CN116862284A - Marine diesel engine health state assessment method based on SOM network and rule fusion - Google Patents
Marine diesel engine health state assessment method based on SOM network and rule fusion Download PDFInfo
- Publication number
- CN116862284A CN116862284A CN202310667387.5A CN202310667387A CN116862284A CN 116862284 A CN116862284 A CN 116862284A CN 202310667387 A CN202310667387 A CN 202310667387A CN 116862284 A CN116862284 A CN 116862284A
- Authority
- CN
- China
- Prior art keywords
- diesel engine
- state
- marine diesel
- health state
- feature
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000036541 health Effects 0.000 title claims abstract description 62
- 238000000034 method Methods 0.000 title claims abstract description 28
- 230000004927 fusion Effects 0.000 title claims abstract description 14
- 238000011156 evaluation Methods 0.000 claims abstract description 38
- 238000012549 training Methods 0.000 claims description 32
- 239000011159 matrix material Substances 0.000 claims description 19
- 230000009467 reduction Effects 0.000 claims description 16
- 210000002569 neuron Anatomy 0.000 claims description 14
- 238000012360 testing method Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 7
- 230000000875 corresponding effect Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 230000001186 cumulative effect Effects 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 238000009795 derivation Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 230000002596 correlated effect Effects 0.000 claims description 2
- 238000012544 monitoring process Methods 0.000 abstract description 5
- 238000013210 evaluation model Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000001816 cooling Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 239000000446 fuel Substances 0.000 description 1
- 239000000295 fuel oil Substances 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 239000010687 lubricating oil Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Mathematical Optimization (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computational Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Analysis (AREA)
- Computational Linguistics (AREA)
- Educational Administration (AREA)
- Pure & Applied Mathematics (AREA)
- Biomedical Technology (AREA)
- Strategic Management (AREA)
- Health & Medical Sciences (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Economics (AREA)
- Marketing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Game Theory and Decision Science (AREA)
- Databases & Information Systems (AREA)
- Algebra (AREA)
- General Business, Economics & Management (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
Abstract
According to the marine diesel engine health state assessment method based on SOM network and rule fusion, state monitoring parameter data of a marine diesel engine is taken as input, a health baseline of the marine diesel engine is trained by adopting the SOM network, and an initial health state CV value of the diesel engine is obtained by measuring the distance between real-time data and the health baseline; and synchronously counting alarm information of the diesel engine, and correcting the initial health state CV value of the diesel engine according to rules so as to obtain the final health state CV value of the diesel engine. The application can integrally show the health state of the marine diesel engine, and provides powerful support for the auxiliary decision of users; by fusing the evaluation model and the evaluation rule, the interpretability and the credibility of the evaluation result of the health state of the diesel engine are improved.
Description
Technical Field
The application belongs to the technical field of ship fault prediction, and particularly relates to a marine diesel engine health state assessment method based on SOM network and rule fusion.
Background
The diesel engine is used as a main power source of the ship and is core equipment for guaranteeing safe and reliable navigation of the ship. However, the diesel engine has complex structure and function, variable running conditions and sea conditions, and is extremely easy to generate various anomalies. Meanwhile, with the extension of the service time of the diesel engine, the parts such as the piston and the cylinder of the diesel engine tend to be degraded to different degrees, so that the health state of the diesel engine is reduced, the use requirement of the diesel engine can not be met when the health state is serious, and the completion of the ship navigation task is greatly influenced. The traditional method for evaluating the health state of the diesel engine is mainly divided into a parameter threshold method and an intelligent evaluation method, wherein the reliability of an evaluation result of the parameter threshold method is strongly dependent on the setting of a threshold value, so that the method has stronger artificial subjective factors, and meanwhile, the health state of the whole diesel engine cannot be comprehensively reflected by a single monitoring parameter, so that the method is restricted to be applied to the engineering field to a certain extent; most of intelligent evaluation methods adopt methods such as a neural network and a support vector machine, and the state of health of the diesel engine is evaluated in a data driving mode, but the method does not have interpretability and is not effectively applied to actual engineering because the experience of personnel accumulation of the marine diesel engine in actual use is not fully considered.
Disclosure of Invention
In order to solve the problems in the prior art, the application provides a marine diesel engine health state evaluation method based on SOM network and rule fusion, which comprises the following steps:
constructing an evaluation index system, wherein the evaluation index system comprises characteristic parameters such as temperature, pressure, rotating speed, vibration and efficiency;
feature reduction, namely performing feature reduction on the mutually related feature indexes in the evaluation index system;
constructing a healthy baseline, inputting the evaluation index system with the reduced characteristics into an SOM network for training, and constructing the healthy baseline of the marine diesel engine;
measuring the state deviation, namely measuring the distance between the test sample and the healthy state baseline to obtain the difference between the diesel engine test sample and the normal state of the diesel engine;
judging alarm rules, synchronously collecting alarm information of the marine diesel engine, and correcting the initial health state of the diesel engine according to the alarm information;
and (3) carrying out fusion evaluation, namely fusing the SOM model-based evaluation method and the rule-based evaluation method by adopting the safest principle, so as to obtain a final marine diesel engine health state value.
Further, for marine diesel engines, the correlated characteristic indicators in the evaluation indicator system include temperature, vibration and pressure.
Further, the feature reduction of the feature index system specifically includes:
1) Sample normalization
Assuming that n feature indexes, each having m samples, are obtained, the obtained feature index data is written as an m×n-dimensional data matrix:
wherein a is mn An mth sample that is an nth index feature;
the corresponding standardized data array is:
X＝(X ij ) m×n
where i=1, 2, …, m, j=1, 2, …, n; a, a ij An ith sample that is the jth feature index;is the average of the j-th feature; sigma (sigma) j Standard deviation for the jth feature;
2) Computing a kernel matrix
Firstly, selecting parameters in a Gaussian radial kernel function, calculating a kernel matrix K, and obtaining a kernel matrix KL through correction;
3) KL correlation matrix eigenvalue decomposition
Calculating characteristic value lambda of KL by using Jacobi iteration method 1 ,...,λ n I.e. the corresponding feature vector v 1 ,...,v n . The eigenvalues are ordered in descending order (by selection ordering) to get lambda 1 '>...>λ n ' and correspondingly adjusting the feature vector to obtain v 1 ',...,v n '. Orthogonalization of the feature vector is carried out by the unit of the Schmidt orthogonalization method to obtain alpha 1 ,...,α n ；
4) Nuclear principal component derivation
Calculating the cumulative contribution rate B of the characteristic values 1 ,...,B n According to a given extraction efficiency p, if B t More than or equal to p, t principal components alpha are extracted 1 ,...,α t . Calculating the projection y=kl·α of the corrected kernel matrix KL onto the extracted feature vector, where α= (α) 1 ,...,α t ) The obtained projection Y is the data obtained after the data is subjected to KPCA dimension reduction.
Further, the healthy baseline construction specifically includes:
network initialization, namely randomly assigning a smaller value to an initial connection weight, setting a larger initial neighborhood, establishing a learning rate initial value, and defining a training ending condition;
searching winning neurons, randomly extracting a sample from a training sample set, normalizing, marking the normalized sample as x, calculating the distance between the sample and each output node, determining the winning neighborhood when the distance is the smallest as the winning neurons;
adjusting weights, and adjusting weights of all nodes in the neuron node and the winning neighbor, wherein a weight adjusting formula is as follows:
w i (t+1)＝w i (t)+α(t)(x(t)-w i (t))
wherein w is i (t) represents the weight of the neuron i in the t-th training round, alpha (t) represents the learning rate of the t-th training round, the learning rate is a function of attenuation along with the training, and the numerical range is between 0 and 1;
ending the examination, namely ending the training when the training reaches the initially defined ending condition, if the learning rate reaches a certain preset minimum value or reaches a preset maximum training frequency, returning to the step of searching winning neurons;
after the SOM network training is finished, the healthy base line of the diesel engine is constructed.
Further, the state offset metric specifically includes:
applied to Euclidean distance and N-dimensional vector x when using SOM network for health evaluation 1 ，x 2 The Euclidean distance between the two is calculated as follows:
wherein d is the distance of state offset; n is the number of indexes; x is x 1i An ith index that is a first state vector; x is x 2i An ith index for the second state vector;
and measuring the distance between the test sample and the healthy state baseline by adopting the Euclidean distance calculation formula to obtain the difference between the diesel engine test sample and the normal state of the diesel engine. And (3) calibrating the obtained distance to obtain the initial health state, namely CV value, of the marine diesel engine.
Further:
final marine diesel engine state of health CV value = min (CV Model ，CV Rules of )。
Compared with the prior art, the marine diesel engine health state evaluation method provided by the application has the beneficial effects that:
the application can integrally show the health state of the marine diesel engine, and provides powerful support for the auxiliary decision of users; by fusing the evaluation model and the evaluation rule, the interpretability and the credibility of the evaluation result of the health state of the diesel engine are improved.
Drawings
FIG. 1 is a flow chart of the diesel engine health assessment of the present application;
FIG. 2 is a schematic diagram of a diesel engine evaluation index system according to the present application;
FIG. 3 is a schematic illustration of the characteristic reduction result of the present application;
FIG. 4 is a schematic diagram of a characteristic reduced index system distribution according to the present application;
FIG. 5 is an initial state of health CV value for a PA6 diesel engine;
fig. 6 shows final state of health CV values for PA6 diesel engines.
Detailed Description
The present application will be further described with reference to the drawings and the detailed description below, so that those skilled in the art can better understand the technical solutions of the present application.
Referring to fig. 1, the application provides a method for evaluating the health state of a marine diesel engine, which takes state monitoring parameter data of the marine diesel engine as input, trains a health baseline of the marine diesel engine by adopting an SOM network, and obtains an initial health state CV value of the diesel engine by measuring the distance between real-time data and the health baseline; and synchronously counting alarm information of the diesel engine, and correcting the initial health state CV value of the diesel engine according to rules so as to obtain the final health state CV value of the diesel engine.
The method specifically comprises the following steps:
step 1, constructing an evaluation index system:
according to the running mechanism of the marine diesel engine and the actual conditions of the state monitoring measuring points, the marine diesel engine is respectively developed from the aspects of a complete machine, a cooling system, a lubricating oil system, an air inlet and outlet system, a fuel oil system and the like, and in order to reflect the health state of the diesel engine as comprehensively as possible, an evaluation index system is composed of characteristic parameters such as temperature, pressure, rotating speed, vibration, efficiency and the like;
step 2, feature reduction:
for marine diesel engines, characteristic parameters such as temperature, vibration, pressure and the like are related to each other, so that characteristic index systems need to be subjected to characteristic reduction;
a) Sample normalization
Writing a batch of data of the obtained n characteristic indexes (each index having m samples) into an (m×n) dimension data matrix
Wherein a is mn An mth sample which is an nth feature index;
in engineering, the value range of the measured variable is greatly changed, and the dimensions are different, so that the variable Z is generally standardized before principal component analysis, and the corresponding standardized data array is as follows:
X＝(X ij ) m×n
where i=1, 2, …, m, j=1, 2, …, n; a, a ij An ith sample that is the jth feature index;is the average of the j-th feature; sigma (sigma) j Standard deviation for the jth feature;
b) Computing a kernel matrix
Firstly, selecting parameters in a Gaussian radial kernel function, calculating a kernel matrix K, and obtaining a kernel matrix KL through correction;
c) KL correlation matrix eigenvalue decomposition
Calculating characteristic value lambda of KL by using Jacobi iteration method 1 ,...,λ n I.e. the corresponding feature vector v 1 ,...,v n . The eigenvalues are ordered in descending order (by selection ordering) to get lambda 1 '>...>λ n ' and correspondingly adjusting the feature vector to obtain v 1 ',...,v n '. Orthogonalization of the feature vector is carried out by the unit of the Schmidt orthogonalization method to obtain alpha 1 ,...,α n ；
d) Nuclear principal component derivation
Calculating the cumulative contribution rate B of the characteristic values 1 ,...,B n According to a given extraction efficiency p, if B t More than or equal to p, t principal components alpha are extracted 1 ,...,α t . Calculating the projection y=kl·α of the corrected kernel matrix KL onto the extracted feature vector, where α= (α) 1 ,...,α t ) The method comprises the steps of carrying out a first treatment on the surface of the The obtained projection Y is the data obtained after the data is subjected to KPCA dimension reduction.
Step 3, healthy baseline construction:
inputting the index system with the reduced characteristics into an SOM network for training, so as to construct a healthy baseline of the marine diesel engine;
a) Network initialization: randomly assigning a smaller value to the initial connection weight, setting a larger initial neighborhood (the neighborhood is gradually contracted in the training process), establishing a learning rate initial value, and defining a training ending condition;
b) Searching for winning neurons: randomly extracting a sample from a training sample set, normalizing the sample, marking the normalized sample as x, calculating the distance between the sample and each output node, determining a winning neighborhood when the distance is the smallest as a winning neuron;
c) Adjusting weight values: the weights of all the nodes in the neuron node and the winning neighbor are adjusted, and the weight adjustment formula is as follows:
w i (t+1)＝w i (t)+α(t)(x(t)-w i (t))
wherein w is i (t) represents the weight of the neuron i in the t-th training round, alpha (t) represents the learning rate of the t-th training round, the learning rate is a function of attenuation along with the training, and the numerical range is between 0 and 1;
d) Ending the inspection: ending training when the training reaches the initially defined ending condition, if the learning rate reaches a certain preset minimum value or reaches a preset maximum training frequency, returning to the step b);
after SOM network training is finished, the healthy base line of the diesel engine is constructed;
step 4, measuring state offset:
the distance is used for measuring the state offset, the Euclidean distance is applied to the health evaluation by using the SOM network, and the N-dimensional vector x is used for the health evaluation 1 ，x 2 The Euclidean distance between the two is calculated as follows:
wherein d is the distance of state offset; n is the number of indexes; x is x 1i An ith index that is a first state vector; x is x 2i An ith index for the second state vector;
and measuring the distance between the test sample and the healthy state baseline by adopting the Euclidean distance calculation formula to obtain the difference between the diesel engine test sample and the normal state of the diesel engine. Performing calibration treatment on the obtained distance to obtain an initial health state, namely a CV value, of the marine diesel engine;
step 5, judging alarm rules:
and synchronously collecting alarm information of the marine diesel engine, and correcting the initial health state of the diesel engine according to the alarm information. If a general alarm appears, correcting the CV value of the health state of the diesel engine to be 0.7; if the deceleration alarm occurs, correcting the CV value of the health state of the diesel engine to be 0.6; and if a stopping alarm occurs, correcting the CV value of the health state of the diesel engine to be 0.
Step 6, fusion evaluation:
the two evaluation methods based on the SOM model and the rule are fused by adopting the safest principle, and the final CV value=min (CV Model ，CV Rules of )；
Taking a marine PA6 diesel engine as an example, the specific implementation method of the application comprises the following steps:
constructing a health state evaluation index system of the PA6 diesel engine according to the structural composition, the operation mechanism and the actual condition of the current monitoring parameters of the PA6 diesel engine, as shown in figure 2;
and (3) carrying out standardization processing on the 22-dimensional state characteristic index of the PA6 diesel engine according to the characteristic reduction step, carrying out characteristic contribution rate analysis, and when the contribution rate exceeds 90%, enabling the characteristic reduction to meet the requirement, wherein the characteristic matrix is the result after redundant information is removed. The process of the reduction of the health state characteristics of the PA6 diesel engine is shown in fig. 3, and the distribution condition of the reduced indexes is shown in fig. 4;
inputting the reduced characteristic index into an SOM network for training, obtaining a health state baseline of the PA6 diesel engine after learning is completed, measuring the deviation degree of the test data and the health state baseline by adopting the distance, and obtaining an initial health state deviation value of the PA6 diesel engine after calibration, as shown in figure 5;
and counting alarm information of the PA6 diesel engine, such as information of ultra-high rotation speed parking alarm, general alarm of low fuel inlet pressure, high crankcase pressure deceleration alarm and the like, correcting the health state CV value according to a state judgment rule, and obtaining the final health state CV value of the PA6 diesel engine after the evaluation result based on the SOM network and the judgment result based on the rule are fused according to the safest principle. When the general, the speed reduction and the stop alarm of the diesel engine respectively occur, the health state evaluation result of the PA6 diesel engine based on the application is shown in figure 6.
In summary, the present application is not limited to the preferred embodiments, but is intended to cover modifications and equivalent arrangements included within the scope of the appended claims and their equivalents.
Claims (6)
1. The marine diesel engine health state evaluation method based on SOM network and rule fusion is characterized by comprising the following steps:
constructing an evaluation index system, wherein the evaluation index system comprises characteristic parameters such as temperature, pressure, rotating speed, vibration and efficiency;
feature reduction, namely performing feature reduction on the mutually related feature indexes in the evaluation index system;
constructing a healthy baseline, inputting the evaluation index system with the reduced characteristics into an SOM network for training, and constructing the healthy baseline of the marine diesel engine;
measuring the state deviation, namely measuring the distance between the test sample and the healthy state baseline to obtain the difference between the diesel engine test sample and the normal state of the diesel engine;
judging alarm rules, synchronously collecting alarm information of the marine diesel engine, and correcting the initial health state of the diesel engine according to the alarm information;
and (3) carrying out fusion evaluation, namely fusing the SOM model-based evaluation method and the rule-based evaluation method by adopting the safest principle, so as to obtain a final marine diesel engine health state value.
2. The marine diesel engine health state assessment method based on SOM network and rule fusion according to claim 1, wherein: for marine diesel engines, the correlated characteristic indicators in the evaluation indicator system include temperature, vibration and pressure.
3. The marine diesel engine health state assessment method based on SOM network and rule fusion according to claim 2, wherein the feature reduction of the feature index system specifically comprises:
1) Sample normalization
Assuming that n feature indexes, each having m samples, are obtained, the obtained feature index data is written as an m×n-dimensional data matrix:
wherein a is mn An mth sample that is an nth index feature;
the corresponding standardized data array is:
X＝(X ij ) m×n
where i=1, 2, …, m, j=1, 2, …, n; a, a ij An ith sample that is the jth feature index;is the average of the j-th feature; sigma (sigma) j Standard deviation for the jth feature;
2) Computing a kernel matrix
Firstly, selecting parameters in a Gaussian radial kernel function, calculating a kernel matrix K, and obtaining a kernel matrix KL through correction;
3) KL correlation matrix eigenvalue decomposition
Calculating characteristic value lambda of KL by using Jacobi iteration method 1 ,...,λ n I.e. the corresponding feature vector v 1 ,...,v n . The eigenvalues are ordered in descending order (by selection ordering) to get lambda 1 '>...>λ n ' and correspondingly adjusting the feature vector to obtain v 1 ',...,v n '. Orthogonalization of the feature vector is carried out by the unit of the Schmidt orthogonalization method to obtain alpha 1 ,...,α n ；
4) Nuclear principal component derivation
Calculating the cumulative contribution rate B of the characteristic values 1 ,...,B n According to a given extraction efficiency p, if B t More than or equal to p, t principal components alpha are extracted 1 ,...,α t . Calculating the projection y=kl·α of the corrected kernel matrix KL onto the extracted feature vector, where α= (α) 1 ,...,α t ) The obtained projection Y is the data obtained after the data is subjected to KPCA dimension reduction.
4. The marine diesel engine health state assessment method based on SOM network and rule fusion according to claim 3, wherein the health baseline construction specifically comprises:
network initialization, namely randomly assigning a smaller value to an initial connection weight, setting a larger initial neighborhood, establishing a learning rate initial value, and defining a training ending condition;
searching winning neurons, randomly extracting a sample from a training sample set, normalizing, marking the normalized sample as x, calculating the distance between the sample and each output node, determining the winning neighborhood when the distance is the smallest as the winning neurons;
adjusting weights, and adjusting weights of all nodes in the neuron node and the winning neighbor, wherein a weight adjusting formula is as follows:
w i (t+1)＝w i (t)+α(t)(x(t)-w i (t))
wherein w is i (t) represents the weight of the neuron i in the t-th training round, alpha (t) represents the learning rate of the t-th training round, the learning rate is a function of attenuation along with the training, and the numerical range is between 0 and 1;
ending the examination, namely ending the training when the training reaches the initially defined ending condition, if the learning rate reaches a certain preset minimum value or reaches a preset maximum training frequency, returning to the step of searching winning neurons;
after the SOM network training is finished, the healthy base line of the diesel engine is constructed.
5. The method for estimating a health state of a marine diesel engine based on a SOM network and rule fusion according to claim 4, wherein the state deviation metric specifically comprises:
applied to Euclidean distance and N-dimensional vector x when using SOM network for health evaluation 1 ，x 2 The Euclidean distance between the two is calculated as follows:
wherein d is the distance of state offset; n is the number of indexes; x is x 1i An ith index that is a first state vector; x is x 2i An ith index for the second state vector;
and measuring the distance between the test sample and the healthy state baseline by adopting the Euclidean distance calculation formula to obtain the difference between the diesel engine test sample and the normal state of the diesel engine. And (3) calibrating the obtained distance to obtain the initial health state, namely CV value, of the marine diesel engine.
6. The marine diesel engine health state assessment method based on SOM network and rule fusion according to claim 5, wherein:
final marine diesel engine state of health CV value = min (CV Model ，CV Rules of )。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310667387.5A CN116862284A (en) | 2023-06-07 | 2023-06-07 | Marine diesel engine health state assessment method based on SOM network and rule fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310667387.5A CN116862284A (en) | 2023-06-07 | 2023-06-07 | Marine diesel engine health state assessment method based on SOM network and rule fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116862284A true CN116862284A (en) | 2023-10-10 |
Family
ID=88224171
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310667387.5A Pending CN116862284A (en) | 2023-06-07 | 2023-06-07 | Marine diesel engine health state assessment method based on SOM network and rule fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116862284A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117171921A (en) * | 2023-11-02 | 2023-12-05 | 上海市环境科学研究院 | Method and device for evaluating and processing health state of diesel particulate tail gas purification device |
-
2023
- 2023-06-07 CN CN202310667387.5A patent/CN116862284A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117171921A (en) * | 2023-11-02 | 2023-12-05 | 上海市环境科学研究院 | Method and device for evaluating and processing health state of diesel particulate tail gas purification device |
| CN117171921B (en) * | 2023-11-02 | 2024-01-30 | 上海市环境科学研究院 | Method and device for evaluating and processing health state of diesel particulate tail gas purification device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107941537B (en) | A kind of mechanical equipment health state evaluation method | |
| CN113779496B (en) | Power equipment state evaluation method and system based on equipment panoramic data | |
| CN111648992B (en) | Gas turbine compressor fault identification early warning method | |
| CN109740254B (en) | Ship diesel engine abrasive particle type identification method based on information fusion | |
| CN113033011B (en) | Ship mechanical health state assessment method and system | |
| CN112904810B (en) | Process industry nonlinear process monitoring method based on effective feature selection | |
| CN109298633A (en) | Chemical production process fault monitoring method based on adaptive piecemeal Non-negative Matrix Factorization | |
| CN112364567A (en) | Residual life prediction method based on consistency check of similarity of degraded tracks | |
| CN115187832A (en) | Energy system fault diagnosis method based on deep learning and gram angular field image | |
| CN116862284A (en) | Marine diesel engine health state assessment method based on SOM network and rule fusion | |
| CN114297918A (en) | Aero-engine residual life prediction method based on full-attention depth network and dynamic ensemble learning | |
| CN111680875A (en) | Unmanned aerial vehicle state risk fuzzy comprehensive evaluation method based on probability baseline model | |
| CN113687972A (en) | Method, device and equipment for processing abnormal data of business system and storage medium | |
| CN110570013A (en) | Single-station online wave period data prediction diagnosis method | |
| CN110427019B (en) | Industrial process fault classification method and control device based on multivariate discriminant analysis | |
| CN115375026A (en) | Method for predicting service life of aircraft engine in multiple fault modes | |
| Chou et al. | SHM data anomaly classification using machine learning strategies: A comparative study | |
| CN113673600A (en) | Industrial signal abnormity early warning method, system, storage medium and computing equipment | |
| Bodo et al. | Fault classification driven by maintenance management for smart maintenance applications | |
| CN117633688A (en) | Large-scale power data anomaly detection method based on ridge regression-k-means clustering-LOF-LSTM fusion algorithm | |
| CN115729761B (en) | Hard disk fault prediction method, system, equipment and medium | |
| CN113902145B (en) | Chemical pump maintenance strategy optimization method based on 2-type fuzzy language term | |
| CN116204825A (en) | Production line equipment fault detection method based on data driving | |
| CN114154266B (en) | Gas turbine fault prediction method based on bias rank correlation flow causal structure learning | |
| CN115878986A (en) | Degradation trend evaluation method for variable working condition bearing |
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 |