CN112052952A - Monitoring parameter optimization selection method in diesel engine fault diagnosis based on genetic algorithm - Google Patents

Monitoring parameter optimization selection method in diesel engine fault diagnosis based on genetic algorithm Download PDF

Info

Publication number
CN112052952A
CN112052952A CN202010837056.8A CN202010837056A CN112052952A CN 112052952 A CN112052952 A CN 112052952A CN 202010837056 A CN202010837056 A CN 202010837056A CN 112052952 A CN112052952 A CN 112052952A
Authority
CN
China
Prior art keywords
diesel engine
fault
monitoring parameters
genetic algorithm
monitoring
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202010837056.8A
Other languages
Chinese (zh)
Other versions
CN112052952B (en
Inventor
王忠巍
张驰
徐荣
冯力东
马龙
倪小明
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.)
Harbin Engineering University
Original Assignee
Harbin Engineering University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Harbin Engineering University filed Critical Harbin Engineering University
Priority to CN202010837056.8A priority Critical patent/CN112052952B/en
Publication of CN112052952A publication Critical patent/CN112052952A/en
Application granted granted Critical
Publication of CN112052952B publication Critical patent/CN112052952B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M15/00—Testing of engines
    • G01M15/04—Testing internal-combustion engines

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biophysics (AREA)
  • Evolutionary Biology (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Genetics & Genomics (AREA)
  • Evolutionary Computation (AREA)
  • Chemical & Material Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Physiology (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Combustion & Propulsion (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

The invention relates to a monitoring parameter optimization selection method in diesel engine fault diagnosis based on a genetic algorithm, and belongs to the field of diesel engine monitoring parameter optimization selection. Firstly, describing a causal relationship between a fault and an abnormal monitoring parameter by adopting a fault characteristic matrix; then screening out monitoring parameters necessary for realizing fault discrimination by utilizing the conditional entropy, and calculating the attribute importance of the remaining monitoring parameters; coding by adopting a binary coding mode; constructing a fitness function of the genetic algorithm according to the requirement of an optimization target and by combining the conditional entropy and the attribute importance; and designing three operators of selection, intersection and variation in genetic operation, realizing optimal calculation and obtaining an optimal sensor arrangement scheme. The method is based on the genetic algorithm, realizes the optimized selection of the monitoring parameters of the diesel engine system, greatly reduces the difficulty of early preparation work, and improves the optimization efficiency.

Description

Monitoring parameter optimization selection method in diesel engine fault diagnosis based on genetic algorithm
Technical Field
The invention relates to a monitoring parameter optimization selection method in diesel engine fault diagnosis based on a genetic algorithm, and belongs to the technical field of diesel engine monitoring parameter optimization selection.
Background
With the continuous improvement of the automation degree and the working performance of the diesel engine, the diesel engine becomes one of the most important power equipment and plays an extremely important role in the fields of ship transportation, engineering machinery, railway machinery, fixed power stations and the like. While the product quality is improved, the economic cost is reduced and the labor efficiency is improved, people can realize that once a diesel engine breaks down, if the diesel engine cannot be processed in time, huge economic loss can be brought, and serious safety accidents of machine damage and human death can be caused in serious cases. Therefore, the normal and ordered foundation and premise of industrial production is to ensure the safe, reliable and efficient operation of the diesel engine. The diesel engine fault diagnosis technology is required by modern production.
The diesel engine fault diagnosis technology comprises three steps: state monitoring, fault detection and fault identification. The state monitoring task is to acquire various parameters representing the running state of the diesel engine by utilizing advanced sensors and signal processing technology, so as to realize the description and evaluation of the running state of the diesel engine. The state monitoring is the first step of diesel engine fault diagnosis, and the output data information is the basis of fault detection and fault identification.
Theoretically, the more the number of signal acquisition devices such as sensors and the like is, the wider the distribution of measurement points is, the more clear the acquired data can reflect the running state of the diesel engine. However, the data volume collected in the state monitoring is increased due to the arrangement of too many sensors, which can also avoid the situations of redundant information volume increase, data information waste, difficulty in distinguishing core information and the like; meanwhile, modern diesel engines develop towards high power density, the structure of the modern diesel engines is more and more compact, and the installation of signal acquisition equipment such as sensors and the like is restricted by the space structure of the diesel engines. Therefore, it is important to select the least number of monitoring parameters to fully describe the diesel engine system for the purpose of fault detection.
A good diesel engine monitoring parameter selection scheme can realize the differentiation of diesel engine faults by using the least monitoring parameters. However, the operation conditions of the diesel engine are complex and changeable, and a complex multidimensional mapping relationship exists between the fault and the abnormal symptom, so that it is very difficult to select an optimal monitoring parameter scheme. The genetic algorithm is used as a heuristic intelligent optimization algorithm, a mathematical model is not needed during optimization calculation, prior knowledge is not needed, and the importance and the sensitivity to faults of each monitoring point can be measured according to the relation between the characteristic value of each measuring point signal and a fault mode, so that the optimization of the selection scheme of the monitoring parameters of the diesel engine by adopting the genetic algorithm is feasible.
Disclosure of Invention
The invention aims to provide a method for optimally selecting monitoring parameters in diesel engine fault diagnosis based on a genetic algorithm, so as to solve the problem of optimally selecting the monitoring parameters in the conventional diesel engine fault diagnosis.
A monitoring parameter optimization selection method in diesel engine fault diagnosis based on a genetic algorithm comprises the following steps:
acquiring a cause-and-effect relationship between a fault and an abnormal monitoring parameter in a diesel engine system by using an expert system, and expressing the cause-and-effect relationship as a fault characteristic matrix;
screening out monitoring parameters necessary for realizing fault discrimination by utilizing the conditional entropy, and then calculating the attribute importance of the remaining monitoring parameters;
thirdly, coding the individuals in a binary coding mode;
fourthly, according to the requirement of the optimization target, combining the conditional entropy and the attribute importance degree to construct a fitness function;
and step five, designing and selecting, crossing and mutating three genetic operators to realize optimization calculation of the genetic algorithm.
Further, in step one, specifically, let a diesel engine system be ∑ S, F, where F ═ F1,…fnIndicates n possible faults of the diesel engine system, S ═ S1,…,skExpressing monitored k operation parameters, utilizing an expert system to obtain the cause-effect relationship between faults and abnormal monitoring parameters in the diesel engine system, and expressing the cause-effect relationship as a fault characteristic matrix R (sigma) (R)ij)n×kThe fault signature is described by a ternary set { +1,0, -1}, i.e.
Figure BDA0002640092360000021
The form of the fault signature matrix is as follows:
Figure BDA0002640092360000031
in the fault signature matrix R (sigma), the rows represent n faults which may occur in the diesel engine, the columns represent k monitored operating parameters of the diesel engine, and the element R in the matrixijThere are three possible values { +1,0, -1}, which respectively indicate when a fault occurs fiAt the time of occurrence, the operating parameter s of the diesel enginejWhether or not the corresponding abnormal change and the form thereof occur, when rijWhen 0, it indicates a failure fiDoes not cause sjChange of (i.e. fault f)iAnd parameter sjNo cause-effect relationship exists between the two; when r isij1 and rijWhen the measured values are-1, the occurrence of a failure f is indicatediTime, operating parameter sjIncrease and decrease compared to normal data.
Further, in step two, the conditional entropy H (F | S) of the failure set F with respect to the full set S of monitoring parameters is calculated, and then the subset { S } of the failure set F with respect to the single monitoring parameter is calculatedtConditional entropy of H (F | { s)t}) to screen out the necessary detection parameters to achieve fault discrimination, and then calculate the attribute importance SGF(s) of the remaining monitoring parametersiS', F) for guiding a search process of a genetic algorithm.
Further, for a diesel engine system, the conditional entropy is defined as follows:
given a diesel engine system ∑ S, F, whereF＝{f1,…fnIndicates n possible faults of the diesel engine system, S ═ S1,…,skDenotes k operating parameters that can be monitored, and the conditional entropy of F with respect to S is defined as:
Figure BDA0002640092360000032
wherein FiIs a quotient F/S ═ F1,F2,…,FmElement in (1), m is the base number of the quotient set F/S;
Figure BDA0002640092360000033
the symbol | represents the cardinality of the set, i.e., the number of elements of the set;
the attribute importance is defined as follows:
given a diesel engine system ∑ S, F, where F ═ F1,…fnIndicates n possible faults of the diesel engine system, S ═ S1,…,skDenotes k operating parameters that can be monitored, let S' be a proper subset of S, i.e. S
Figure BDA0002640092360000044
For any one monitored parameter stEpsilon S-S', its significance SGF (S)tS', F) is defined as: SGF(s)t,S',F)＝H(F|S')-H(D|S'∪{st})。
Further, in step three, specifically, let binary string X ═ X1,…,xk}∈{0,1}kRepresenting a selection scheme of the monitoring parameters, the length of the binary string being equal to the number of candidate monitoring parameters, xi1 means that the monitoring parameter of item i is selected, xj0 means that the jth monitored parameter is not selected, and the coding form is as follows:
Figure BDA0002640092360000041
further, in step four, specifically, according to the optimization objective: monitoring the distinguishing capability of the least number of parameters and no influence on faults, and constructing a fitness function of the genetic algorithm by combining the conditional entropy and the attribute importance:
Figure BDA0002640092360000042
noting that two operators of the fitness function are functions f1,f2I.e. by
Figure BDA0002640092360000043
At function f1Wherein x is an individual in the population; card (x) represents the number of genes whose gene positions in individual x take 1, i.e., the number of selected monitor pairs; k represents the number of all genes in the individual, i.e., the number of all candidate monitoring parameters, and as can be seen from this expression, when the number of selected monitoring parameters is small, f1The larger the value of the function of (c) is,
at function f2H (x) represents that in the case of x individuals, the failure set F is relative to the monitoring parameter subset SxConditional entropy of (a), i.e. H (F | S) ═ H (x)x) In which S isxRepresents a selected combination of monitors in individual x; h0Representing conditional entropy of the fault set F relative to the sensor corpus S, i.e. H0＝H(F|S)，
Alpha and beta are functions f1And function f2The coefficient of (a).
Furthermore, in the fifth step, three genetic operations of selection, crossing and variation are carried out on individuals in the population, excellent individuals are reserved, inferior individuals are eliminated, a new population is formed, and the evolution process of excellence and inferiority is realized.
Further, the selecting operation: and (3) adopting a proportion selection operator to realize selection operation: summing the fitness values of all individuals in the population and calculating the fitness proportion of each individual as the probability that the individual is inherited to the next generation of the population, and then determining the number of times that the individual is selected by adopting a roulette method;
and (3) cross operation: adopting a single-point crossover operator to realize crossover operation: randomly pairing all individuals in the population in pairs, randomly setting a certain gene position as a cross point, and interchanging partial structures of two individuals before and after the cross point according to the cross rate to form two new individuals;
mutation operation: and (3) realizing mutation operation by adopting a few-home mutation operators: for individuals in the population, a certain gene position in the individual is designated as a variation point according to the variation rate, and then the variation point is obtained to carry out inverse operation, namely the variation point is originally 1 and becomes 0 after variation; the mutation point is originally 0 and becomes 1 after mutation, thereby generating a new individual.
The main advantages of the invention are:
1. when the method is adopted for optimizing and selecting the monitoring parameters, a mathematical model of a diesel engine system does not need to be established, and prior knowledge related to the diesel engine is also not needed, so that the difficulty of preparation work is greatly reduced.
2. The invention adopts the genetic algorithm to carry out optimization calculation, can effectively solve the problem of combined explosion among different monitoring parameters when the number of the monitoring parameters of the diesel engine is more, and greatly improves the optimization efficiency.
3. The method can be used for solving the problem of optimizing and selecting the monitoring parameters of the diesel engine system and can also be used in other engineering systems.
Drawings
FIG. 1 is a flow chart of the method for optimizing and selecting the monitoring parameters in the diesel engine fault diagnosis based on the genetic algorithm.
Detailed Description
The technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, a monitoring parameter optimization selection method in diesel engine fault diagnosis based on a genetic algorithm includes the following steps:
acquiring a cause-and-effect relationship between a fault and an abnormal monitoring parameter in a diesel engine system by using an expert system, and expressing the cause-and-effect relationship as a fault characteristic matrix;
screening out monitoring parameters necessary for realizing fault discrimination by utilizing the conditional entropy, and then calculating the attribute importance of the remaining monitoring parameters;
thirdly, coding the individuals in a binary coding mode;
fourthly, according to the requirement of the optimization target, combining the conditional entropy and the attribute importance degree to construct a fitness function;
and step five, designing and selecting, crossing and mutating three genetic operators to realize optimization calculation of the genetic algorithm.
Further, in step one, specifically, let a diesel engine system be ∑ S, F, where F ═ F1,…fnIndicates n possible faults of the diesel engine system, S ═ S1,…,skExpressing monitored k operation parameters, utilizing an expert system to obtain the cause-effect relationship between faults and abnormal monitoring parameters in the diesel engine system, and expressing the cause-effect relationship as a fault characteristic matrix R (sigma) (R)ij)n×kThe fault signature is described by a ternary set { +1,0, -1}, i.e.
Figure BDA0002640092360000061
The form of the fault signature matrix is as follows:
Figure BDA0002640092360000071
in the fault signature matrix R (sigma), the rows represent n faults which may occur in the diesel engine, the columns represent k monitored operating parameters of the diesel engine, and the element R in the matrixijThere are three possible values { +1,0, -1}, which respectively indicate when a fault occurs fiAt the time of occurrence, the operating parameter s of the diesel enginejWhether or not the corresponding abnormal change and the form thereof occur, when rijWhen 0, it indicates a failure fiDoes not cause sjChange of, i.e. failurefiAnd parameter sjNo cause-effect relationship exists between the two; when r isij1 and rijWhen the measured values are-1, the occurrence of a failure f is indicatediTime, operating parameter sjIncrease and decrease compared to normal data.
Further, in step two, the conditional entropy H (F | S) of the failure set F with respect to the full set S of monitoring parameters is calculated, and then the subset { S } of the failure set F with respect to the single monitoring parameter is calculatedtConditional entropy of H (F | { s)t}) to screen out the necessary detection parameters to achieve fault discrimination, and then calculate the attribute importance SGF(s) of the remaining monitoring parametersiS', F) for guiding a search process of a genetic algorithm.
Further, for a diesel engine system, the conditional entropy is defined as follows:
given a diesel engine system ∑ S, F, where F ═ F1,…fnIndicates n possible faults of the diesel engine system, S ═ S1,…,skDenotes k operating parameters that can be monitored, and the conditional entropy of F with respect to S is defined as:
Figure BDA0002640092360000072
wherein FiIs a quotient F/S ═ F1,F2,…,FmElement in (1), m is the base number of the quotient set F/S;
Figure BDA0002640092360000073
the symbol | represents the cardinality of the set, i.e., the number of elements of the set;
the invention introduces the conditional entropy to evaluate the distinguishing capability of the monitoring parameter set on the fault, and the lower the conditional entropy, the stronger the distinguishing capability of the corresponding sensor set on the fault. The invention introduces attribute importance to evaluate the fault distinguishing capability of a single sensor, and the definition of the attribute importance is as follows:
given a diesel engine system ∑ S, F, where F ═ F1,…fnIndicates n possible faults of the diesel engine system, S ═ S1,…,skDenotes k operating parameters that can be monitored, let S' be a proper subset of SI.e. by
Figure BDA0002640092360000084
For any one monitored parameter stEpsilon S-S', its significance SGF (S)tS', F) is defined as: SGF(s)t,S',F)＝H(F|S')-H(D|S'∪{st})。
Further, in step three, specifically, let binary string X ═ X1,…,xk}∈{0,1}kRepresenting a selection scheme of the monitoring parameters, the length of the binary string being equal to the number of candidate monitoring parameters, xi1 means that the monitoring parameter of item i is selected, xj0 means that the jth monitored parameter is not selected, and the coding form is as follows:
Figure BDA0002640092360000081
further, in step four, specifically, according to the optimization objective: 1) the number of monitoring parameters is minimum; 2) the fault distinguishing capability is not influenced, and a fitness function of the genetic algorithm is constructed by combining the conditional entropy and the attribute importance:
Figure BDA0002640092360000082
noting that two operators of the fitness function are functions f1,f2I.e. by
Figure BDA0002640092360000083
At function f1Wherein x is an individual in the population; card (x) represents the number of genes whose gene positions in individual x take 1, i.e., the number of selected monitor pairs; k represents the number of all genes in the individual, i.e., the number of all candidate monitoring parameters, and as can be seen from this expression, when the number of selected monitoring parameters is small, f1The larger the value of the function, and thus, this ensures that the individual has a higher fitness when fewer monitoring parameters are selected.
At function f2H (x) represents that in the case of x individuals, the failure set F is relative to the monitoring parameter subset SxConditional entropy of (a), i.e. H (F | S) ═ H (x)x) In which S isxRepresents a selected combination of monitors in individual x; h0Representing conditional entropy of the fault set F relative to the sensor corpus S, i.e. H0The coefficient τ is introduced to ensure that H (F | S) is closer to H (x) for the conditional entropy H (x)0The individual has higher fitness.
Alpha and beta are functions f1And function f2The coefficient of (a). They have two roles: representing the function weight and adjusting the individual fitness. When calculating the individual fitness, let α ═ λ × μ and β ═ λ × η be1And function f2We can give different weights μ and η to the two sub-functions as needed to ensure that the evolution can progress towards two different directions with the least number of monitoring parameters and the strongest fault discrimination capability. Attribute importance is introduced herein to evaluate the ability of a certain monitoring parameter to distinguish faults. When the monitoring parameter with the highest attribute importance degree is selected in an individual in the evolution process, the individual should be endowed with higher fitness, and the method is realized by setting a lambda value with a larger parameter.
Furthermore, in the fifth step, three genetic operations of selection, crossing and variation are carried out on individuals in the population, excellent individuals are reserved, inferior individuals are eliminated, a new population is formed, and the evolution process of excellence and inferiority is realized.
Further, the selecting operation: and (3) adopting a proportion selection operator to realize selection operation: summing the fitness values of all individuals in the population and calculating the fitness proportion of each individual as the probability that the individual is inherited to the next generation of the population, and then determining the number of times that the individual is selected by adopting a roulette method;
and (3) cross operation: adopting a single-point crossover operator to realize crossover operation: randomly pairing all individuals in the population in pairs, randomly setting a certain gene position as a cross point, and interchanging partial structures of two individuals before and after the cross point according to the cross rate to form two new individuals;
mutation operation: and (3) realizing mutation operation by adopting a few-home mutation operators: for individuals in the population, a certain gene position in the individual is designated as a variation point according to the variation rate, and then the variation point is obtained to carry out inverse operation, namely the variation point is originally 1 and becomes 0 after variation; the mutation point is originally 0 and becomes 1 after mutation, thereby generating a new individual.
The working process of the invention is described below with reference to the accompanying drawing 1:
step 1: and determining a fault set F, a monitoring parameter complete set S and a corresponding fault characteristic matrix R (sigma) of the diesel engine system.
Step 2: the conditional entropy H (F | S) of the set of computed faults F relative to the full set of monitored parameters S is calculated.
And step 3: hypothesis monitoring parameter subsets
Figure BDA0002640092360000101
To pair
Figure BDA0002640092360000102
If H (F | S) ≠ H (F | S- { S)t}) the monitoring parameter is a necessary parameter for achieving fault discrimination, S' ═ S { [ S ] } u {, St}。
And 4, step 4: if H (F | S) ═ H (F | S '), S' is the optimal monitoring parameter selection scheme. Otherwise, step 5 is executed.
And 5: calculating attribute importance SGF(s) of remaining monitoring parametersi,S',F)。
Step 6: searching an optimal monitoring parameter selection scheme by using a genetic algorithm:
step 6.1: randomly generating an initial population P ═ g with the size P1,…,gk]p×kAnd setting s in all individualstThe gene site g corresponding to epsilon StIs 1.
Step 6.2: calculating the group P ═ g according to the fitness function1,…,gk]p×kAnd arranging the individuals in the population in a descending order according to the fitness value.
Step 6.3: finding out the individual with the highest fitness, and recording the corresponding monitoring parameter selection scheme as Sb'estAnd calculating H (F | S)b'est). If H (F | S)b'est) H (F | S), then Sb'estOtherwise 6.4 is executed for optimal monitoring parameter scheme.
Step 6.4: randomly pairing all individuals in the population, randomly setting a certain gene position as a cross point, and according to the cross rate PiAnd exchanging the partial structures of the two individuals before and after the point to form two new individuals.
Step 6.5: randomly pairing all individuals in the population, randomly setting a certain gene position as a cross point, and determining the cross rate pcAnd exchanging the partial structures of the two individuals before and after the point, and generating two new individuals.
Step 6.6: for individuals in the population, the variation rate pmThe gene position in an individual is designated as a mutation point, and then the mutation point is mutated, namely the mutation point is originally 1 and is 0 after mutation, and the mutation point is originally 0 and is 1 after mutation.
Step 6.7: and updating the population and returning to the step 6.2 for repeated execution.
The method is implemented in a diesel engine and used for optimizing and selecting the monitoring parameters in the fault diagnosis of the lubricating oil system. Let the diesel engine lubricating oil system be ∑ S, F, where F ═ F1,…f5Indicates 5 possible faults of the diesel engine system, S ═ S1,…,s8Represents 8 operational parameters that can be monitored. The method provided by the invention is used for optimizing and selecting the monitoring parameters. Table 1 shows the results of the optimized selection of the measuring points in the diesel engine lubricating oil system implemented by the invention.
Figure BDA0002640092360000111
TABLE 1 comparison of results before and after optimization
As can be seen from table 1, the optimal monitoring parameter selection scheme is obtained after optimization calculation is performed by the genetic algorithm. Compared with the monitoring parameters before optimization, the number of the monitoring parameters after optimization is reduced from 8 to 4, and the fault distinguishing capability which is the same as that of the monitoring parameter set before optimization is ensured, which shows that the method is completely suitable for solving the problem of optimizing and selecting the monitoring parameters in the fault diagnosis of the diesel engine.

Claims (8)

1. The method for optimizing and selecting the monitoring parameters in the diesel engine fault diagnosis based on the genetic algorithm is characterized by comprising the following steps of:
acquiring a cause-and-effect relationship between a fault and an abnormal monitoring parameter in a diesel engine system by using an expert system, and expressing the cause-and-effect relationship as a fault characteristic matrix;
screening out monitoring parameters necessary for realizing fault discrimination by utilizing the conditional entropy, and then calculating the attribute importance of the remaining monitoring parameters;
thirdly, coding the individuals in a binary coding mode;
fourthly, according to the requirement of the optimization target, combining the conditional entropy and the attribute importance degree to construct a fitness function;
and step five, designing and selecting, crossing and mutating three genetic operators to realize optimization calculation of the genetic algorithm.
2. The method for optimizing and selecting the monitoring parameters in the diesel engine fault diagnosis based on the genetic algorithm as claimed in claim 1, wherein in the step one, specifically, a diesel engine system is set as ∑ S, F, where F ═ F { (S, F }, where F ═ F { (F) } F { (S, F) } is set1,…fnIndicates n possible faults of the diesel engine system, S ═ S1,…,skExpressing monitored k operation parameters, utilizing an expert system to obtain the cause-effect relationship between faults and abnormal monitoring parameters in the diesel engine system, and expressing the cause-effect relationship as a fault characteristic matrix R (sigma) (R)ij)n×kThe fault signature is described by a ternary set { +1,0, -1}, i.e.
Figure FDA0002640092350000011
The form of the fault signature matrix is as follows:
Figure FDA0002640092350000012
Figure FDA0002640092350000013
in the fault signature matrix R (sigma), the rows represent n faults which may occur in the diesel engine, the columns represent k monitored operating parameters of the diesel engine, and the element R in the matrixijThere are three possible values { +1,0, -1}, which respectively indicate when a fault occurs fiAt the time of occurrence, the operating parameter s of the diesel enginejWhether or not the corresponding abnormal change and the form thereof occur, when rijWhen 0, it indicates a failure fiDoes not cause sjChange of (i.e. fault f)iAnd parameter sjNo cause-effect relationship exists between the two; when r isij1 and rijWhen the measured values are-1, the occurrence of a failure f is indicatediTime, operating parameter sjIncrease and decrease compared to normal data.
3. The method for optimizing and selecting the monitoring parameters in the diesel engine fault diagnosis based on the genetic algorithm as claimed in claim 1, wherein in the second step, the conditional entropy H (F | S) of the fault set F relative to the whole set S of the monitoring parameters is calculated, and then the subset { S } formed by the fault set F relative to the single monitoring parameters is calculatedtConditional entropy of H (F | { s)t}) to screen out the necessary detection parameters to achieve fault discrimination, and then calculate the attribute importance SGF(s) of the remaining monitoring parametersiS', F) for guiding a search process of a genetic algorithm.
4. The method for optimally selecting the monitoring parameters in the diesel engine fault diagnosis based on the genetic algorithm as claimed in claim 3, is characterized in that the conditional entropy is defined as follows for a diesel engine system:
given a diesel engine system ∑ S, F, where F ═ F1,…fnIndicates n possible faults of the diesel engine system, S ═ S1,…,skDenotes k operating parameters that can be monitored, and the conditional entropy of F with respect to S is defined as:
Figure FDA0002640092350000021
wherein FiIs a quotient F/S ═ F1,F2,…,FmElement in (1), m is the base number of the quotient set F/S;
Figure FDA0002640092350000022
the symbol | represents the cardinality of the set, i.e., the number of elements of the set;
the attribute importance is defined as follows:
given a diesel engine system ∑ S, F, where F ═ F1,…fnIndicates n possible faults of the diesel engine system, S ═ S1,…,skDenotes k operating parameters that can be monitored, let S' be a proper subset of S, i.e. S
Figure FDA0002640092350000023
For any one monitored parameter stEpsilon S-S', its significance SGF (S)tS', F) is defined as: SGF(s)t,S',F)＝H(F|S')-H(D|S'∪{st})。
5. The method for optimized selection of monitoring parameters in diesel engine fault diagnosis based on genetic algorithm as claimed in claim 1, wherein in step three, specifically, a binary string X ═ X is set1,…,xk}∈{0,1}kRepresenting a selection scheme of the monitoring parameters, the length of the binary string being equal to the number of candidate monitoring parameters, xi1 means that the monitoring parameter of item i is selected, xj0 means that the jth monitored parameter is not selected, and the coding form is as follows:
Figure FDA0002640092350000031
6. the method for optimizing and selecting the monitoring parameters in the diesel engine fault diagnosis based on the genetic algorithm as claimed in claim 1, wherein in the fourth step, specifically, according to the optimization objective: monitoring the distinguishing capability of the least number of parameters and no influence on faults, and constructing a fitness function of the genetic algorithm by combining the conditional entropy and the attribute importance:
Figure FDA0002640092350000032
noting that two operators of the fitness function are functions f1,f2I.e. by
Figure FDA0002640092350000033
At function f1Wherein x is an individual in the population; card (x) represents the number of genes whose gene positions in individual x take 1, i.e., the number of selected monitor pairs; k represents the number of all genes in the individual, i.e., the number of all candidate monitoring parameters, and as can be seen from this expression, when the number of selected monitoring parameters is small, f1The larger the value of the function of (c) is,
at function f2H (x) represents that in the case of x individuals, the failure set F is relative to the monitoring parameter subset SxConditional entropy of (a), i.e. H (F | S) ═ H (x)x) In which S isxRepresents a selected combination of monitors in individual x; h0Representing conditional entropy of the fault set F relative to the sensor corpus S, i.e. H0＝H(F|S)，
Alpha and beta are functions f1And function f2The coefficient of (a).
7. The method for optimizing and selecting the monitoring parameters in the diesel engine fault diagnosis based on the genetic algorithm as claimed in claim 1, wherein in the fifth step, three genetic operations of selection, crossing and variation are performed on individuals in the population, excellent individuals are reserved, poor individuals are eliminated, a new population is formed, and the evolution process of high-quality and low-quality is realized.
8. The method for optimizing and selecting the monitoring parameters in the diesel engine fault diagnosis based on the genetic algorithm according to claim 7,
selecting operation: and (3) adopting a proportion selection operator to realize selection operation: summing the fitness values of all individuals in the population and calculating the fitness proportion of each individual as the probability that the individual is inherited to the next generation of the population, and then determining the number of times that the individual is selected by adopting a roulette method;
and (3) cross operation: adopting a single-point crossover operator to realize crossover operation: randomly pairing all individuals in the population in pairs, randomly setting a certain gene position as a cross point, and interchanging partial structures of two individuals before and after the cross point according to the cross rate to form two new individuals;
mutation operation: and (3) realizing mutation operation by adopting a few-home mutation operators: for individuals in the population, a certain gene position in the individual is designated as a variation point according to the variation rate, and then the variation point is obtained to carry out inverse operation, namely the variation point is originally 1 and becomes 0 after variation; the mutation point is originally 0 and becomes 1 after mutation, thereby generating a new individual.
CN202010837056.8A 2020-08-19 2020-08-19 Monitoring parameter optimization selection method in diesel engine fault diagnosis based on genetic algorithm Active CN112052952B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010837056.8A CN112052952B (en) 2020-08-19 2020-08-19 Monitoring parameter optimization selection method in diesel engine fault diagnosis based on genetic algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010837056.8A CN112052952B (en) 2020-08-19 2020-08-19 Monitoring parameter optimization selection method in diesel engine fault diagnosis based on genetic algorithm

Publications (2)

Publication Number Publication Date
CN112052952A true CN112052952A (en) 2020-12-08
CN112052952B CN112052952B (en) 2021-12-07

Family

ID=73600412

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010837056.8A Active CN112052952B (en) 2020-08-19 2020-08-19 Monitoring parameter optimization selection method in diesel engine fault diagnosis based on genetic algorithm

Country Status (1)

Country Link
CN (1) CN112052952B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114116370A (en) * 2021-08-31 2022-03-01 西南电子技术研究所（中国电子科技集团公司第十研究所） Method for optimizing operation health state monitoring points of complex electronic system

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105574589A (en) * 2016-01-07 2016-05-11 西安工程大学 Transformer oil chromatogram fault diagnosis method based on ecological niche genetic algorithm
CN109443783A (en) * 2018-10-18 2019-03-08 哈尔滨工业大学 A kind of gas turbine based on priori knowledge is deficient to determine Gas path fault diagnosis method
CN110594107A (en) * 2019-10-24 2019-12-20 内蒙古青电云电力服务有限公司 Wind turbine generator fault detection method and device based on rapid gradient elevator

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105574589A (en) * 2016-01-07 2016-05-11 西安工程大学 Transformer oil chromatogram fault diagnosis method based on ecological niche genetic algorithm
CN109443783A (en) * 2018-10-18 2019-03-08 哈尔滨工业大学 A kind of gas turbine based on priori knowledge is deficient to determine Gas path fault diagnosis method
CN110594107A (en) * 2019-10-24 2019-12-20 内蒙古青电云电力服务有限公司 Wind turbine generator fault detection method and device based on rapid gradient elevator

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
黄新波等: "采用遗传算法优化装袋分类回归树组合算法的变压器故障诊断", 《高电压技术》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114116370A (en) * 2021-08-31 2022-03-01 西南电子技术研究所（中国电子科技集团公司第十研究所） Method for optimizing operation health state monitoring points of complex electronic system
CN114116370B (en) * 2021-08-31 2024-04-26 西南电子技术研究所（中国电子科技集团公司第十研究所） Complex electronic system operation health state monitoring point optimization method

Also Published As

Publication number Publication date
CN112052952B (en) 2021-12-07

Similar Documents

Publication Publication Date Title
CN111237134B (en) Offshore double-fed wind driven generator fault diagnosis method based on GRA-LSTM-stacking model
CN107784325B (en) Spiral fault diagnosis method based on data-driven incremental fusion
CN111709490B (en) Fan health state assessment method based on GRU neural network
CN106780121B (en) Power consumption abnormity identification method based on power consumption load mode analysis
CN111967486A (en) Complex equipment fault diagnosis method based on multi-sensor fusion
CN111614491B (en) Power monitoring system oriented safety situation assessment index selection method and system
CN106373025A (en) Outlier detection-based real-time anti-power-theft monitoring method for power utilization information acquisition system
CN106095639A (en) A kind of cluster subhealth state method for early warning and system
CN115099335A (en) Abnormal identification and feature screening method and system for multi-source heterogeneous data
CN108053148A (en) A kind of efficient diagnostic method of power information system failure
CN109165242B (en) Fault diagnosis and early warning method based on entropy sorting and space-time analysis
CN110555478B (en) Fan multi-fault diagnosis method based on depth measurement network of difficult sample mining
CN109921415A (en) A kind of pernicious online defence method of Data Injection Attacks of power grid towards hybrid measurement
CN112147432A (en) BiLSTM module based on attention mechanism, transformer state diagnosis method and system
CN113792754A (en) Method for processing DGA (differential global alignment) online monitoring data of converter transformer by removing different elements and then repairing
CN112257767A (en) Product key part state classification method aiming at class imbalance data
CN112052952B (en) Monitoring parameter optimization selection method in diesel engine fault diagnosis based on genetic algorithm
CN115146718A (en) Depth representation-based wind turbine generator anomaly detection method
CN112380763A (en) System and method for analyzing reliability of in-pile component based on data mining
Fu et al. Trimming outliers using trees: winning solution of the large-scale energy anomaly detection (LEAD) competition
CN115033591A (en) Intelligent detection method and system for electricity charge data abnormity, storage medium and computer equipment
CN115966074A (en) Machine learning-based shield cutter head mud cake early warning method and system
CN115392349A (en) Fault diagnosis method and device for cutting part of heading machine and heading machine
CN114597886A (en) Power distribution network operation state evaluation method based on interval type two fuzzy clustering analysis
CN107508279B (en) Stability simulation method for power network

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