CN113177594B - Air conditioner fault diagnosis method based on Bayesian optimization PCA-extreme random tree - Google Patents

Air conditioner fault diagnosis method based on Bayesian optimization PCA-extreme random tree Download PDF

Info

Publication number
CN113177594B
CN113177594B CN202110475880.8A CN202110475880A CN113177594B CN 113177594 B CN113177594 B CN 113177594B CN 202110475880 A CN202110475880 A CN 202110475880A CN 113177594 B CN113177594 B CN 113177594B
Authority
CN
China
Prior art keywords
pca
tree
fault diagnosis
random tree
air conditioner
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110475880.8A
Other languages
Chinese (zh)
Other versions
CN113177594A (en
Inventor
陆玲霞
秦锋
季文献
于淼
韩宝慧
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.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
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 Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN202110475880.8A priority Critical patent/CN113177594B/en
Publication of CN113177594A publication Critical patent/CN113177594A/en
Application granted granted Critical
Publication of CN113177594B publication Critical patent/CN113177594B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/24323—Tree-organised classifiers
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F11/00—Control or safety arrangements
    • F24F11/30—Control or safety arrangements for purposes related to the operation of the system, e.g. for safety or monitoring
    • F24F11/32—Responding to malfunctions or emergencies
    • F24F11/38—Failure diagnosis
    • 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
    • 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
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G06F18/24155—Bayesian classification

Abstract

The invention discloses a Bayesian optimization-based PCA-extreme random tree air conditioner fault diagnosis method, which comprises the following steps of: 1) acquiring and normalizing operation data of the air conditioner under normal and different faults; 2) reducing the dimension of the normalized data through a PCA algorithm and then using the reduced dimension as the input of an extreme random tree (ExtraTree) model; 3) establishing a limit random tree classification model, training and testing the classifier to obtain a PCA-limit random tree fault diagnosis model for the air conditioner; 4) optimizing the feature quantity and the CART decision tree quantity after PCA dimensionality reduction of the PCA-extreme random tree fault diagnosis model by using a Bayesian optimization algorithm to obtain the optimal feature quantity and the CART decision tree quantity after dimensionality reduction; 5) then, the calculated optimal PCA dimensionality reduction feature quantity value and the calculated CART decision tree quantity value are used as parameters of a PCA-limit random tree model and train samples to obtain a PCA-limit random tree fault diagnosis model, and then the diagnosis model can be used for diagnosing real-time data.

Description

Air conditioner fault diagnosis method based on Bayesian optimization PCA-extreme random tree
Technical Field
The invention relates to the field of fault diagnosis of heating, ventilating and air conditioning systems, in particular to a PCA-extreme random tree air conditioning fault diagnosis method based on Bayesian optimization.
Background
At present, the heating, ventilating and air conditioning system of a large public building is often provided with a plurality of components and a complex structure, and comprises cold and heat source equipment, air treatment equipment, an air conditioning air system, an air conditioning water system, a control and regulation device and the like. The heating, ventilation and air conditioning system has complicated and complicated pipelines, various and dispersed subsystems and difficult data communication, thereby influencing the coordination management of the whole equipment; the system has the characteristics of nonlinearity, complex and changeable structure, mutual coupling of a plurality of system parameters and the like, so that fault diagnosis is difficult. Meanwhile, air conditioner faults are often gradually exposed along with the aging of electronic components of equipment or the blockage of various pipelines, and when people find the faults, the faults are often serious, so that the feasibility and the practical significance are realized for timely detecting and diagnosing the air conditioner faults so as to avoid huge energy consumption and potential safety hazards caused by operation in a fault state.
In recent years, air conditioners are more complex and more intelligent, traditional diagnosis technologies far cannot meet the diagnosis requirements of modern air conditioners, and machine learning is gradually rising under the situation, wherein common diagnosis methods include a Neural Network (NN) and a Support Vector Machine (SVM). The neural network has the advantages of strong learning ability, nonlinear approximation ability and the like, but has the defects of difficult parameter optimization, slow convergence speed, easy overfitting and the like. The support vector machine is used as a more classical classification algorithm, the problems of low convergence speed and overfitting of a neural network do not exist, but the capability is insufficient when large sample data is processed, and the accuracy of solving the problem of multi-classification is low.
The random forest is one of more classical algorithms in ensemble learning, can solve the problems that a neural network is too slow in convergence speed and easy to fall into overfitting, and the like, and can also solve the defect that a support vector machine is insufficient in capability of processing large sample data. More importantly, the random forest can integrate various feature vectors, and the diagnosis accuracy is effectively improved. The extreme random tree is used as a variant algorithm of a random forest, and a characteristic threshold value is randomly selected when the node selects the characteristic, so that the prediction variance is smaller, and the generalization capability of the model is higher. Therefore, the invention provides a PCA-extreme random tree air conditioner fault diagnosis method based on Bayesian optimization, which is characterized in that feature dimension reduction is carried out through PCA, then an extreme random tree is used for constructing a diagnosis model, and global optimization is carried out on the range of feature quantity after feature dimension reduction of the PCA and the decision tree quantity of the extreme random tree through a Bayesian optimization algorithm, so that the model training time is reduced, and meanwhile, the fault detection and classification accuracy is optimized, thereby more quickly and accurately realizing the fault detection and classification of the air conditioner.
At present, the air conditioner fault diagnosis method based on the PCA-limit random tree optimized by the Bayesian optimization provided by the invention is not found in published documents and patents.
Disclosure of Invention
The invention aims to provide a PCA-extreme random tree air conditioner fault diagnosis method based on Bayesian optimization to overcome the defects of the prior art, so that fault detection and classification of an air conditioner can be realized more quickly and accurately.
The invention provides a Bayesian optimization-based PCA-extreme random tree air conditioner fault diagnosis method, which comprises the following steps of:
1) acquiring characteristic data Y ═ Y of air conditioner running in normal and different faults1,y2，…,yN]∈RN×MWherein N represents the number of data, M represents the number of features, and the features include the number of measured variables and manipulated variables; carrying out normalization processing on the data;
2) reducing the dimension of the features in the normalized data by using a PCA algorithm and then using the features as training samples;
3) establishing a limit random tree classification model, and training by using a training sample to obtain a PCA-limit random tree fault diagnosis model for the air conditioner;
4) and optimizing the feature quantity and the limit random tree quantity after the PCA dimension reduction of the PCA-limit random tree fault diagnosis model by using a Bayes optimization algorithm to obtain the optimal feature quantity and the limit random tree quantity after the dimension reduction.
5) And reducing the dimension of the normalized data again according to the obtained optimal PCA dimension-reduced characteristic quantity value to update the training sample, and training the updated training sample by taking the extreme random tree quantity value as a parameter of the extreme random tree fault diagnosis model to obtain the final PCA-extreme random tree fault diagnosis model.
6) And normalizing the air conditioner operation data collected in real time, reducing the dimension according to the obtained optimal PCA dimension-reduced characteristic quantity value, and inputting the reduced value to a final PCA-limit random tree fault diagnosis model for fault diagnosis.
Further, in the step 1), the data normalization process adopts a z-score normalization method.
Further, in the step 2), a specific process of performing dimension reduction on the features in the normalized data by using a PCA algorithm is as follows:
2.1) firstly carrying out zero equalization on the characteristics, namely, firstly solving the characteristic mean value and then subtracting the mean value;
2.2) calculating a covariance matrix;
2.3) solving the eigenvalue of the covariance matrix and the corresponding eigenvector, sorting the eigenvalues from large to small, selecting the first k eigenvectors, and then respectively taking the corresponding k eigenvectors as column vectors to form an eigenvector matrix;
and 2.4) reducing the dimension of the original data according to the feature vector matrix, and calculating new features to be used as training samples.
Further, the specific process of step 3) is as follows:
and 3.1) randomly selecting features from the training samples to obtain a plurality of training subsets, generating a plurality of corresponding CART decision trees, taking the data after dimensionality reduction as the input of each tree, and simultaneously randomly selecting the features from each decision tree to construct the decision tree.
3.2) training decision trees, each tree dividing features at each node of the trees according to a threshold value; the threshold is obtained through random selection, then the bifurcation value at the moment is calculated, all the features in the node are traversed, the bifurcation values of all the features are obtained according to the method, the feature corresponding to the minimum bifurcation value is selected as the classification feature of the node, and the PCA-extreme random tree model is obtained.
Further, in step 3.2), the minimum GINI value is used as the bifurcation value, and the GINI calculation formula is as follows:
Figure BDA0003047377140000031
wherein j denotes an index of a feature, IjNumber of classes, p, for feature jj，iIs the probability that the feature j is classified into the ith class.
Further, in the step 4), the objective function z of the bayesian optimization algorithm is an average accuracy obtained after five-fold cross validation, and the parameter to be optimized is the feature quantity x after the dimensionality reduction of the PCA algorithm1And number of CART decision trees of extreme random numbers x2Specifically, the following is shown:
Figure BDA0003047377140000032
wherein ω isiI is 1,2 is the coefficient of the ith Gaussian distribution, XiFor a plurality of ith independent variables xiA vector of parameters, E (X)i) Is a vector XiMathematical expectation, K (X)i，Xi) Is a vector XiThe autocovariance matrix of (2).
Further, in the step 6), each tree in the PCA-limit random tree fault diagnosis model classifies the data points collected in real time, and then each tree votes to determine the final classification of the data points.
The invention has the beneficial effects that: the extreme random tree is used as a classification algorithm, so that the training precision is improved under the condition of ensuring the network training speed, the overfitting problem is reduced, and the generalization capability of the model is enhanced. In addition, the use of PCA reduces the number of features in the model, allowing the speed of the model to be further increased. Meanwhile, the Bayesian optimization is used for optimizing the parameters of the model, so that the training time of the model is shortened, and the accuracy of fault detection and classification is further improved, thereby more quickly and accurately realizing the fault detection and classification of the air conditioner.
Drawings
Fig. 1 is a general flow chart of an air conditioner fault diagnosis method based on a PCA-limit random tree optimized by bayes provided by the invention.
FIG. 2 is a flow chart of optimization parameters of a Bayesian optimization algorithm for searching the dimensionality reduction quantities and the decision tree quantities of PCA in an embodiment of the present invention.
Detailed Description
In order to make the technical scheme and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. The specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention to these embodiments. In addition, the technical solutions according to the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
The technical scheme of the invention is specifically explained below with reference to the accompanying drawings.
The invention discloses a Bayesian optimization-based PCA-extreme random tree air conditioner fault diagnosis method, which has a general flow diagram as shown in figure 1 and specifically comprises the following steps:
1) the method comprises the steps of obtaining characteristic data of normal and different faults of the air conditioner during operation, wherein the faults comprise various faults of crude oil surplus, condenser water flow reduction, evaporator water flow reduction, refrigerant quantity lower than a nominal value, refrigerant quantity exceeding a nominal value and the like, each fault comprises a plurality of characteristic (measured variable and operation variable) data points during the operation of the air conditioner, and the air conditioning system is set to have M characteristics, such as temperature, pressure, flow, valve position, electric energy consumption of a compressor, working efficiency and the like. Each data point is an observation vector yn with dimension 1 × M, and assuming that the data packet has N data points, the observation matrix, i.e. the input data, is Y ═ Y1，y2，...，yN]∈RN×MAnd normalizing the characteristic data to obtain
Figure BDA0003047377140000044
Preferably, the normalization process for the fault data of the air conditioner may be performed using a z-score normalization method.
2) Reducing the dimension of the data obtained after the normalization processing through a PCA algorithm and then taking the data as the input of a limit random tree model; the PCA algorithm dimension reduction specific process is as follows:
2.1) firstly carrying out zero equalization on the characteristics, namely, firstly solving the characteristic mean value and then subtracting the mean value;
2.2) calculating a covariance matrix C, wherein the calculation formula is as follows:
Figure BDA0003047377140000041
2.3) solving the eigenvalue of the covariance matrix and the corresponding eigenvector, sorting the eigenvalues from large to small, selecting the largest k eigenvectors, and then taking the corresponding k eigenvectors as column vectors respectively to form an eigenvector matrix P epsilon RM×k；
2.4) reducing the dimension of the original data according to P and calculating a new observation matrix
Figure BDA0003047377140000045
As a training sample, the calculation formula is:
Figure BDA0003047377140000042
3) establishing a limit random tree classification model, training and testing the classifier to obtain a limit random tree fault diagnosis model for the air conditioner, and evaluating the performance of the model by adopting five-fold cross validation;
in this embodiment, a CART decision tree is used as the classification tree, the CART decision tree is optimized based on ID3, when the CART decision tree is used as the classification tree, a minimum GINI value is used as a basis for node splitting, that is, a bifurcation value, and a GINI calculation formula is as follows:
Figure BDA0003047377140000043
assume that the GINI of a certain feature j, j-1, 2jThen, first, a threshold is set to classify the feature as IjA class, wherein pj，iRefers to the probability that a feature j is classified into the ith class.
The specific process of the extreme random tree algorithm is as follows:
3.1) randomly selecting features from the training samples to obtain a plurality of training subsets, generating a plurality of corresponding CART decision trees, and observing an observation matrix
Figure BDA0003047377140000052
As input to each tree, while each decision tree randomly selects features to build the decision tree.
3.2) training decision trees, each tree dividing features at each node of the trees according to a threshold value; and (3) randomly selecting threshold values of the characteristic values in the extreme random tree, then calculating GINI at the moment, traversing all the characteristics in the node, obtaining the bifurcation values of all the characteristics, namely the GINI, according to the method, selecting the characteristic corresponding to the minimum bifurcation value as the classification characteristic of the node, and obtaining the PCA-extreme random tree fault diagnosis model.
3.3) five-fold cross-validation was used to evaluate the performance of the model.
In the five-fold cross validation in the step 3.3), the training sample is divided into 5 parts, 4 parts of the training sample are used as a training set each time, the rest 1 part of the training sample is used as a test set, the obtained average accuracy is used as an evaluation index, and the process is repeated for 5 times to ensure that each part of data is used as the test set. The process of testing and centrally diagnosing certain data is as follows: each tree classifies this data and votes on the trees to determine the final classification of the data point. In this step, the probability that a data point belongs to a certain class is calculated for each tree, then the probabilities obtained by all the trees are added, and finally the class with the highest probability is taken as the final classification.
4) And optimizing the feature quantity and the CART decision tree quantity after the PCA dimensionality reduction of the PCA-extreme random tree fault diagnosis model by using a Bayesian optimization algorithm to obtain the optimal feature quantity and the CART decision tree quantity after the dimensionality reduction.
Specifically, the specific flow of the bayesian optimization algorithm is shown in fig. 2, the bayesian objective function z in the invention is the average accuracy obtained after five-fold cross validation, and the parameter to be optimized is the feature quantity x after dimensionality reduction of the PCA algorithm1And number of CART decision trees of extreme random numbers x2The specific expression is as follows:
Figure BDA0003047377140000051
where f (x) is the objective function z, the parameter to be optimized, i.e. the argument x ═ x1，x2，ωiIs the coefficient of the ith Gaussian distribution, XiFor a plurality of i-th independent variables xiA vector of parameters, E (X)i) Is a vector XiMathematical expectation, K (X)i，Xi) Is a vector XiThe autocovariance matrix of (2). GP tableA gaussian process is shown.
In optimization, a plurality of initial parameter configuration parameter sets { X } are generated randomly1，X2Z }; then constructing a Gaussian mixture model by using the initial parameters; and selecting a new parameter point x by an AC Function (Acquisition Function)1′、x2'; then, calculating the average accuracy obtained after five-fold cross validation through a PCA-extreme random tree fault diagnosis model according to the parameter points, namely an objective function z', and then reconstructing a new parameter set by forming new parameters and initial parameters to select the parameter points by a Gaussian mixture model; since too many selections are time consuming, the selections are repeated 25 times in this embodiment.
5) And (4) carrying out dimensionality reduction on the normalized data again according to the obtained optimal PCA dimensionality reduced characteristic quantity value to update the training sample, and training the updated training sample by taking the extreme random tree quantity value as a parameter of the extreme random tree fault diagnosis model to obtain a final PCA-extreme random tree fault diagnosis model.
6) And normalizing the air conditioner operation data collected in real time, reducing the dimension according to the obtained optimal PCA dimension-reduced characteristic quantity value, and inputting the reduced value to a final PCA-limit random tree fault diagnosis model for fault diagnosis.
In this example, data was from the institute of heating, refrigeration and air Conditioning Engineers (ASHRAE) engineering package ASHRAE-D-16919- > 20101110, a 90 ton centrifugal chiller was selected in the laboratory to simulate 8 possible failure data packages for the chiller system, and data packages under normal conditions. The whole data packet provides 65 data, including 29 temperature data, 5 pressure data, 5 flow data, 7 valve position data, power consumption of the compressor, working efficiency and the like, which is a relatively detailed diagnosis data of the air conditioner fault. And carrying out normalization processing on each type of feature data.
In this embodiment, the empirical selection method is used to select the reduced-dimension feature quantity of PCA as 20, the CART decision tree quantity of the extreme random tree is 500, and the obtained average accuracy is 97.95%. In the same case, the accuracy of the BP nerve is 97.56%, the accuracy of the CNN neural network is 96.58%, and the effect of the extreme random tree is greatly improved compared with the prior art.
After Bayes parameter optimization, when the feature quantity after dimensionality reduction of PCA is 36 and the CART decision tree quantity of the extreme random tree is 738, the obtained average accuracy is 98.38%, which is obviously improved compared with the accuracy of the unoptimized algorithm. If training speed is pursued, Bayesian optimization is used to obtain that when the feature quantity after dimensionality reduction of PCA is 31 and the CART decision tree quantity of the extreme random tree is 146, the average accuracy is 98.28%, and at this time, the number of decision trees is greatly reduced under the condition of slightly sacrificing accuracy. According to actual measurement, when the feature quantity of the PCA after dimensionality reduction is 20 and the CART decision tree quantity of the extreme random tree is 500, the training speed is 2min 38 s; when the number of features after dimensionality reduction of PCA is 31, and the number of CART decision trees of the extreme random tree is 146, the training speed is 59.8 s. Compared with an unoptimized algorithm, the method has the advantages that the accuracy is improved, and meanwhile, the speed is greatly increased, so that the fault detection and classification of the air conditioner are realized more quickly and accurately.
The above-described embodiment is merely a preferable embodiment of the present invention, and does not limit the scope of the present invention, and modifications, equivalent substitutions, improvements, and the like may be made without departing from the scope of the present invention as set forth in the claims.

Claims (7)

1. A PCA-limit random tree air conditioner fault diagnosis method based on Bayesian optimization is characterized by comprising the following steps:
1) acquiring characteristic data Y ═ Y of air conditioner running in normal and different faults1,y2,…,yN]∈RN×MWherein N represents the number of data, M represents the number of features, and the features include the number of measured variables and manipulated variables; carrying out normalization processing on the data;
2) reducing the dimension of the features in the normalized data by using a PCA algorithm and then using the features as training samples;
3) establishing a limit random tree classification model, and training by using a training sample to obtain a PCA-limit random tree fault diagnosis model for the air conditioner;
4) optimizing the feature quantity and the limit random tree quantity after PCA dimension reduction of the PCA-limit random tree fault diagnosis model by using a Bayes optimization algorithm to obtain the optimal feature quantity and the limit random tree quantity after dimension reduction;
5) carrying out dimensionality reduction on the normalized data again according to the obtained optimal PCA dimensionality reduced characteristic quantity value to update a training sample, and training the updated training sample by taking the extreme random tree quantity value as a parameter of the extreme random tree fault diagnosis model to obtain a final PCA-extreme random tree fault diagnosis model;
6) and normalizing the air conditioner operation data collected in real time, reducing the dimension according to the obtained optimal PCA dimension-reduced characteristic quantity value, and inputting the reduced value to a final PCA-limit random tree fault diagnosis model for fault diagnosis.
2. The air conditioner fault diagnosis method based on the PCA-limit random tree optimized by the Bayes as recited in claim 1, wherein: in the step 1), a z-score standardization method is adopted in data normalization processing.
3. The air conditioner fault diagnosis method based on the PCA-limit random tree optimized by the Bayes as recited in claim 1, wherein: in the step 2), the specific process of using the PCA algorithm to reduce the dimension of the features in the normalized data is as follows:
2.1) firstly carrying out zero equalization on the characteristics, namely, firstly solving the characteristic mean value and then subtracting the mean value;
2.2) calculating a covariance matrix;
2.3) solving the eigenvalue of the covariance matrix and the corresponding eigenvector, sorting the eigenvalues from large to small, selecting the first k eigenvectors, and then respectively taking the corresponding k eigenvectors as column vectors to form an eigenvector matrix;
and 2.4) reducing the dimension of the original data according to the feature vector matrix, and calculating new features to be used as training samples.
4. The air conditioner fault diagnosis method based on the PCA-limit random tree optimized by the Bayes as recited in claim 1, wherein: the specific process of the step 3) is as follows:
3.1) randomly selecting features from the training samples to obtain a plurality of training subsets, generating a plurality of corresponding CART decision trees, taking the data after dimensionality reduction as the input of each tree, and simultaneously randomly selecting features from each decision tree to construct a decision tree;
3.2) training decision trees, each tree dividing features at each node of the trees according to a threshold value; the threshold is obtained through random selection, then the bifurcation value at the moment is calculated, all the features in the node are traversed, the bifurcation values of all the features are obtained according to the method, the feature corresponding to the minimum bifurcation value is selected as the classification feature of the node, and the PCA-extreme random tree model is obtained.
5. The air conditioner fault diagnosis method based on the PCA-limit random tree optimized by the Bayes as recited in claim 4, wherein: in the step 3.2), the minimum GINI value is used as the bifurcation value, and the GINI calculation formula is as follows:
Figure FDA0003047377130000021
wherein j denotes an index of a feature, IjNumber of classes, p, for feature jj,iIs the probability that the feature j is classified into the ith class.
6. The air conditioner fault diagnosis method based on the PCA-limit random tree optimized by the Bayes as recited in claim 1, wherein: in the step 4), the objective function z of the Bayesian optimization algorithm is the average accuracy obtained after five-fold cross validation, and the parameter to be optimized is the feature quantity x after dimensionality reduction of the PCA algorithm1And number of CART decision trees of extreme random numbers x2Specifically, the following are shown:
Figure FDA0003047377130000022
wherein omegaiI is 1,2 is the coefficient of the ith Gaussian distribution, XiFor a plurality of ith independent variables xiA vector of parameters, E (X)i) Is a vector XiMathematical expectation, K (X)i,Xi) Is a vector XiThe autocovariance matrix of (2).
7. The air conditioner fault diagnosis method based on the PCA-limit random tree optimized by the Bayes as recited in claim 1, wherein: in the step 6), each tree in the PCA-extreme random tree fault diagnosis model classifies the data points collected in real time, and then each tree votes to determine the final classification of the data points.
CN202110475880.8A 2021-04-29 2021-04-29 Air conditioner fault diagnosis method based on Bayesian optimization PCA-extreme random tree Active CN113177594B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110475880.8A CN113177594B (en) 2021-04-29 2021-04-29 Air conditioner fault diagnosis method based on Bayesian optimization PCA-extreme random tree

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110475880.8A CN113177594B (en) 2021-04-29 2021-04-29 Air conditioner fault diagnosis method based on Bayesian optimization PCA-extreme random tree

Publications (2)

Publication Number Publication Date
CN113177594A CN113177594A (en) 2021-07-27
CN113177594B true CN113177594B (en) 2022-06-17

Family

ID=76925403

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110475880.8A Active CN113177594B (en) 2021-04-29 2021-04-29 Air conditioner fault diagnosis method based on Bayesian optimization PCA-extreme random tree

Country Status (1)

Country Link
CN (1) CN113177594B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113792762B (en) * 2021-08-24 2023-06-20 华南理工大学 Water chilling unit fault diagnosis method, system and medium based on Bayesian optimization LightGBM
CN113673623B (en) * 2021-08-31 2023-10-20 重庆大学 Bridge damage identification method based on improved principal component analysis algorithm
CN113780405B (en) * 2021-09-07 2023-07-25 科希曼电器有限公司 Air conditioner parameter regression optimization method based on deep neural network
CN115095953B (en) * 2022-06-16 2024-04-09 青岛海信日立空调系统有限公司 Training method and device for fault diagnosis model

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2849099B1 (en) * 2013-09-11 2021-07-28 Dassault Systèmes A computer-implemented method for designing an industrial product modeled with a binary tree.
CN106991432A (en) * 2017-03-07 2017-07-28 南京信息工程大学 The feature selection approach of forest optimized algorithm based on scoring
CN110175195B (en) * 2019-04-23 2022-11-29 哈尔滨工业大学 Mixed gas detection model construction method based on extreme random tree
CN112327219B (en) * 2020-10-29 2024-03-12 国网福建省电力有限公司南平供电公司 Distribution transformer fault diagnosis method with automatic feature mining and parameter automatic optimizing functions

Also Published As

Publication number Publication date
CN113177594A (en) 2021-07-27

Similar Documents

Publication Publication Date Title
CN113177594B (en) Air conditioner fault diagnosis method based on Bayesian optimization PCA-extreme random tree
Liu et al. Data-driven and association rule mining-based fault diagnosis and action mechanism analysis for building chillers
CN111832647A (en) Abnormal flow detection system and method
CN110929847A (en) Converter transformer fault diagnosis method based on deep convolutional neural network
CN110674120A (en) Wind power plant data cleaning method and device
JP7078873B2 (en) Machine learning device, demand control system, and air conditioning control system
Li et al. Diagnosis for multiple faults of chiller using ELM-KNN model enhanced by multi-label learning and specific feature combinations
CN112327168A (en) XGboost-based electric vehicle battery consumption prediction method
CN111237988A (en) Control method and system for subway vehicle-mounted air conditioning unit
CN112884012A (en) Building energy consumption prediction method based on support vector machine principle
CN111695288B (en) Transformer fault diagnosis method based on Apriori-BP algorithm
Li et al. Machine learning based diagnosis strategy for refrigerant charge amount malfunction of variable refrigerant flow system
Aqlan et al. Integrating artificial neural networks and cluster analysis to assess energy efficiency of buildings
CN110084301B (en) Hidden Markov model-based multi-working-condition process working condition identification method
CN111242204A (en) Operation and maintenance management and control platform fault feature extraction method
Ye et al. Multi-classification decision-making method for interval-valued intuitionistic fuzzy three-way decisions and its application in the group decision-making
CN114384423A (en) Fuel cell health state identification method based on typical operation parameters
CN113469252A (en) Extra-high voltage converter valve operation state evaluation method considering unbalanced samples
CN113177577A (en) Bearing fault diagnosis method based on improved convolutional neural network
CN110427019B (en) Industrial process fault classification method and control device based on multivariate discriminant analysis
CN113450562A (en) Road network traffic state discrimination method based on clustering and graph convolution network
CN112700162A (en) Method and device for evaluating running state of rail transit air conditioner
CN113523904A (en) Cutter wear detection method
CN115730228A (en) Central air conditioner energy consumption analysis method based on BP neural network
CN114357412A (en) Efficiency self-control optimizing control system of heating ventilation air conditioning system

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