CN112036479A - Ship air conditioning system fault identification method and device and storage medium - Google Patents
Ship air conditioning system fault identification method and device and storage medium Download PDFInfo
- Publication number
- CN112036479A CN112036479A CN202010890717.3A CN202010890717A CN112036479A CN 112036479 A CN112036479 A CN 112036479A CN 202010890717 A CN202010890717 A CN 202010890717A CN 112036479 A CN112036479 A CN 112036479A
- Authority
- CN
- China
- Prior art keywords
- evidence
- conditioning system
- vector
- probability distribution
- classification
- 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
- 238000004378 air conditioning Methods 0.000 title claims abstract description 61
- 238000000034 method Methods 0.000 title claims abstract description 46
- 238000013145 classification model Methods 0.000 claims abstract description 48
- 238000012549 training Methods 0.000 claims abstract description 38
- 230000004927 fusion Effects 0.000 claims abstract description 36
- 238000012360 testing method Methods 0.000 claims abstract description 21
- 239000013598 vector Substances 0.000 claims description 68
- 238000005315 distribution function Methods 0.000 claims description 23
- 238000005457 optimization Methods 0.000 claims description 14
- 238000004422 calculation algorithm Methods 0.000 claims description 12
- 230000002068 genetic effect Effects 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 4
- 238000012937 correction Methods 0.000 claims description 4
- 238000003745 diagnosis Methods 0.000 abstract description 9
- 238000011160 research Methods 0.000 abstract description 6
- 230000006870 function Effects 0.000 description 15
- 238000013528 artificial neural network Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 230000015572 biosynthetic process Effects 0.000 description 6
- 238000006243 chemical reaction Methods 0.000 description 6
- 238000003786 synthesis reaction Methods 0.000 description 6
- 238000002474 experimental method Methods 0.000 description 5
- 238000009499 grossing Methods 0.000 description 5
- 239000008186 active pharmaceutical agent Substances 0.000 description 4
- 238000010606 normalization Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 239000003507 refrigerant Substances 0.000 description 3
- 239000000498 cooling water Substances 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000014509 gene expression Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000001308 synthesis method Methods 0.000 description 2
- 230000002194 synthesizing effect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000013075 data extraction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000037213 diet Effects 0.000 description 1
- 235000005911 diet Nutrition 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 238000001704 evaporation Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000004205 output neuron Anatomy 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000005057 refrigeration Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000010189 synthetic method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B63—SHIPS OR OTHER WATERBORNE VESSELS; RELATED EQUIPMENT
- B63B—SHIPS OR OTHER WATERBORNE VESSELS; EQUIPMENT FOR SHIPPING
- B63B79/00—Monitoring properties or operating parameters of vessels in operation
- B63B79/20—Monitoring properties or operating parameters of vessels in operation using models or simulation, e.g. statistical models or stochastic models
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B63—SHIPS OR OTHER WATERBORNE VESSELS; RELATED EQUIPMENT
- B63J—AUXILIARIES ON VESSELS
- B63J2/00—Arrangements of ventilation, heating, cooling, or air-conditioning
- B63J2/02—Ventilation; Air-conditioning
-
- 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
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/257—Belief theory, e.g. Dempster-Shafer
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- Combustion & Propulsion (AREA)
- Chemical & Material Sciences (AREA)
- Ocean & Marine Engineering (AREA)
- Mechanical Engineering (AREA)
- Biomedical Technology (AREA)
- Physiology (AREA)
- Genetics & Genomics (AREA)
- Probability & Statistics with Applications (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Air Conditioning Control Device (AREA)
Abstract
The invention provides a fault identification method, a fault identification device and a storage medium for a ship air conditioning system, wherein the method comprises the following steps: acquiring the operation data of a ship air-conditioning system, and dividing the operation data of the ship air-conditioning system into a training sample and a test sample set; respectively training the three groups of classification models by using the training samples; classifying the test samples by using the trained classification models respectively, wherein the classification result of each classification model correspondingly generates an evidence body; probability distribution is carried out on all the obtained evidence bodies based on the evidence distance, and the fusion weight of each evidence body is obtained; and fusing the corresponding model classification results according to the fusion weight of each evidence body to obtain a final classification result. The invention adopts a data driving method and applies the idea of information fusion to the fault diagnosis research of the air conditioning system.
Description
Technical Field
The invention relates to the technical field of turbine engineering, in particular to a ship air conditioning system fault diagnosis method and device based on improved evidence theory decision-level fusion and a storage medium.
Background
With the development of large-scale, specialized and intelligent ships, the performance requirements on ship equipment are higher and higher. The ship air conditioning system is used as an important auxiliary device and can guarantee air conditioning and diet refrigeration of ships. However, due to the complex internal structure of the air conditioning system and the complex and variable working environment of the ship, the instability and failure of the system operation condition are caused, the living comfort of the crew is reduced, and the service life of the equipment is shortened, so that a fault diagnosis technology of the ship air conditioning system with more excellent performance needs to be researched.
Most of ship air conditioning systems are still overhauled by traditional workers, which not only requires the maintainers to have accurate judgment and rich experience knowledge, but also wastes time and labor when being executed. In addition, the research of the intelligent fault diagnosis technology on the ship air conditioning system is started later, and the application is less. Compared with the prior art, the intelligent advanced technology is used for fault diagnosis of the air conditioning system on land, and particularly, a data driving method is adopted for theoretical research on the background of big data, so that good results are obtained, and the method has good reference significance.
The ship air-conditioning system is highly complex, coupling characteristics, nonlinearity and non-Gaussian exist among system operation data, most of research information sources are single, only one classification model is trained and optimized, longitudinal comparison is carried out, the recognition effect is improved on the basis of the original model, the multiple classification models are transversely compared, the fault judgment is not carried out by integrating the results of the multiple classification models, and the ship air-conditioning system has certain limitation.
Disclosure of Invention
The invention provides a fault identification method and device for a ship air conditioning system and a storage medium. The problem that a single classification model fault recognition result has limitation due to characteristics of a ship air conditioning system is solved.
The technical means adopted by the invention are as follows:
a fault identification method for a ship air conditioning system comprises the following steps:
acquiring the operation data of a ship air conditioning system, and dividing the operation data of the ship air conditioning system into a training sample and a test sample set, wherein the operation data of the ship air conditioning system comprises operation condition data and characteristic variable data;
dividing the training samples into three groups of classification models for training, and taking the characteristic variable data as the input of the models and the operating condition data as the output of the classification models;
classifying the test samples by using the trained classification models respectively, wherein the classification result of each classification model correspondingly generates an evidence body;
probability distribution is carried out on all the obtained evidence bodies based on the evidence distance, and the fusion weight of each evidence body is obtained;
and fusing the corresponding model classification results according to the fusion weight of each evidence body to obtain a final classification result.
Further, the probability distribution is performed on all the obtained evidence bodies based on the evidence distances to obtain the fusion weight of each evidence body, including:
calculating a basic probability distribution function of each evidence body, and constructing an evidence vector according to the basic probability distribution function;
calculating the centroid vector of each evidence body according to the basic probability distribution function;
constructing an evidence weight vector;
calculating the distance between each evidence vector and the centroid vector, constructing an objective optimization function according to the distance between each evidence vector and the centroid vector, and solving the evidence weight vector according to the objective optimization function.
Further, probability distribution is performed on all the obtained evidence bodies based on the evidence distance to obtain the fusion weight of each evidence body, and the method further comprises the following steps:
and correcting the evidence vector according to the evidence weight vector to obtain a corrected evidence vector.
Further, acquiring the operation data of the ship air-conditioning system comprises carrying out standardized processing on the operation data of the ship air-conditioning system, so that the operation data can meet the input requirement of a classification model.
Further, the training of the three groups of classification models by using the training samples respectively further comprises:
and optimizing the trained classification model by using a genetic algorithm.
A ship air conditioning system fault recognition device comprises:
the system comprises a data acquisition unit, a data processing unit and a data processing unit, wherein the data acquisition unit is used for acquiring the operation data of the ship air-conditioning system and dividing the operation data of the ship air-conditioning system into a training sample and a test sample set, and the operation data of the ship air-conditioning system comprises operation condition data and characteristic variable data;
the model training unit is used for respectively training three groups of classification models by utilizing the training samples, and the characteristic variable data is used as the input of the models and the operation condition data is used as the output of the classification models;
the generating unit is used for classifying the test samples by using the trained classification models, and the classification result of each classification model correspondingly generates an evidence body;
the probability distribution unit is used for carrying out probability distribution on all the obtained evidence bodies based on the evidence distances to obtain the fusion weight of each evidence body;
and the fusion unit is used for fusing the corresponding model classification results according to the fusion weight of each evidence body to obtain the final classification result.
Further, the probability distribution unit includes:
the calculation module is used for calculating a basic probability distribution function of each evidence body, constructing an evidence vector according to the basic probability distribution function, and calculating a centroid vector of each evidence body according to the basic probability distribution function;
and the evidence weight vector construction and calculation module is used for constructing an evidence weight vector, calculating the distance between each evidence vector and the centroid vector, constructing an objective optimization function according to the distance between each evidence vector and the centroid vector, and solving the evidence weight vector according to the objective optimization function.
Further, the probability distribution unit further includes:
and the correction module is used for correcting the evidence vector according to the evidence weight vector to obtain a corrected evidence vector.
Further, the model training unit includes:
and the optimization module is used for optimizing the trained classification model by utilizing a genetic algorithm.
A computer-readable storage medium having a set of computer instructions stored therein; the computer instruction set is used for realizing the fault identification method of the ship air conditioning system when being executed by a processor.
Compared with the prior art, the invention has the following advantages:
1. the invention adopts a data driving method and applies the idea of information fusion to the fault diagnosis research of the air conditioning system. On the aspect of feature level fusion, firstly extracting and preprocessing feature samples, then establishing at least two long-tested pattern recognition models, carrying out transverse comparison analysis on recognition results of the models, combining the function optimization capability of an intelligent genetic algorithm on the basis of the models, and trying to improve the classification performance of the models; and in decision-level fusion, hard output of each model is converted into soft probability output, basic probability assignment conversion is carried out according to corresponding criteria, uncertainty information is separated from each model, the output result of each local diagnosis model is used as an evidence body to carry out decision-level fusion through a DS evidence theory, and the result of the fusion decision is used as a final decision of a combined model, so that the fault of the air conditioning system is judged more comprehensively and accurately.
2. The method can achieve the effects of energy conservation, efficiency improvement and emission reduction, can avoid the occurrence of major accidents, and lays a certain foundation for realizing green navigation and safe driving and protecting navigation and the development of unmanned ships.
For the reasons, the invention can be widely popularized in the field of turbine engineering.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
Fig. 1 is a flowchart of a fault identification method for a ship air conditioning system according to the invention.
FIG. 2 is a schematic diagram of the identification result of the improved evidence theory method.
Detailed Description
In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the 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.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
As shown in fig. 1, the present invention provides a fault identification method for a ship air conditioning system, including:
and S1, obtaining the operation data of the ship air conditioning system, and dividing the operation data of the ship air conditioning system into a training sample set and a testing sample set, wherein the operation data of the ship air conditioning system comprises operation condition data and characteristic variable data.
The invention uses the ship air-conditioning system experiment platform to carry out the fault experiment, and 6 working condition samples are preferably collected. Respectively in normal operating mode y1Leakage of suction valve y2Leakage of exhaust valve y3Expansion valve ice plug y4Condenser fouling y5And refrigerant containing non-condensable gas y6。
Some irrelevant and redundant variables are abandoned in the selection of the characteristic variables, characteristics which are complete in information and easy to realize are selected as far as possible, and the selected characteristic variables are 7 in total and respectively: compressor inlet pressure PsCompressor discharge pressure PdIntake air temperature T of condensercAir supply temperature T of air supply bloweraTemperature T of cooling waterinThe outlet water temperature T of the cooling wateroutAnd evaporator evaporating pressure Pe。
And then, carrying out normalization processing on the sample data, wherein the normalization method adopted in the text is min-max normalization (also called dispersion normalization), and mapping the sample data to the range between [0 and 1] so that the operation is faster.
And S2, training three groups of classification models respectively by using the training samples, and taking the characteristic variable data as the input of the models and the operating condition data as the output of the classification models.
As a preferred embodiment of the present invention, a BP neural network classification model (BP), a support vector machine classification model (SVM), and a probabilistic neural network classification model (PNN) are preferably used as the classification models.
Establishing a BP neural network classification model, establishing the BP neural network model by using a training sample, predicting a test input sample by using a sim function to obtain a test output matrix, and converting the matrix into a test output label value, namely a working condition type.
And establishing a classification model of the support vector machine, and determining internal structure parameters, a kernel function, a penalty factor and a smoothing factor of the SVM.
Establishing a probabilistic neural network classification model to determine a smoothing factor of the probabilistic neural network model, establishing a PNN network by using a training sample, and identifying a test sample.
As a more preferable embodiment of the present invention, the three classification models are further optimized by a genetic algorithm in this embodiment. Specifically, the method comprises the following steps: the GA optimizes the BP, the initial weight and the threshold of the traditional BP are randomly given, and a gradient descent learning algorithm is adopted, so that the BP is low in convergence speed and easy to fall into a local minimum point, and therefore the GA can be used for optimizing the weight and the threshold of a BP neural network, the BP is helped to jump out of a possible local minimum point to a certain extent, and the BP is closer to a global optimal solution. The GA optimizes the SVM, the penalty factor C and the kernel parameter sigma have direct influence on an SVM recognition result, the model error is in a trend of decreasing first and then increasing along with the increase of C, and the fitting condition is changed from an over-learning phenomenon to an under-learning phenomenon along with the increase of sigma. Based on a genetic algorithm, the method optimizes the penalty factor and the nuclear parameter of the SVM. Before GA optimizes PNN, the smoothing factor is estimated by empirical statistics or selected by a trial and error method before the establishment of a PNN model, however, the methods have strong subjectivity or are too complicated in process, the calculated amount is increased, and ideal smoothing factor parameters are difficult to obtain. The method adopts a genetic algorithm to optimize the smoothing factor of the PNN, and achieves better classification precision while ensuring the stability of the model
And S3, classifying the test samples by using the trained classification models respectively, and generating an evidence body corresponding to the classification result of each classification model.
And S4, carrying out probability distribution on all the obtained evidence bodies based on the evidence distance to obtain the fusion weight of each evidence body.
Specifically, in most of the previous applications of evidence theories, the basic probability distribution function is established on the experience knowledge of experts, and the individual professional knowledge is used for providing support to various propositions to different degrees, so that the subjectivity is strong, the tendency is inevitable, the fields which are adept by the experts are possibly deviated, and the opinions are different when different problems are met. Aiming at the problem, the method establishes three algorithm models, performs feature level fusion and constructs a basic probability distribution function.
In the embodiment, the probability construction of uncertain propositions is performed by using the network approximation error of the BP neural network, and the listed error formula is as follows:
the DS evidence theoretical probability distribution function is thus constructed as follows:
in the formula, m (A)j) Representing the mass function of the j-th fault, representing that the sample to be detected is identified as AjM (Θ) represents the mass function of uncertainty, represents the fundamental probability distribution of identifying uncertainty, yi、diRespectively representing the output value and the expected value of the ith output neuron.
The basic probability assignment transformation of the SVM is that after an SVM model is built by using a training set, the training set is brought back to the SVM model to obtain a recognition accuracy rate r, and a basic probability distribution of uncertainty propositions is constructed by using a recognition error rate (1-r), which is specifically as follows:
in the formula, r represents the recognition accuracy of the training set,m(Ai) Representing the mass function of the ith fault, representing that the sample to be detected is identified as AiM (Θ) represents the mass function of uncertainty, represents the fundamental probability distribution of identifying uncertainty, piAnd the output of the ith type mode of the sample to be tested on the SVM is shown.
The PNN is subjected to the probability assignment function conversion by using the formulas (1) and (2) with reference to the probability assignment mode of the BP neural network in this embodiment.
The DS evidence theory considers each group of evidence as equally important, however, the support degree provided by different evidences for each proposition is different, even the condition that the evidences are highly conflicted and cannot be effectively synthesized occurs, so that different weight coefficients can be given to each group of evidence to serve as influence factors of each group of evidence in the synthesis process.
Therefore, decision-level weighted fusion based on evidence distance is introduced, the opinions of objective experts are integrated to make final decisions, the problems that different expert systems can give different probability assignments and even severe conflicts are solved, and the fusion between normal evidences cannot be influenced.
(1) The basic probability distribution of each evidence obtained according to the above formulas (2) and (3) is used as the corresponding basic probability distribution function EiAnd setting an identification frame Θ { Y1, Y2.. Yl }, where Y1, Y2.. Yl represents 1 condition that can occur, and 1 is 6 in this embodiment.
(2) Calculating the centroid vector of each evidence body
(3) Assume evidence fusion weight vector W ═ W (W)1，w2，...，wn)；
(4) Evidence vector is corrected to be M'E＝(E′1，E′2，...，E′n)＝(w1E1；w2E2；...；wnEn)；
(5) Calculating the distance d between the 3 groups of evidence vectors and the centroid vectori＝||WiEi-CE||；
(6) Optimizing a function according to a constructed objective
Calculating W;
and S5, fusing the corresponding model classification results according to the fusion weight of each evidence body to obtain a final classification result.
Corresponding to the method, the invention also provides a fault recognition device for the ship air conditioning system, which comprises the following steps: the device comprises a data acquisition unit, a model training unit, a generation unit, a probability distribution unit and a fusion unit. Specifically, the method comprises the following steps:
the system comprises a data acquisition unit and a data processing unit, wherein the data acquisition unit is used for acquiring the operation data of the ship air-conditioning system and dividing the operation data of the ship air-conditioning system into a training sample and a test sample set, and the operation data of the ship air-conditioning system comprises operation condition data and characteristic variable data.
And the model training unit is used for respectively training at least two classification models by utilizing the training samples, and takes the characteristic variable data as the input of the models and the operating condition data as the output of the classification models. Further, the model training unit comprises an optimization module for optimizing the trained classification model by using a genetic algorithm.
And the generating unit is used for classifying the test samples by using the trained classification models, and the classification result of each classification model correspondingly generates an evidence body.
And the probability distribution unit is used for carrying out probability distribution on all the obtained evidence bodies based on the evidence distances to obtain the fusion weight of each evidence body. The probability distribution unit includes:
and the calculation module is used for calculating a basic probability distribution function of each evidence body, constructing an evidence vector according to the basic probability distribution function, and calculating a centroid vector of each evidence body according to the basic probability distribution function.
And the evidence weight vector construction and calculation module is used for constructing an evidence weight vector, calculating the distance between each evidence vector and the centroid vector, constructing an objective optimization function according to the distance between each evidence vector and the centroid vector, and solving the evidence weight vector according to the objective optimization function.
And the correction module is used for correcting the evidence vector according to the evidence weight vector to obtain a corrected evidence vector.
And the fusion unit is used for fusing the corresponding model classification results according to the fusion weight of each evidence body to obtain the final classification result.
The present invention also provides a computer-readable storage medium having a set of computer instructions stored therein; the computer instruction set is used for realizing the fault identification method of the ship air conditioning system when being executed by a processor.
The scheme and effect of the present invention will be further explained by specific application examples.
The method comprises the steps of carrying out fault diagnosis research by taking a ship air conditioning system experiment platform as an object, preprocessing collected data samples and selecting characteristics to form experiment samples, dividing the experiment samples into training samples and testing samples, establishing three models by using the training samples, optimizing initial parameters of each model by using a genetic algorithm, giving initial values to the initial parameters, predicting each testing sample, taking output results of each model as evidence bodies of an evidence theory, carrying out basic probability assignment conversion, carrying out basic probability assignment redistribution on probability output of each model based on an evidence distance probability distribution method, carrying out evidence synthesis, and carrying out fusion decision to obtain a final prediction result.
1. Fault data extraction and processing
A group of sample examples are selected from the experimental samples of 6 working conditions, and are respectively input into each local diagnosis model, and the obtained output results are shown in table 1. Wherein y is0In normal operation, y1For suction valve leakage, y2In order for the exhaust valve to leak,y3is an expansion valve plug, y4For condenser fouling, y5The refrigerant contains non-condensable gas.
2. Setting a basic probability distribution function of an evidence body
For each group of samples, the output results of the local models are respectively used as an evidence body 1, an evidence body 2 and an evidence body 3. The basic probability distribution functions of the evidence bodies 1 and 3 can be respectively given by the expressions (2) and (3), and the basic probability distribution function of the evidence body 2 is given by the expression (3), and the basic probability conversion results of the evidence bodies in each group are shown in table 2.
TABLE 1 local diagnostic model output results
TABLE 2 basic probability distribution function values for evidence bodies
3. Re-assignment of evidence weighting coefficients
Taking the output of each local model in table 1 as an evidence body, the identification framework is Θ ═ { y ═ y0，y1，y2，y3，y4，y5The centroid vector of the above three sets of evidences is calculated as M ═ M (0.0810, 0.3854, 0.0559, 0.0525, 0.0285, 0.3967), and the weight vector is calculated as W ═ W (W ═ W1，w2，w3) The original evidence can then be modified as shown in table 3 below.
TABLE 3 weight correction for each evidence body
Calculating the distance d between the corrected evidence and the centroid vectori＝||WiEi-CE | |, and solving the weight vector of the evidence body group as: w ═ (0.3392, 0.2823, 0.3785), and converted into a phaseFor the weight vector: (w)1，w2，w3)＝(0.8962，0.7458，1)。
4. Information fusion based on evidence distance method
The relative weights are brought into the original evidence table 3 for evidence synthesis, and the results are shown in table 4.
TABLE 4 synthetic results based on evidence distance
As can be seen from the above table synthesis results, the evidence body 1, the evidence body 2 and the evidence body 3 are respectively given relative weight coefficients 0.8962, 0.7458 and 1 as credibility factors in the evidence synthesis process, and convert the non-credible parts into uncertainty probabilities m (Θ) of the evidence bodies. It can be seen that under the rule of synthesizing the evidence distances, the relative weight coefficient of the evidence body 2 with misjudgment is the lowest, so that the uncertainty probability is the highest, the influence degree is the lowest in the evidence synthesizing process, and the evidence bodies 1 and 3 have correct judgment y5The supporting degree is highest, the influence degree is larger, and the reason that the whole decision-making system can correctly identify the fault working condition is.
5. Validation of improved evidence distance synthesis method
The simulation was performed on all test samples using this synthesis method, and the results are shown in fig. 2. The average accuracy of various working conditions can be calculated to be 94.58% by using the method, wherein the identification accuracy of the fault working condition of the refrigerant containing the non-condensable gas is improved to 92.5% from 90% of that of the traditional synthetic method.
The evidence distance-based synthesis rule does not need to perform basic probability assignment conversion on each specific model, can play a role in processing the high-conflict evidence, and can stably fuse ideal results while retaining information lost due to probability assignment conversion. Compared with the prior art, the traditional DS fusion decision is weak in the face of conflicting evidences, but the reason is that each evidence body is regarded as equally important, however, the influence of the evidence body on the final decision is different, and the advantages of local experts are fully played in the practical application process to perform advantage complementation, so that the most favorable decision is made.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A fault identification method for a ship air conditioning system is characterized by comprising the following steps:
acquiring the operation data of a ship air conditioning system, and dividing the operation data of the ship air conditioning system into a training sample and a test sample set, wherein the operation data of the ship air conditioning system comprises operation condition data and characteristic variable data;
training three groups of classification models respectively by using the training samples, and taking the characteristic variable data as the input of the models and the operation condition data as the output of the classification models;
classifying the test samples by using the trained classification models respectively, wherein the classification result of each classification model correspondingly generates an evidence body;
probability distribution is carried out on all the obtained evidence bodies based on the evidence distance, and the fusion weight of each evidence body is obtained;
and fusing the corresponding model classification results according to the fusion weight of each evidence body to obtain a final classification result.
2. The ship air conditioning system fault identification method according to claim 1, wherein the probability distribution is performed on all obtained evidence bodies based on evidence distances to obtain the fusion weight of each evidence body, and the method comprises the following steps:
calculating a basic probability distribution function of each evidence body, and constructing an evidence vector according to the basic probability distribution function;
calculating the centroid vector of each evidence body according to the basic probability distribution function;
constructing an evidence weight vector;
calculating the distance between each evidence vector and the centroid vector, constructing an objective optimization function according to the distance between each evidence vector and the centroid vector, and solving the evidence weight vector according to the objective optimization function.
3. The ship air conditioning system fault identification method according to claim 2, wherein probability distribution is performed on all obtained evidence bodies based on evidence distances to obtain a fusion weight of each evidence body, and further comprising:
and correcting the evidence vector according to the evidence weight vector to obtain a corrected evidence vector.
4. The method for identifying the fault of the ship air-conditioning system according to claim 1, wherein the step of obtaining the operation data of the ship air-conditioning system comprises the step of carrying out standardized processing on the operation data of the ship air-conditioning system so that the operation data can meet the input requirements of a classification model.
5. The method for identifying the fault of the ship air conditioning system according to claim 1, wherein the three groups of classification models are respectively trained by using training samples, and further comprising:
and optimizing the trained classification model by using a genetic algorithm.
6. A ship air conditioning system fault recognition device is characterized by comprising:
the system comprises a data acquisition unit, a data processing unit and a data processing unit, wherein the data acquisition unit is used for acquiring the operation data of the ship air-conditioning system and dividing the operation data of the ship air-conditioning system into a training sample and a test sample set, and the operation data of the ship air-conditioning system comprises operation condition data and characteristic variable data;
the model training unit is used for respectively training three groups of classification models by utilizing the training samples, and the characteristic variable data is used as the input of the models and the operation condition data is used as the output of the classification models;
the generating unit is used for classifying the test samples by using the trained classification models, and the classification result of each classification model correspondingly generates an evidence body;
the probability distribution unit is used for carrying out probability distribution on all the obtained evidence bodies based on the evidence distances to obtain the fusion weight of each evidence body;
and the fusion unit is used for fusing the corresponding model classification results according to the fusion weight of each evidence body to obtain the final classification result.
7. The marine air conditioning system fault identification device of claim 6, wherein the probability distribution unit comprises:
the calculation module is used for calculating a basic probability distribution function of each evidence body, constructing an evidence vector according to the basic probability distribution function, and calculating a centroid vector of each evidence body according to the basic probability distribution function;
and the evidence weight vector construction and calculation module is used for constructing an evidence weight vector, calculating the distance between each evidence vector and the centroid vector, constructing an objective optimization function according to the distance between each evidence vector and the centroid vector, and solving the evidence weight vector according to the objective optimization function.
8. The marine air conditioning system fault identification device of claim 7, wherein the probability distribution unit further comprises:
and the correction module is used for correcting the evidence vector according to the evidence weight vector to obtain a corrected evidence vector.
9. The marine air conditioning system fault recognition device of claim 6, wherein the model training unit comprises:
and the optimization module is used for optimizing the trained classification model by utilizing a genetic algorithm.
10. A computer-readable storage medium having a set of computer instructions stored therein; the set of computer instructions, when executed by a processor, implements the marine air conditioning system fault identification method of any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010890717.3A CN112036479A (en) | 2020-08-29 | 2020-08-29 | Ship air conditioning system fault identification method and device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010890717.3A CN112036479A (en) | 2020-08-29 | 2020-08-29 | Ship air conditioning system fault identification method and device and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112036479A true CN112036479A (en) | 2020-12-04 |
Family
ID=73586309
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010890717.3A Pending CN112036479A (en) | 2020-08-29 | 2020-08-29 | Ship air conditioning system fault identification method and device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112036479A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112700162A (en) * | 2021-01-12 | 2021-04-23 | 上海电气集团股份有限公司 | Method and device for evaluating running state of rail transit air conditioner |
| CN116975502A (en) * | 2023-09-21 | 2023-10-31 | 天津津轨汇海科技发展有限公司 | Energy efficiency monitoring management system for electromechanical equipment of subway train air conditioning system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110530631A (en) * | 2019-08-21 | 2019-12-03 | 贵州大学 | A kind of gear list type fault detection method based on hybrid classifer |
| CN111368885A (en) * | 2020-02-24 | 2020-07-03 | 大连理工大学 | Aero-engine gas circuit fault diagnosis method based on deep learning and information fusion |
-
2020
- 2020-08-29 CN CN202010890717.3A patent/CN112036479A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110530631A (en) * | 2019-08-21 | 2019-12-03 | 贵州大学 | A kind of gear list type fault detection method based on hybrid classifer |
| CN111368885A (en) * | 2020-02-24 | 2020-07-03 | 大连理工大学 | Aero-engine gas circuit fault diagnosis method based on deep learning and information fusion |
Non-Patent Citations (6)
| Title |
|---|
| 卢卡·萨巴等: "卵巢肿瘤影像学", 31 March 2017, 天津科技翻译出版有限公司, pages: 385 * |
| 司马莉萍等: "基于SVM和D-S证据理论的电力变压器内部故障部位识别", 电力自动化设备, 30 November 2012 (2012-11-30), pages 72 - 77 * |
| 孔慧芳等: "基于神经网络和证据理论的DCT电控系统故障诊断", 合肥工业大学学报（自然科学版）, 30 September 2015 (2015-09-30), pages 1193 - 1196 * |
| 张梅军等: "EEMD方法及其在机械故障诊断中的应用", 31 October 2015, 国防工业出版社, pages: 128 * |
| 张永礼等: "神经网络优化算法在技术经济领域中的应用", 30 November 2015, 冶金工业出版社, pages: 35 * |
| 王小艺等: "一种基于最优权重分配的D-S改进算法", 系统工程理论与实践, 30 November 2006 (2006-11-30), pages 103 - 107 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112700162A (en) * | 2021-01-12 | 2021-04-23 | 上海电气集团股份有限公司 | Method and device for evaluating running state of rail transit air conditioner |
| CN116975502A (en) * | 2023-09-21 | 2023-10-31 | 天津津轨汇海科技发展有限公司 | Energy efficiency monitoring management system for electromechanical equipment of subway train air conditioning system |
| CN116975502B (en) * | 2023-09-21 | 2023-12-19 | 天津津轨汇海科技发展有限公司 | Energy efficiency monitoring management system for electromechanical equipment of subway train air conditioning system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113256066B (en) | PCA-XGboost-IRF-based job shop real-time scheduling method | |
| CN111680875B (en) | Unmanned aerial vehicle state risk fuzzy comprehensive evaluation method based on probability baseline model | |
| CN110837223A (en) | Combustion optimization control method and system for gas turbine | |
| CN108256173A (en) | A kind of Gas path fault diagnosis method and system of aero-engine dynamic process | |
| CN111140986A (en) | Operating state detection method and device of air conditioning system, storage medium and air conditioner | |
| CN112036479A (en) | Ship air conditioning system fault identification method and device and storage medium | |
| CN115099296A (en) | Sea wave height prediction method based on deep learning algorithm | |
| CN111597760A (en) | Method for obtaining gas path parameter deviation value under small sample condition | |
| CN111340069A (en) | Incomplete data fine modeling and missing value filling method based on alternate learning | |
| CN111242204A (en) | Operation and maintenance management and control platform fault feature extraction method | |
| CN115751441A (en) | Heat supply system heating station heat regulation method and system based on secondary side flow | |
| CN114973350B (en) | Cross-domain facial expression recognition method irrelevant to source domain data | |
| CN110826587B (en) | Vortex shaft engine fault detection method based on improved weighted support vector machine | |
| CN114219228B (en) | Stadium evacuation evaluation method based on EM clustering algorithm | |
| CN111695288A (en) | Transformer fault diagnosis method based on Apriori-BP algorithm | |
| CN118038130B (en) | Mixed expert pathology large model system and method based on sparse routing | |
| CN115204041A (en) | Large aircraft approach and approach stage oil consumption prediction method | |
| CN110929761A (en) | Balance method for collecting samples in situation awareness framework of intelligent system security system | |
| CN113469252A (en) | Extra-high voltage converter valve operation state evaluation method considering unbalanced samples | |
| CN110414407B (en) | Ship propeller air entrainment identification method based on evidence reasoning rule and self-adaptive lifting | |
| CN117112992A (en) | Fault diagnosis method for polyester esterification stage | |
| CN114021445B (en) | Ocean vortex mixing non-locality prediction method based on random forest model | |
| CN113742860B (en) | Scroll engine power estimation method based on DBN-Bayes algorithm | |
| CN114861977A (en) | Distillation integration mode perception algorithm and system for unbalanced power data | |
| CN115479773A (en) | Aeroengine cross-domain fault diagnosis method based on dynamic countermeasure residual error 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 |