CN114925787B - Intelligent common rail oil injector fault identification method - Google Patents

Intelligent common rail oil injector fault identification method Download PDF

Info

Publication number
CN114925787B
CN114925787B CN202210844307.4A CN202210844307A CN114925787B CN 114925787 B CN114925787 B CN 114925787B CN 202210844307 A CN202210844307 A CN 202210844307A CN 114925787 B CN114925787 B CN 114925787B
Authority
CN
China
Prior art keywords
value
characteristic
data
characteristic value
historical
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.)
Active
Application number
CN202210844307.4A
Other languages
Chinese (zh)
Other versions
CN114925787A (en
Inventor
庄福如
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.)
Shandong Xin Ya Gelin Baoer Fuel System Co ltd
Original Assignee
Shandong Xin Ya Gelin Baoer Fuel System 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 Shandong Xin Ya Gelin Baoer Fuel System Co ltd filed Critical Shandong Xin Ya Gelin Baoer Fuel System Co ltd
Priority to CN202210844307.4A priority Critical patent/CN114925787B/en
Publication of CN114925787A publication Critical patent/CN114925787A/en
Application granted granted Critical
Publication of CN114925787B publication Critical patent/CN114925787B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/2431—Multiple classes
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/10—Internal combustion engine [ICE] based vehicles
    • Y02T10/40—Engine management systems

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention relates to the technical field of equipment fault identification, in particular to an intelligent common rail fuel injector fault identification method which comprises the steps of obtaining multiple groups of historical characteristic vectors of engines in different years, and calculating the sensitive index of each characteristic value to different fault types and the data reliability of each year according to multiple characteristic values in the historical characteristic vectors; counting a first number of fault types and normal states, and dividing a plurality of groups of historical feature vectors into a plurality of optimal categories equal to the first number; and acquiring a real-time characteristic vector of the engine, respectively substituting the real-time characteristic vector into each optimal class, calculating the membership degree of the real-time characteristic vector in each optimal class, and confirming the fault type of the oil sprayer according to the membership degree. The method avoids the problem of inaccurate classification caused by discrete data, thereby confirming the fault type of the oil injector in real time based on the optimal classification result and greatly improving the efficiency of fault identification.

Description

Intelligent common rail oil injector fault identification method
Technical Field
The invention relates to the technical field of equipment fault identification, in particular to an intelligent common rail oil sprayer fault identification method.
Background
Relevant practices show that the service life of a diesel engine fuel injector can reach about 10 years at most, but in recent years, system faults occur frequently in many diesel engines, the system faults are found through system scientific inspection to be caused by the fuel injector faults, and the working state of the fuel injector directly influences the economy, the dynamic property, the tail gas emission and the like of the diesel engine, so that the fuel injector needs to be subjected to fault detection in time.
Faults of a common oil injector correspond to various phenomena, and a plurality of similar fault phenomena may occur in different faults, so that a great amount of time is needed when fault detection of the oil injector is carried out manually, and a method capable of intelligently identifying fault types of the oil injector is urgently needed.
Disclosure of Invention
In order to solve the technical problems, the invention aims to provide an intelligent common rail injector fault identification method, which adopts the following technical scheme:
acquiring a plurality of groups of historical characteristic vectors of the engine in different years, wherein one group of historical characteristic vectors comprises a plurality of characteristic values; counting the characteristic values in the multiple groups of historical characteristic vectors to respectively obtain the sensitivity index of each characteristic value to different fault types; acquiring a characteristic value set of each characteristic value according to a plurality of groups of historical characteristic vectors under one year, and calculating the data credibility of each year based on all the characteristic value sets under different years;
counting a first number of fault types and normal states, clustering multiple groups of historical feature vectors by using an FCM (fuzzy c-means) algorithm, constructing a loss function by using the sensitive indexes and the data credibility based on a clustering result, and continuously iterating the loss function to obtain a plurality of optimal categories equal to the first number;
and acquiring a real-time characteristic vector of the engine, respectively substituting the real-time characteristic vector into each optimal class, calculating the membership degree of the real-time characteristic vector in each optimal class, and confirming the fault type of the oil sprayer according to the membership degree.
Further, the characteristic values are respectively the oil return flow of the oil injector, the temperature of the diesel engine, the exhaust color of the exhaust pipe, the change time of the exhaust color, the oil injection pressure of the oil injector, the vibration frequency of the engine and the delay time of the engine.
Further, the calculation formula of the sensitivity index is as follows:
Figure 229641DEST_PATH_IMAGE001
wherein,
Figure 956901DEST_PATH_IMAGE002
indicating the characteristic value H versus the fault type
Figure 533376DEST_PATH_IMAGE003
The sensitivity index of (a);
Figure 476055DEST_PATH_IMAGE004
to be in a fault type
Figure 352745DEST_PATH_IMAGE003
The lower eigenvalue H has a value of
Figure 722677DEST_PATH_IMAGE005
The number of occurrences of time;
Figure 52027DEST_PATH_IMAGE006
is a value of the characteristic value H
Figure 63977DEST_PATH_IMAGE005
The total number of occurrences in all historical feature vectors;
Figure 529593DEST_PATH_IMAGE007
the number of kinds of numerical values of the eigenvalue H.
Further, the method for calculating the data reliability of each year based on all feature value sets under different years comprises the following steps:
calculating the characteristic value range and the characteristic value variance of each characteristic value set in the current year to obtain the data standard degree of each characteristic value in the current year, wherein the calculation formula of the data standard degree is as follows:
Figure 319695DEST_PATH_IMAGE008
wherein,
Figure 890003DEST_PATH_IMAGE009
the standard degree of the data is;
Figure 689332DEST_PATH_IMAGE010
is the extreme difference of the characteristic value;
Figure 291346DEST_PATH_IMAGE011
is the variance of the eigenvalue;
Figure 517928DEST_PATH_IMAGE012
the number of characteristic values in the characteristic value set;
Figure 572602DEST_PATH_IMAGE013
is the corresponding numerical value of the current year;
selecting the standard degree of minimum data with the same characteristic value under different years, and taking the characteristic value range and the characteristic value variance corresponding to the standard degree of the minimum data as the standard data of the corresponding characteristic value;
acquiring standard data of each characteristic value, calculating the data deviation degree of each characteristic value in each year based on the standard data, and calculating the data reliability of the current year by combining the data deviation degrees of all the characteristic values in the current year;
the calculation formula of the data deviation degree is as follows:
Figure 910043DEST_PATH_IMAGE014
wherein,
Figure 100984DEST_PATH_IMAGE015
is the degree of data deviation;
Figure 232888DEST_PATH_IMAGE016
the extreme difference of the characteristic value in the standard data;
Figure 771929DEST_PATH_IMAGE017
the variance of the characteristic value in the standard data;
Figure 647481DEST_PATH_IMAGE010
the characteristic value range of the corresponding characteristic value in any one year;
Figure 942196DEST_PATH_IMAGE011
the variance of the eigenvalue of the corresponding eigenvalue in any one year;
Figure 261313DEST_PATH_IMAGE018
to adjust the coefficient;
the calculation formula of the data reliability is as follows:
Figure 274269DEST_PATH_IMAGE019
wherein,
Figure 704244DEST_PATH_IMAGE020
the data credibility;
Figure 119045DEST_PATH_IMAGE021
is as follows
Figure 874642DEST_PATH_IMAGE005
The degree of data deviation of the seed characteristic values;
Figure 109315DEST_PATH_IMAGE022
the number of categories of eigenvalues.
Further, the formula of the loss function is as follows:
Figure 592249DEST_PATH_IMAGE023
wherein,
Figure 869079DEST_PATH_IMAGE024
is a loss function;
Figure 44845DEST_PATH_IMAGE025
the number of historical feature vectors;
Figure 517546DEST_PATH_IMAGE026
is the number of categories;
Figure 804171DEST_PATH_IMAGE027
as historical feature vectors
Figure 678717DEST_PATH_IMAGE028
In a category
Figure 759805DEST_PATH_IMAGE029
Degree of membership in (1);
Figure 985381DEST_PATH_IMAGE030
as historical feature vectors
Figure 810118DEST_PATH_IMAGE028
Data credibility of the belonged year TY;
Figure 801820DEST_PATH_IMAGE031
is a category
Figure 53810DEST_PATH_IMAGE029
Corresponding sensitivity index matrix, i.e. the class by all kinds of characteristic value pairs
Figure 750371DEST_PATH_IMAGE029
A one-dimensional matrix formed by the sensitive indexes of the corresponding fault types;
Figure 129530DEST_PATH_IMAGE032
as historical feature vectors under year TY
Figure 493516DEST_PATH_IMAGE028
；
Figure 401560DEST_PATH_IMAGE033
Is a category
Figure 116575DEST_PATH_IMAGE029
The cluster central point of (2) is also the central historical characteristic vector;
Figure 299426DEST_PATH_IMAGE034
is the Euclidean distance;
Figure 252338DEST_PATH_IMAGE035
is the Hadamard product;
Figure 865372DEST_PATH_IMAGE036
as a result of historical feature vectors
Figure 536525DEST_PATH_IMAGE028
And categories
Figure 523067DEST_PATH_IMAGE029
A one-dimensional difference matrix is formed by the differences of the same eigenvalue among the central historical eigenvectors;
Figure 596065DEST_PATH_IMAGE037
history feature vector in the process of two iterations
Figure 642649DEST_PATH_IMAGE028
Absolute value difference between corresponding numerical values of the categories;
Figure 817410DEST_PATH_IMAGE038
is a category
Figure 591331DEST_PATH_IMAGE029
The degree of dependence on all kinds of characteristic values.
Further, the historical feature vector
Figure 266638DEST_PATH_IMAGE028
In a category
Figure 467812DEST_PATH_IMAGE029
The calculation formula of the membership degree in (1) is as follows:
Figure 864290DEST_PATH_IMAGE039
wherein,
Figure 441902DEST_PATH_IMAGE040
is a category
Figure 240225DEST_PATH_IMAGE041
I.e. the central historical feature vector.
Further, the classes
Figure 612300DEST_PATH_IMAGE029
The calculation formula of the dependence degree on all kinds of characteristic values is as follows:
computing categories based on sets of historical feature vectors
Figure 496074DEST_PATH_IMAGE029
Between any two lower characteristic valuesSetting a sensitive value threshold, and when the sensitive value is greater than the sensitive value threshold, confirming that the corresponding sensitive value is a hypersensitive value;
calculating categories based on hypersensitivity value and amount of hypersensitivity value
Figure 611797DEST_PATH_IMAGE029
And for the dependency degrees of all the characteristic values, the calculation formula of the dependency degree is as follows:
Figure 544854DEST_PATH_IMAGE042
wherein,
Figure 87831DEST_PATH_IMAGE038
is a category
Figure 193321DEST_PATH_IMAGE029
The degree of dependence on all species characteristic values;
Figure 847157DEST_PATH_IMAGE043
is as follows
Figure 620072DEST_PATH_IMAGE005
Individual hypersensitivity value;
Figure 68370DEST_PATH_IMAGE044
is the number of hypersensitivity values.
Further, the formula for calculating the sensitivity value is as follows:
Figure 926736DEST_PATH_IMAGE045
wherein,
Figure 384262DEST_PATH_IMAGE046
is a sensitive value;
Figure 743175DEST_PATH_IMAGE047
is a category
Figure 159113DEST_PATH_IMAGE029
The values of the characteristic value H and the characteristic value W under the corresponding fault types are
Figure 239195DEST_PATH_IMAGE005
The number of occurrences of time;
Figure 234833DEST_PATH_IMAGE048
the values for the characteristic value H and the characteristic value W are
Figure 451182DEST_PATH_IMAGE005
The total number of occurrences in all historical feature vectors;
Figure 506863DEST_PATH_IMAGE049
the combination types of the occurrence of the eigenvalue H and the eigenvalue W.
The embodiment of the invention at least has the following beneficial effects: based on the sensitivity of various characteristic values of the engine to fault types, historical characteristic data are optimally classified, each type corresponds to one fault type, the accuracy of data classification on the whole is guaranteed, the problem of inaccurate classification caused by discrete data is avoided, the fault type of the fuel injector is confirmed in real time based on the optimal classification result, and the efficiency of fault recognition is greatly improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a flowchart illustrating steps of an intelligent common rail injector fault identification method according to an embodiment of the present invention.
Detailed Description
In order to further explain the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description is provided for the method for identifying the fault of the intelligent common rail injector according to the present invention, and the specific implementation manner, structure, features and effects thereof are described in detail with reference to the accompanying drawings and preferred embodiments. In the following description, the different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following describes a specific scheme of the intelligent common rail injector fault identification method provided by the invention in detail with reference to the accompanying drawings.
Referring to fig. 1, a flow chart illustrating steps of an intelligent common rail injector fault identification method according to an embodiment of the invention is shown, and the method includes the following steps:
s001, acquiring multiple groups of historical feature vectors of the engine in different years, wherein one group of historical feature vectors comprises multiple feature values; counting the characteristic values in the multiple groups of historical characteristic vectors to respectively obtain the sensitivity index of each characteristic value to different fault types; and obtaining the characteristic value set of each characteristic value according to a plurality of groups of historical characteristic vectors in one year, and calculating the data credibility of each year based on all the characteristic value sets in different years.
Specifically, when the sprayer breaks down, can cause the influence to each part in the engine, different faults can cause different influences, so can be through the detection index that detects each part of engine, confirm the fault type, wherein, the eigenvalue that needs to detect is: the oil return flow of an oil injector oil return pipe of the engine; the temperature of the diesel engine is obtained by a temperature sensor; the method comprises the following steps of (1) taking a picture of the vicinity of an exhaust pipeline when a diesel engine is started to obtain an exhaust image, and taking the average gray value of the exhaust image as the exhaust color; acquiring the exhaust color after the exhaust color is changed and the exhaust color change time; the oil injection pressure of the oil injector is obtained by the pressure sensor; the vibration frequency of the engine is obtained by a vibration sensor; the time interval between the starting time and the starting vibration time of the engine is the delay time of the engine; these feature values are combined into a feature vector for the engine.
According to common fault types, the fault types of the fuel injector can be divided into: the first is that the atomizer atomizes badly, and the trouble main phenomenon does: the power of the diesel engine is reduced, black smoke is emitted in exhaust, and the running sound of the machine is abnormal; the second is that the needle valve is stuck, and the main phenomena of the fault are as follows: engine power droop, engine shudder; the third is that the needle valve and the guide surface of the needle valve hole are worn, and the main phenomena of the failure are as follows: the power is reduced, and the engine is difficult to start; the fourth is oil dripping of an oil nozzle, and the main fault phenomenon is as follows: when the diesel engine is at low temperature, the diesel engine is difficult to start, the exhaust pipe emits white smoke, and the exhaust pipe becomes black after being heated; the fifth is that the oil return amount is too high, and the main phenomena of the fault are as follows: the oil injection pressure is reduced, the oil injection time is prolonged, and the power of the engine is reduced; the sixth is normal, i.e. no fault.
Furthermore, firstly, historical data, namely a plurality of groups of historical feature vectors of the engine in different years, are collected, and a plurality of groups of historical feature vectors are collected in one year; counting the quantity of each characteristic value of the historical characteristic vectors to respectively obtain the sensitive indexes of each characteristic value to different fault types, wherein the sensitive indexes refer to the special degree of a certain characteristic value to a target fault type, namely, in the range of the characteristic value corresponding to the fault type, and the determinability of the certain characteristic value to the fault type is determined, the calculation formula of the sensitive indexes is as follows:
Figure 74241DEST_PATH_IMAGE050
wherein,
Figure 873570DEST_PATH_IMAGE002
indicating the characteristic value H versus the fault type
Figure 212934DEST_PATH_IMAGE003
Is determined by the sensitivity of the sensor to the measured temperature,
Figure 439516DEST_PATH_IMAGE002
the larger the value of (A), the more the fault type is indicated by the characteristic value
Figure 228612DEST_PATH_IMAGE003
The more sensitive, the more capable of judging whether the fault type occurs according to the characteristic value H
Figure 831632DEST_PATH_IMAGE003
；
Figure 271840DEST_PATH_IMAGE004
To be in a fault type
Figure 420056DEST_PATH_IMAGE003
The lower eigenvalue H has a value of
Figure 211294DEST_PATH_IMAGE005
The number of occurrences of time;
Figure 103158DEST_PATH_IMAGE006
is a value of the characteristic value H
Figure 397873DEST_PATH_IMAGE005
The total number of occurrences in all historical feature vectors;
Figure 714061DEST_PATH_IMAGE007
the number of kinds of numerical values of the eigenvalue H.
And the sensitive indexes of each characteristic value to different fault types can be obtained based on the historical data and a calculation formula of the sensitive indexes.
Dividing historical data according to years to obtain multiple groups of historical feature vectors in each year; acquiring a characteristic value set of each characteristic value in the current year, wherein one characteristic value corresponds to one characteristic value set, calculating characteristic value range and characteristic value variance in the characteristic value set, and calculating the data standard degree of each characteristic value in the current year, wherein the calculation formula of the data standard degree is as follows:
Figure 727016DEST_PATH_IMAGE008
wherein,
Figure 422571DEST_PATH_IMAGE009
the data standard degree is obtained;
Figure 837372DEST_PATH_IMAGE010
the characteristic value is extremely poor;
Figure 592969DEST_PATH_IMAGE011
is the variance of the eigenvalue;
Figure 93221DEST_PATH_IMAGE012
the number of characteristic values in the characteristic value set;
Figure 326887DEST_PATH_IMAGE013
is the corresponding numerical value of the current year.
Taking the eigenvalue H as an example, obtaining the data standard degree of the eigenvalue H in each year based on a calculation formula of the data standard degree, and taking the eigenvalue range and the eigenvalue variance corresponding to the minimum data standard degree as the standard data of the eigenvalue H; and similarly, obtaining standard data corresponding to each characteristic value.
Calculating the data deviation degree of the corresponding characteristic value in each year based on the standard data of each characteristic value, wherein the calculation formula of the data deviation degree is as follows:
Figure 596194DEST_PATH_IMAGE051
wherein,
Figure 513904DEST_PATH_IMAGE015
is the degree of data deviation;
Figure 767031DEST_PATH_IMAGE016
the extreme difference of the characteristic value in the standard data;
Figure 538809DEST_PATH_IMAGE017
the variance of the characteristic value in the standard data;
Figure 193781DEST_PATH_IMAGE010
the characteristic value corresponding to the characteristic value in any year is extremely poor,
Figure 25602DEST_PATH_IMAGE011
the variance of the eigenvalue corresponding to the eigenvalue in any one year;
Figure 766025DEST_PATH_IMAGE018
to adjust the coefficients, the empirical value of 100 is taken.
The data deviation degree of each feature value in the corresponding historical feature vector under one year can be obtained by the calculation formula of the data deviation degree, the data reliability of a plurality of groups of historical feature vectors under the corresponding year is calculated by combining the data deviation degrees of all the feature values under one year, the data reliability refers to the data acquisition accuracy of each feature value in the historical feature vectors, and is used for representing the data reliability acquired under the corresponding year, so that the calculation formula of the data reliability is as follows:
Figure 341494DEST_PATH_IMAGE052
wherein,
Figure 850973DEST_PATH_IMAGE020
the data credibility is obtained;
Figure 850765DEST_PATH_IMAGE021
is a first
Figure 812905DEST_PATH_IMAGE005
The degree of data deviation of the seed characteristic values;
Figure 926486DEST_PATH_IMAGE022
the number of categories of eigenvalues.
The data reliability of each year can be obtained by the calculation formula of the data reliability.
Step S002, counting the first quantity of fault types and normal states, clustering multiple groups of historical characteristic vectors by using an FCM algorithm, constructing a loss function by using sensitive indexes and data credibility based on a clustering result, and continuously iterating the loss function to obtain multiple optimal categories equal to the first quantity.
Specifically, because the acquired historical feature vectors may be discrete and are not well distinguished, the invention classifies a plurality of groups of historical feature vectors by using a clustering algorithm, so that the classified classes have smaller difference, and the analysis of subsequent data is more facilitated.
According to the above, if there are 5 fault types and 1 normal state, the first number of fault types and normal states is 6, so that the FCM algorithm is used to cluster the multiple sets of historical feature vectors to classify the multiple sets of historical feature vectors into 6 categories.
Constructing a loss function based on the clustering result to ensure that 6 optimal classes are obtained by clustering, namely the difference between the classes of the 6 classes is minimum, and then the calculation formula of the loss function is as follows:
Figure 24892DEST_PATH_IMAGE053
wherein,
Figure 464094DEST_PATH_IMAGE024
is a loss function;
Figure 913530DEST_PATH_IMAGE025
the number of historical feature vectors;
Figure 830802DEST_PATH_IMAGE026
as a number of categories；
Figure 783714DEST_PATH_IMAGE027
As historical feature vectors
Figure 377507DEST_PATH_IMAGE028
In a category
Figure 802321DEST_PATH_IMAGE029
Degree of membership in (2);
Figure 772552DEST_PATH_IMAGE030
as historical feature vectors
Figure 330703DEST_PATH_IMAGE028
Data credibility of the belonged year TY;
Figure 360976DEST_PATH_IMAGE031
is a category
Figure 4578DEST_PATH_IMAGE029
Corresponding sensitivity index matrix, i.e. the class by all kinds of characteristic value pairs
Figure 247340DEST_PATH_IMAGE029
A one-dimensional matrix formed by the sensitive indexes of the corresponding fault types;
Figure 440424DEST_PATH_IMAGE032
as historical feature vectors under year TY
Figure 392331DEST_PATH_IMAGE028
；
Figure 772497DEST_PATH_IMAGE033
Is a category
Figure 832332DEST_PATH_IMAGE029
The cluster central point of (2), namely the central historical feature vector;
Figure 614343DEST_PATH_IMAGE034
is the Euclidean distance;
Figure 737151DEST_PATH_IMAGE035
is the Hadamard product;
Figure 604613DEST_PATH_IMAGE036
as a result of historical feature vectors
Figure 454757DEST_PATH_IMAGE028
And categories
Figure 842008DEST_PATH_IMAGE029
A one-dimensional difference matrix is formed by the differences of the same eigenvalue among the central historical eigenvectors;
Figure 119405DEST_PATH_IMAGE037
history feature vector in the process of two iterations
Figure 224896DEST_PATH_IMAGE028
Absolute value difference between corresponding numerical values of the categories;
Figure 878731DEST_PATH_IMAGE038
is a category
Figure 369755DEST_PATH_IMAGE029
The degree of dependence on all kinds of characteristic values.
Wherein the historical feature vector
Figure 849014DEST_PATH_IMAGE028
In a category
Figure 956647DEST_PATH_IMAGE029
Degree of membership in
Figure 899327DEST_PATH_IMAGE027
The calculation formula of (2) is as follows:
Figure 776016DEST_PATH_IMAGE039
wherein,
Figure 411528DEST_PATH_IMAGE040
is a category
Figure 740878DEST_PATH_IMAGE041
I.e. the central historical feature vector.
In the clustering process, in order to highlight the characteristics of each category and enable the subsequent data to be easily classified, the trusted characteristic value and the untrusted characteristic value are required to be more trusted and less untrusted characteristic value are required to be less trusted during clustering, that is, each category has less characteristics, and the category judgment can be realized by a small amount of characteristics, so that the category judgment is simpler, and the category is calculated based on the historical data
Figure 487248DEST_PATH_IMAGE029
Setting a sensitive value threshold value for a sensitive value between any two following characteristic values, and when the sensitive value is greater than the sensitive value threshold value, determining that the sensitive value is a hypersensitive value, wherein the calculation formula of the sensitive value is as follows:
Figure 952864DEST_PATH_IMAGE054
wherein,
Figure 21927DEST_PATH_IMAGE046
is a sensitive value;
Figure 307415DEST_PATH_IMAGE047
is a category
Figure 123055DEST_PATH_IMAGE029
The values of the characteristic value H and the characteristic value W under the corresponding fault types are
Figure 443178DEST_PATH_IMAGE005
The number of occurrences of time;
Figure 669760DEST_PATH_IMAGE048
is the value of the eigenvalue H and the eigenvalue W is
Figure 458856DEST_PATH_IMAGE005
The total number of occurrences in all historical feature vectors;
Figure 796296DEST_PATH_IMAGE049
the combination types of the occurrence of the eigenvalue H and the eigenvalue W.
Calculation of classes based on hypersensitivity values
Figure 987237DEST_PATH_IMAGE029
And the degree of dependence on all kinds of characteristic values is calculated according to the following formula:
Figure 384720DEST_PATH_IMAGE042
wherein,
Figure 644800DEST_PATH_IMAGE038
is a category
Figure 539594DEST_PATH_IMAGE029
The degree of dependence on all kinds of characteristic values;
Figure 99888DEST_PATH_IMAGE043
is as follows
Figure 153426DEST_PATH_IMAGE005
Individual hypersensitivity value;
Figure 431961DEST_PATH_IMAGE044
is the number of hypersensitivity values.
Further, continuously iterating the loss function by using a gradient descent method to cluster and divide the multiple groups of historical feature vectors into 6 optimal categories, wherein each optimal category corresponds to one fault type.
And S003, acquiring a real-time eigenvector of the engine, respectively substituting the real-time eigenvector into each optimal category, calculating the membership degree of the real-time eigenvector in each optimal category, and confirming the fault type of the oil sprayer according to the membership degree.
Specifically, the real-time eigenvectors are obtained from the engine, and are respectively substituted into each optimal category, and the membership degree of the real-time eigenvectors in each optimal category is obtained by using the calculation formula of the membership degree in step S002.
It should be noted that, in the embodiment of the present invention, the data reliability of the year to which the real-time feature vector belongs is set to 1.
Setting a membership threshold, when the membership is greater than the membership threshold, determining that the fault type corresponding to the real-time eigenvector is the fault type corresponding to the optimal category, and similarly, judging each obtained membership, and finally determining the fault type of the oil injector, wherein one or more fault types exist in the oil injector, and the condition that no fault exists is normal.
In summary, the embodiment of the invention provides an intelligent common rail injector fault identification method, which includes acquiring multiple sets of historical feature vectors of engines in different years, and calculating sensitivity indexes of each feature value to different fault types and data reliability of each year according to multiple feature values in the historical feature vectors; counting a first number of fault types and normal states, and dividing a plurality of groups of historical feature vectors into a plurality of optimal categories equal to the first number; and acquiring a real-time characteristic vector of the engine, respectively substituting the real-time characteristic vector into each optimal class, calculating the membership degree of the real-time characteristic vector in each optimal class, and confirming the fault type of the oil sprayer according to the membership degree. The method avoids the problem of inaccurate classification caused by discrete data, thereby confirming the fault type of the oil injector in real time based on the optimal classification result and greatly improving the efficiency of fault identification.
It should be noted that: the sequence of the above embodiments of the present invention is only for description, and does not represent the advantages or disadvantages of the embodiments. And specific embodiments thereof have been described above. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and should not be taken as limiting the scope of the present invention, which is intended to cover any modifications, equivalents, improvements, etc. within the spirit of the present invention.

Claims (6)

1. An intelligent common rail injector fault identification method is characterized by comprising the following specific steps:
acquiring a plurality of groups of historical characteristic vectors of the engine in different years, wherein one group of historical characteristic vectors comprises a plurality of characteristic values; counting the characteristic values in the multiple groups of historical characteristic vectors to respectively obtain the sensitivity index of each characteristic value to different fault types; acquiring a feature value set of each feature value according to a plurality of groups of historical feature vectors in one year, and calculating the data credibility of each year based on all the feature value sets in different years;
counting a first number of fault types and normal states, clustering multiple groups of historical feature vectors by using an FCM (fuzzy c-means) algorithm, constructing a loss function by using the sensitive indexes and the data credibility based on a clustering result, and continuously iterating the loss function to obtain a plurality of optimal categories equal to the first number;
acquiring real-time characteristic vectors of an engine, substituting the real-time characteristic vectors into each optimal category, calculating the membership degree of the real-time characteristic vectors in each optimal category, and confirming the fault type of an oil sprayer according to the membership degree;
the calculation formula of the sensitivity index is as follows:
Figure 848341DEST_PATH_IMAGE002
wherein,
Figure DEST_PATH_IMAGE003
indicating the characteristic value H versus the fault type
Figure 903933DEST_PATH_IMAGE004
The sensitivity index of (2);
Figure DEST_PATH_IMAGE005
to be in a fault type
Figure 892749DEST_PATH_IMAGE004
The lower eigenvalue H has a value of
Figure 129564DEST_PATH_IMAGE006
The number of occurrences of time;
Figure DEST_PATH_IMAGE007
is the value of the characteristic value H
Figure 229239DEST_PATH_IMAGE006
The total number of occurrences in all historical feature vectors;
Figure 618632DEST_PATH_IMAGE008
the number of kinds of numerical values which are characteristic values H;
the method for calculating the data credibility of each year based on all feature value sets under different years comprises the following steps:
calculating the characteristic value range and the characteristic value variance of each characteristic value set in the current year to obtain the data standard degree of each characteristic value in the current year, wherein the calculation formula of the data standard degree is as follows:
Figure 214567DEST_PATH_IMAGE010
wherein,
Figure DEST_PATH_IMAGE011
the data standard degree is obtained;
Figure 502460DEST_PATH_IMAGE012
is the extreme difference of the characteristic value;
Figure DEST_PATH_IMAGE013
is the variance of the eigenvalue;
Figure 966677DEST_PATH_IMAGE014
the number of characteristic values in the characteristic value set;
Figure DEST_PATH_IMAGE015
is a numerical value corresponding to the current year;
selecting minimum data standard degrees of the same characteristic value under different years, and taking characteristic value range and characteristic value variance corresponding to the minimum data standard degrees as standard data of corresponding characteristic values;
acquiring standard data of each characteristic value, calculating the data deviation degree of each characteristic value in each year based on the standard data, and calculating the data reliability of the current year by combining the data deviation degrees of all the characteristic values in the current year;
the calculation formula of the data deviation degree is as follows:
Figure DEST_PATH_IMAGE017
wherein,
Figure 87209DEST_PATH_IMAGE018
is the degree of data deviation;
Figure DEST_PATH_IMAGE019
is a standard numberThe characteristic value is extremely poor;
Figure 417827DEST_PATH_IMAGE020
the variance of the characteristic value in the standard data;
Figure 957131DEST_PATH_IMAGE012
the characteristic value range of the corresponding characteristic value in any one year;
Figure 585558DEST_PATH_IMAGE013
the variance of the eigenvalue of the corresponding eigenvalue in any one year;
Figure DEST_PATH_IMAGE021
is an adjustment factor;
the calculation formula of the data credibility is as follows:
Figure DEST_PATH_IMAGE023
wherein,
Figure 74177DEST_PATH_IMAGE024
the data credibility is obtained;
Figure DEST_PATH_IMAGE025
is a first
Figure 575697DEST_PATH_IMAGE006
The degree of data deviation of the seed characteristic values;
Figure 228395DEST_PATH_IMAGE026
is the number of kinds of eigenvalues.
2. An intelligent common rail injector fault identification method as claimed in claim 1, wherein the characteristic values are respectively an oil return flow of an injector, a diesel engine temperature, an exhaust color of an exhaust pipe, an exhaust color change time, an injection pressure of the injector, a vibration frequency of an engine, and a delay time of the engine.
3. An intelligent common rail injector fault identification method as claimed in claim 1, characterized in that the loss function is calculated by the formula:
Figure 644202DEST_PATH_IMAGE028
wherein,
Figure DEST_PATH_IMAGE029
is a loss function;
Figure 941322DEST_PATH_IMAGE030
the number of historical feature vectors;
Figure DEST_PATH_IMAGE031
is the number of categories;
Figure 118138DEST_PATH_IMAGE032
as historical feature vectors
Figure DEST_PATH_IMAGE033
In a category
Figure 195815DEST_PATH_IMAGE034
Degree of membership in (2);
Figure DEST_PATH_IMAGE035
as historical feature vectors
Figure 212050DEST_PATH_IMAGE033
The data credibility of the belonged year TY;
Figure 35781DEST_PATH_IMAGE036
is a category
Figure 66054DEST_PATH_IMAGE034
The corresponding sensitivity index matrix, i.e. the class is formed by all kinds of characteristic values
Figure 473770DEST_PATH_IMAGE034
A one-dimensional matrix formed by the sensitive indexes of the corresponding fault types;
Figure DEST_PATH_IMAGE037
as historical feature vectors under year TY
Figure 388637DEST_PATH_IMAGE033
；
Figure 830988DEST_PATH_IMAGE038
Is a category
Figure 235425DEST_PATH_IMAGE034
The cluster central point of (2) is also the central historical characteristic vector;
Figure DEST_PATH_IMAGE039
is a Euclidean distance;
Figure 959798DEST_PATH_IMAGE040
is the Hadamard product;
Figure DEST_PATH_IMAGE041
as a result of historical feature vectors
Figure 921764DEST_PATH_IMAGE033
And categories
Figure 172617DEST_PATH_IMAGE034
A one-dimensional difference matrix is formed by the differences of the same eigenvalue among the central historical eigenvectors;
Figure 59539DEST_PATH_IMAGE042
history feature vector in the process of two iterations
Figure 271209DEST_PATH_IMAGE033
Absolute value difference between corresponding numerical values of the categories;
Figure DEST_PATH_IMAGE043
is a category
Figure 432937DEST_PATH_IMAGE034
The degree of dependence on all kinds of characteristic values.
4. An intelligent common rail injector fault identification method as claimed in claim 3, characterized in that the historical feature vector
Figure 335034DEST_PATH_IMAGE033
In a category
Figure 222219DEST_PATH_IMAGE034
The calculation formula of the membership degree in (1) is as follows:
Figure DEST_PATH_IMAGE045
wherein,
Figure 91824DEST_PATH_IMAGE046
is a category
Figure DEST_PATH_IMAGE047
I.e. the central historical feature vector.
5. The intelligent common rail injector fault identification method of claim 3, characterized in that the categories
Figure 938469DEST_PATH_IMAGE034
The calculation formula of the dependence degree on all kinds of characteristic values is as follows:
computing categories based on sets of historical feature vectors
Figure 695072DEST_PATH_IMAGE034
Setting a sensitive value threshold value for a sensitive value between any two next characteristic values, and determining that the corresponding sensitive value is a hypersensitive value when the sensitive value is greater than the sensitive value threshold value;
calculating categories based on hypersensitivity value and amount of hypersensitivity value
Figure 362945DEST_PATH_IMAGE034
And the degree of dependence on all kinds of characteristic values is calculated according to the following formula:
Figure DEST_PATH_IMAGE049
wherein,
Figure 47742DEST_PATH_IMAGE043
is a category
Figure 583897DEST_PATH_IMAGE034
The degree of dependence on all kinds of characteristic values;
Figure 663848DEST_PATH_IMAGE050
is as follows
Figure 329054DEST_PATH_IMAGE006
Individual hypersensitivity value;
Figure DEST_PATH_IMAGE051
is the number of hypersensitivity values.
6. An intelligent common rail injector fault identification method as claimed in claim 5, characterized in that the calculation formula of the sensitivity value is as follows:
Figure DEST_PATH_IMAGE053
wherein,
Figure 579775DEST_PATH_IMAGE054
is a sensitive value;
Figure DEST_PATH_IMAGE055
is a category
Figure 247517DEST_PATH_IMAGE034
The values of the characteristic value H and the characteristic value W under the corresponding fault types are
Figure 448821DEST_PATH_IMAGE006
The number of occurrences of time;
Figure 35660DEST_PATH_IMAGE056
the values for the characteristic value H and the characteristic value W are
Figure 665356DEST_PATH_IMAGE006
The total number of occurrences in all historical feature vectors;
Figure DEST_PATH_IMAGE057
is the kind of combination in which the eigenvalue H and the eigenvalue W occur.
CN202210844307.4A 2022-07-19 2022-07-19 Intelligent common rail oil injector fault identification method Active CN114925787B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210844307.4A CN114925787B (en) 2022-07-19 2022-07-19 Intelligent common rail oil injector fault identification method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210844307.4A CN114925787B (en) 2022-07-19 2022-07-19 Intelligent common rail oil injector fault identification method

Publications (2)

Publication Number Publication Date
CN114925787A CN114925787A (en) 2022-08-19
CN114925787B true CN114925787B (en) 2022-10-11

Family

ID=82815848

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210844307.4A Active CN114925787B (en) 2022-07-19 2022-07-19 Intelligent common rail oil injector fault identification method

Country Status (1)

Country Link
CN (1) CN114925787B (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003322047A (en) * 2002-04-26 2003-11-14 Toyota Motor Corp Failure diagnosing device for intra-cylinder injection type internal-combustion engine
CN111520231A (en) * 2019-12-30 2020-08-11 哈尔滨工程大学 Common rail fuel injector sensitive fault feature extraction method based on CHDE and PWFP
CN111520267A (en) * 2019-12-30 2020-08-11 哈尔滨工程大学 Common rail fuel injector fault diagnosis method based on FOA-VMD and HDE

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4325589B2 (en) * 2004-07-06 2009-09-02 株式会社デンソー Common rail injector
CA2798410C (en) * 2010-05-20 2018-04-03 Cummins Inc. Injector performance test

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003322047A (en) * 2002-04-26 2003-11-14 Toyota Motor Corp Failure diagnosing device for intra-cylinder injection type internal-combustion engine
CN111520231A (en) * 2019-12-30 2020-08-11 哈尔滨工程大学 Common rail fuel injector sensitive fault feature extraction method based on CHDE and PWFP
CN111520267A (en) * 2019-12-30 2020-08-11 哈尔滨工程大学 Common rail fuel injector fault diagnosis method based on FOA-VMD and HDE

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
客车国Ⅲ柴油机电控高压共轨喷油器故障处理(1例);王素英;《商用汽车》;20091004(第10期);全文 *

Also Published As

Publication number Publication date
CN114925787A (en) 2022-08-19

Similar Documents

Publication Publication Date Title
EP0637739B1 (en) Testing an internal combustion engine
Sharma et al. Misfire detection in an IC engine using vibration signal and decision tree algorithms
Şentürk Fuzzy regression control chart based on α-cut approximation
CN110909977A (en) Power grid fault diagnosis method based on ADASYN-DHSD-ET
CN113361944B (en) Safe production index abnormity rapid sensing method based on Internet of things and artificial intelligence
CN118094449B (en) Intelligent production monitoring method, system and device based on industrial Internet
CN115268350A (en) Fault early warning method and system for voltage stabilizing transformer
CN110807245A (en) Automatic modeling method and system for equipment fault early warning
CN117609929B (en) Industrial production line fault online diagnosis method and system based on big data
CN114925787B (en) Intelligent common rail oil injector fault identification method
CN110427019B (en) Industrial process fault classification method and control device based on multivariate discriminant analysis
CN110826600B (en) Engine surge prediction method based on adaptive resonance network online incremental learning
CN115130617B (en) Detection method for continuous increase of self-adaptive satellite data mode
Wang et al. A novel scheme for intelligent fault diagnosis of marine diesel engine using the multi-information fusion technology
CN115861318A (en) Cotton processing production quality detection method
CN116433049A (en) Power consumption abnormality detection method based on fuzzy rough entropy
CN115129503A (en) Equipment fault data cleaning method and system
CN114781083A (en) Engine steady-state data hierarchical analysis and steady-state data characteristic value extraction method
CN114240120A (en) Situation threat level assessment method
CN115761211A (en) Petrochemical enterprise pump room equipment temperature monitoring method and device based on RGB image and thermal imaging coupling
KR100581673B1 (en) Data Classification Method
CN111597934A (en) System and method for processing training data for statistical applications
CN113378956B (en) Equipment degradation data convenient labeling method based on secondary relaxation clustering
CN118244071B (en) Method and system for detecting partial discharge of dry type iron core reactor
CN117009791B (en) Method and system for identifying faults of aeroengine

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
PE01 Entry into force of the registration of the contract for pledge of patent right

Denomination of invention: An Intelligent Method for Fault Identification of Common Rail Injectors

Effective date of registration: 20230406

Granted publication date: 20221011

Pledgee: Liaocheng Branch of Postal Savings Bank of China Co.,Ltd.

Pledgor: SHANDONG XIN YA GELIN BAOER FUEL SYSTEM Co.,Ltd.

Registration number: Y2023980037432

PE01 Entry into force of the registration of the contract for pledge of patent right