical fault diagnosis optimization method as claimed in claim 1, it is characterised in that in described preparation process, to the described known diagnosis sample characterizing described known normal condition, carry out variation process, to build described unknown diagnostic sample；Wherein,
Described variation processes: will characterize the described known diagnosis sample of described known fault conditions, and the wavelet packet random superposition obtained after wavelet analysis is to the described known diagnosis sample characterizing described known normal condition；Or characterizing the described known diagnosis sample of described known fault conditions, the intrinsic mode function obtained after EMD decomposes is added on the described known diagnosis sample characterizing described known normal condition.
3. mechanical fault diagnosis optimization method as claimed in claim 2, it is characterized in that, in described second step, by fault diagnosis expert system or expert, described diagnostic sample is analyzed, or by field service personnel, the plant equipment diagnosed as unknown running status is overhauled, to determine the Mechanical Running Condition that described diagnostic sample characterizes, and the described diagnostic sample having determined that the Mechanical Running Condition characterized is updated to described known diagnosis sample.
4. mechanical fault diagnosis algorithm as claimed in claim 3, it is characterized in that, in described 3rd step, by extracting the fault signature of the described known diagnosis sample of recent renewal and the described unknown diagnostic sample in described diagnostic sample storehouse, and the frequency band energy ratio of relatively more described unknown diagnostic sample and the described fault signature of the described known diagnosis sample of recent renewal, delete the frequency band energy of described fault signature than the described unknown diagnostic sample beyond set point.
5. mechanical fault diagnosis optimization method as claimed in claim 1, it is characterised in that training generates described fault diagnosis algorithm and includes:
Extract described known diagnosis sample and the fault signature of described unknown diagnostic sample, and the input with described fault signature for described fault diagnosis algorithm, train described known normal condition or described known incipient fault state or known fault conditions that the diagnostic result of described fault diagnosis algorithm characterizes with described known diagnosis sample respectively, or the unknown running status that described unknown diagnostic sample characterizes is corresponding；
After training generates described fault diagnosis algorithm, by verifying that the accuracy rate of diagnosis of the described fault diagnosis algorithm generated is trained in Sample Storehouse checking, if accuracy rate of diagnosis is not less than design load, then train successfully, otherwise adjust described fault diagnosis algorithm, and be again trained, until training successfully.
6. mechanical fault diagnosis optimization method as claimed in claim 1, it is characterized in that, in described diagnosis algorithm, use the diagnostic sample characterizing machinery current operating conditions that the diagnosis of described fault diagnosis algorithm obtains from scene, after the described diagnostic result of final output, if the mechanical current operating conditions that described diagnostic result and described diagnostic sample characterize does not correspond, then redefine the Mechanical Running Condition that described diagnostic sample characterizes, and the described diagnostic sample having determined that the Mechanical Running Condition characterized is updated to described known diagnosis sample.
7. the Diagnosis system of mechanical failure utilizing mechanical fault diagnosis optimization method as claimed in claim 1, it is characterised in that include in-situ processing unit and server, wherein,
Described in-situ processing unit, for obtaining the diagnostic sample characterizing machinery current operating conditions from scene and obtaining fault diagnosis algorithm from described server, and use described fault diagnosis algorithm to diagnose described diagnostic sample, if the diagnostic result of described diagnostic sample is unknown running status, then the described diagnostic sample of this fault diagnosis is sent to described server；
Described server, for in preparation process, build the diagnostic sample storehouse comprising all known diagnosis samples at present and build some unknown diagnostic sample, and described known diagnosis sample is respectively used to characterize the known normal condition of machinery, known incipient fault state and known fault conditions, and described unknown diagnostic sample is for characterizing the unknown running status of machinery；With in diagnosis algorithm, when the described diagnostic result that described in-situ processing unit obtains after described diagnostic sample is diagnosed is described unknown running status, determine the mechanical running status that described diagnostic sample characterizes, and after described diagnostic sample is updated to known diagnosis sample, delete the described unknown diagnostic sample similar to the described known diagnosis sample of recent renewal；And respectively in described preparation process and described diagnosis algorithm, according to described known diagnosis sample all of in described diagnostic sample storehouse and described unknown diagnostic sample, train and generate fault diagnosis algorithm, and the diagnostic result of described fault diagnosis algorithm is corresponding with described known normal condition, described known incipient fault state, described known fault conditions and described unknown running status respectively.
8. Diagnosis system of mechanical failure as claimed in claim 7, it is characterised in that described in-situ processing unit includes,
Gather subelement, for obtaining the described diagnostic sample characterizing machinery current operating conditions from scene；
Storing sub-units, for preserving the described diagnostic sample for this diagnosis, and preserves the described fault diagnosis algorithm obtained from described server；
Operator unit, for calling described fault diagnosis algorithm in described storing sub-units and the described diagnostic sample for this diagnosis, and uses described fault diagnosis algorithm that described diagnostic sample is diagnosed, and final exports described diagnostic result；
Communicator unit, for obtaining fault diagnosis algorithm from described server, and when the described diagnostic result of this diagnosis is unknown running status, for sending the described diagnostic sample of this diagnosis to described server.
9. Diagnosis system of mechanical failure as claimed in claim 8, it is characterised in that described server includes,
NE, for setting up and the communication connection of described communicator unit；
Database Unit, for building the diagnostic sample storehouse comprising all known diagnosis samples at present, and storage known diagnosis sample, unknown diagnostic sample and described fault diagnosis algorithm；
Running status confirmation unit, for determining the mechanical running status that described diagnostic sample characterizes, and is updated to known diagnosis sample by described diagnostic sample；
Unknown diagnostic sample construction unit, for the described known diagnosis sample characterizing described known normal condition, carrying out variation process, to build described unknown diagnostic sample；
Unknown diagnostic sample deletes unit, after extracting the described known diagnosis sample of recent renewal and the fault signature of the described unknown diagnostic sample in described diagnostic sample storehouse, compare the frequency band energy ratio of described unknown diagnostic sample and the described fault signature of the described known diagnosis sample of recent renewal, and the frequency band energy of the fault signature described in deletion is than the described unknown diagnostic sample beyond set point；
Fault diagnosis algorithm training unit, for according to all described known diagnosis sample in described diagnostic sample storehouse and described unknown diagnostic sample, training and generate fault diagnosis algorithm, the diagnostic result of wherein said fault diagnosis algorithm is corresponding with described known normal condition, described known incipient fault state, described known fault conditions and described unknown running status respectively；
Fault diagnosis algorithm authentication unit, for checking Sample Storehouse is carried out fault diagnosis, the accuracy rate of diagnosis of the described fault diagnosis algorithm that checking training generates, if accuracy rate of diagnosis is not less than design load, then train successfully, and described fault diagnosis algorithm is stored in Database Unit, otherwise adjust described fault diagnosis algorithm, and be again trained, until train successfully；
Update information transmitting unit, for training after successfully at described fault diagnosis training unit, the information of described fault diagnosis algorithm will be updated, by described NE, send to described in-situ processing unit, so that current described fault diagnosis algorithm is replaced with the described fault diagnosis algorithm trained after successfully by described in-situ processing unit.
10. Diagnosis system of mechanical failure as claimed in claim 8 or 9, it is characterized in that, described in-situ processing subelement also includes sample amendment subelement, for in described diagnosis algorithm, when the mechanical current operating conditions that described diagnostic sample characterizes and described diagnostic result do not correspond, and determine by field service personnel and revise that the Mechanical Running Condition that described diagnostic sample characterizes is after normal condition or incipient fault state or malfunction, described diagnostic sample is sent to described server by described communicator unit, and described diagnostic sample is updated to described known diagnosis sample.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610069441.6A CN105740822B (en) | 2016-02-01 | 2016-02-01 | A kind of mechanical fault diagnosis optimization method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610069441.6A CN105740822B (en) | 2016-02-01 | 2016-02-01 | A kind of mechanical fault diagnosis optimization method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105740822A true CN105740822A (en) | 2016-07-06 |
| CN105740822B CN105740822B (en) | 2019-02-19 |
Family
ID=56242124
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610069441.6A Active CN105740822B (en) | 2016-02-01 | 2016-02-01 | A kind of mechanical fault diagnosis optimization method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105740822B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107092247A (en) * | 2017-06-16 | 2017-08-25 | 温州大学 | A kind of packaging production line method for diagnosing faults based on status data |
| CN110207967A (en) * | 2019-06-13 | 2019-09-06 | 大连海事大学 | A kind of state identification method and system based on wavelet pack energy feature and cross-correlation |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050050423A1 (en) * | 2003-07-14 | 2005-03-03 | Fuji Xerox Co., Ltd. | Fault diagnosis apparatus |
| CN102243140A (en) * | 2011-04-18 | 2011-11-16 | 杨彦利 | Mechanical equipment state monitoring method based on sub-band signal analysis |
| CN102721537A (en) * | 2012-06-01 | 2012-10-10 | 西安交通大学 | Mechanical impact type fault diagnosis method based on variable space-scale frame |
| CN103743554A (en) * | 2013-06-28 | 2014-04-23 | 国家电网公司 | High-voltage circuit breaker mechanical failure diagnosis method based on vibration signal analysis |
| CN103914735A (en) * | 2014-04-17 | 2014-07-09 | 北京泰乐德信息技术有限公司 | Failure recognition method and system based on neural network self-learning |
-
2016
- 2016-02-01 CN CN201610069441.6A patent/CN105740822B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050050423A1 (en) * | 2003-07-14 | 2005-03-03 | Fuji Xerox Co., Ltd. | Fault diagnosis apparatus |
| CN102243140A (en) * | 2011-04-18 | 2011-11-16 | 杨彦利 | Mechanical equipment state monitoring method based on sub-band signal analysis |
| CN102721537A (en) * | 2012-06-01 | 2012-10-10 | 西安交通大学 | Mechanical impact type fault diagnosis method based on variable space-scale frame |
| CN103743554A (en) * | 2013-06-28 | 2014-04-23 | 国家电网公司 | High-voltage circuit breaker mechanical failure diagnosis method based on vibration signal analysis |
| CN103914735A (en) * | 2014-04-17 | 2014-07-09 | 北京泰乐德信息技术有限公司 | Failure recognition method and system based on neural network self-learning |
Non-Patent Citations (1)
| Title |
|---|
| 田涛等: "基于智能理论的高压断路器机械故障诊断", 《江苏电机工程》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107092247A (en) * | 2017-06-16 | 2017-08-25 | 温州大学 | A kind of packaging production line method for diagnosing faults based on status data |
| CN107092247B (en) * | 2017-06-16 | 2019-11-22 | 温州大学 | A kind of packaging production line method for diagnosing faults based on status data |
| CN110207967A (en) * | 2019-06-13 | 2019-09-06 | 大连海事大学 | A kind of state identification method and system based on wavelet pack energy feature and cross-correlation |
| CN110207967B (en) * | 2019-06-13 | 2020-12-01 | 大连海事大学 | State identification method and system based on wavelet packet energy characteristics and cross correlation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105740822B (en) | 2019-02-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| DE102017128693A1 (en) | Feature and limit setting for threat detection in an industrial plant control system | |
| CN107770797A (en) | A kind of association analysis method and system of wireless network alarm management | |
| CN106104398A (en) | Distributed big data in Process Control System | |
| CN110296124B (en) | Remote fault diagnosis system and method based on expert system | |
| WO2021121695A1 (en) | Method, apparatus and system for detecting abnormal operating states of a device | |
| RU2755354C1 (en) | Diagnostic system and method for vehicle data processing | |
| CN113516244B (en) | Intelligent operation and maintenance method and device, electronic equipment and storage medium | |
| Ghosh et al. | FBMTP: An automated fault and behavioral anomaly detection and isolation tool for PLC-controlled manufacturing systems | |
| CN116610092A (en) | Method and system for vehicle analysis | |
| CN113935487B (en) | Visual satellite fault diagnosis knowledge generation method, device and system | |
| CN104462842A (en) | Excavating diagnosis method of failure data based on bayesian network | |
| EP3885854B1 (en) | Data analysis method, device and system | |
| CN111123223A (en) | General development platform, management system and method for radar health management | |
| CN115437339A (en) | Remote diagnosis method and device, electronic equipment and storage medium | |
| CN112860676A (en) | Data cleaning method applied to big data mining and business analysis and cloud server | |
| CN115510950A (en) | Aircraft telemetry data anomaly detection method and system based on time convolution network | |
| CN105740822A (en) | Mechanical fault diagnosis optimization method and system | |
| CN109409780B (en) | Change processing method, device, computer equipment and storage medium | |
| CN114238474A (en) | Data processing method, device and equipment based on drainage system and storage medium | |
| JP2016029516A (en) | Data analysis method and data analysis system | |
| CN105606353A (en) | Mechanical fault diagnosis method and system | |
| CN108427744B (en) | Intelligent data association relation determining method based on information operation and maintenance | |
| CN113360306B (en) | Real-time fault detection method for airborne information system | |
| US20230319085A1 (en) | Attack path generation method and attack path generation device | |
| Hu et al. | Research on application of equipment fault diagnosis technology based on FTA |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: An optimization method and system for mechanical fault diagnosis Effective date of registration: 20220217 Granted publication date: 20190219 Pledgee: Zhejiang Xiaoshan rural commercial bank Limited by Share Ltd. Jingjiang branch Pledgor: HANGZHOU JIE DRIVE TECHNOLOGY Co.,Ltd. Registration number: Y2022330000233 |