CN114941890A - Central air conditioner fault diagnosis method and system based on image and depth blurring - Google Patents
Central air conditioner fault diagnosis method and system based on image and depth blurring Download PDFInfo
- Publication number
- CN114941890A CN114941890A CN202210571075.XA CN202210571075A CN114941890A CN 114941890 A CN114941890 A CN 114941890A CN 202210571075 A CN202210571075 A CN 202210571075A CN 114941890 A CN114941890 A CN 114941890A
- Authority
- CN
- China
- Prior art keywords
- data
- image
- fault diagnosis
- layer
- central air
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000003745 diagnosis Methods 0.000 title claims abstract description 52
- 238000006243 chemical reaction Methods 0.000 claims abstract description 21
- 238000000605 extraction Methods 0.000 claims abstract description 15
- 238000012545 processing Methods 0.000 claims abstract description 14
- 238000004378 air conditioning Methods 0.000 claims description 10
- 238000013507 mapping Methods 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 7
- 238000005457 optimization Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- 238000002372 labelling Methods 0.000 claims description 4
- 238000012163 sequencing technique Methods 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 108091034117 Oligonucleotide Proteins 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 5
- 230000004075 alteration Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Classifications
-
- 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
-
- 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
- F24F11/50—Control or safety arrangements characterised by user interfaces or communication
- F24F11/54—Control or safety arrangements characterised by user interfaces or communication using one central controller connected to several sub-controllers
-
- 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
- F24F11/50—Control or safety arrangements characterised by user interfaces or communication
- F24F11/61—Control or safety arrangements characterised by user interfaces or communication using timers
-
- 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
- F24F11/62—Control or safety arrangements characterised by the type of control or by internal processing, e.g. using fuzzy logic, adaptive control or estimation of values
- F24F11/63—Electronic processing
- F24F11/64—Electronic processing using pre-stored data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
- G06N5/048—Fuzzy inferencing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/28—Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
- G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Chemical & Material Sciences (AREA)
- Combustion & Propulsion (AREA)
- Software Systems (AREA)
- Mechanical Engineering (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Signal Processing (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Fuzzy Systems (AREA)
- Human Computer Interaction (AREA)
- Probability & Statistics with Applications (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Databases & Information Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Automation & Control Theory (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention provides a central air conditioner fault diagnosis method and system based on image and depth blurring, wherein the system mainly comprises four modules: the system comprises a data processing module, a digital-image conversion module, a residual error depth fuzzy module and a fault diagnosis module; the data processing module performs feature extraction and feature sorting on the data set; the digital image conversion module converts the data set after the characteristic extraction into a corresponding two-dimensional gray image and generates a rich image data set by using a sliding window method; the residual depth blurring module trains a residual depth blurring model for fault diagnosis by using a two-dimensional grayscale image data set. The real-time fault diagnosis module of the central air conditioner can acquire data through a plurality of sensors and input the data into the fault diagnosis model to judge whether faults exist or not. The fault diagnosis method adopts a strategy of combining a kernel slow characteristic analysis algorithm, a digital-to-image conversion algorithm and a residual error depth fuzzy model to construct a fault diagnosis model of the central air conditioner, and can efficiently and accurately diagnose faults.
Description
Technical Field
The invention belongs to the field of air conditioner fault diagnosis, and particularly relates to a central air conditioner fault diagnosis method and system based on image and depth blurring.
Background
The central air conditioner can provide a comfortable indoor environment by adjusting air temperature and humidity of different rooms and areas. Due to the fact that the number of components is large, the operation environment is complex and changeable, the operation is not proper or the components are corroded by natural factors, and various faults of the central air conditioner can occur. The long-time operation of the air conditioner in a fault state not only shortens the service life of the equipment, but also causes serious energy waste. Therefore, it is necessary to find an efficient and accurate method and system for diagnosing faults of a central air conditioner.
The central air conditioner presents various operation modes along with the change of the external environment, has extremely strong time dynamic characteristics and nonlinear characteristics, and presents spatial correlation characteristics by the correlation among all characteristic variables. The existing central air conditioner fault diagnosis method generally does not consider the characteristics, and has the following problems: (1) an unprocessed original data set is used when a fault diagnosis model is established, and the characteristic variables are complex and the fault characteristics are not obvious; (2) the method mainly utilizes numerical data to establish a model, and destroys the spatial correlation characteristics among all characteristic variables; (3) the fault diagnosis model is complex in structure and has no interpretability.
Disclosure of Invention
In order to solve the problems in the prior art, a central air conditioner fault diagnosis method and system based on image and depth blurring are provided.
The technical scheme adopted by the invention for solving the technical problems is as follows:
the technical scheme provides a central air conditioner fault diagnosis method based on image and depth blurring, which comprises the following steps:
step (1): data processing
Carrying out data acquisition and marking on data of normal operation and various fault operations of the central air conditioner; preprocessing the acquired data; establishing a kernel slow characteristic analysis model for carrying out characteristic extraction and characteristic sequencing on data to obtain a kernel slow characteristic data set;
step (2): number map conversion
Converting the kernel slow characteristic data set after characteristic extraction into a corresponding two-dimensional gray image, and generating a rich image data set by using a sliding window method;
and (3): building and training residual depth fuzzy model
Establishing a residual depth fuzzy model; training a residual error depth fuzzy model for fault diagnosis through the acquired two-dimensional gray image data set;
and (4): central air-conditioning fault diagnosis
Measuring and collecting data when the central air conditioner operates to obtain newly collected data, executing the first step and the second step on the newly collected data, inputting the processed data into the established residual error depth fuzzy model, and judging whether the fault exists or not.
In the step (1), the specific method for data acquisition and labeling is as follows: by air-conditioning a plurality of central air-conditionersA sensor is arranged at the position to collect data of normal operation and various fault operations; data acquisition is carried out by taking one day as a unit, and sampling time intervals are one minute; representing the acquired raw data set as
It contains data for N days, and there are T time samples and P feature variables for each day of data. Thus, data for a day may be represented as
Wherein T is an element of [1,2]The time range is represented by the time range,
the value corresponding to the p-th characteristic variable in the continuous T time sequences; and after the data acquisition is finished, marking the data with a corresponding operating condition label.
In the step (1), the specific method for data preprocessing is as follows:
the raw data set X was normalized per day using the Z-score algorithm as follows:
wherein mean (x) p (t)) is x p Mean value of (t), std (x) p (t)) is x p (t) standard deviation;
defining the data normalized by a certain day as
Wherein
Therefore, the data of each day are standardized to obtain a data set
In the step (1), the specific method for extracting the features comprises the following steps: determining optimal parameters by using the standardized normal operation data, and establishing a kernel slow characteristic analysis model; first, the data is mapped using an implicit nonlinear mapping function φ (-) to
Mapping to high dimensionality
The slow feature is then obtained by solving the following optimization problem:
wherein the output data y j (t) is x from the input data φ (t) the jth slow feature extracted,
is y j (t) the first derivative with respect to the time variable t,<.>is defined as
K is a sampling interval;
further, the data after feature extraction is represented as
Extracting slowly-changing features from all fault data sets based on the trained kernel slow feature analysis model to obtain N-day kernel slow feature data sets which are recorded as
In the step (2), the specific method for converting the numerical map comprises the following steps:
carrying out Min-Max normalization processing on data of each day in a kernel slow feature data set Y after feature extraction, and multiplying an output result by 255, wherein the formula is as follows:
wherein, min (y) η (t)) and max (y) η (t)) are each y η (t) (η ═ 1, 2.., R) minimum and maximum values of the vector, and the acquired N-day dataset was recorded as
Converting the data set Z obtained in the step into a corresponding two-dimensional gray image; arranging the characteristic variables according to the slow change degree of the characteristic variables in the image; specifically, the characteristic variable which changes slowest is converted into a pixel in a first column of the image, then the characteristic variable which changes second slowest is converted into a pixel in a second column of the image, and so on, so that a converted two-dimensional gray image is obtained; wherein the conversion of data to an image is performed by converting the input data to an unsigned 8 integer type, followed by using a data format conversion instruction; the larger the numerical value is, the deeper the gray scale is;
expanding the generated image data set by using a sliding window method; setting the lag parameter to L, the first image is generated by data of the M line to the (M + L-1) line; if the sliding distance of the sliding window is set as q, intercepting from the (M + q) line data to the (M + L + q-1) line data, and generating a second image by analogy; if the data in a day has R characteristic variables and T samples, repeating the above operation can obtain (T-L)/q +1 images, and the size of each image is L multiplied by R.
The residual depth fuzzy model in the step (3) comprises an input layer, a hidden layer and an output layer; the model is realized by stacking fuzzy reasoning modules from bottom to top layer by layer; the detailed structure and principle of the model are as follows:
1) the input layer preliminarily acquires the input image information of each fault type and transmits the information to the hidden layer through the nodes of the equivalent mapping;
2) in the hidden layer, the fuzzy inference modules are stacked layer by layer, and have s layers in total; the 1 st hidden layer is provided with s fuzzy inference modules, the 2 nd hidden layer is provided with (s-1) fuzzy inference modules, and by analogy, the s th hidden layer is provided with only one fuzzy inference module;
3) in the hidden layer, the output variable y of all fuzzy inference modules of the previous layer l And the residual error of the expected result is weighted and then is used as the input quantity of the fuzzy inference module of the next layer, and the output of the ith layer is as follows:
where l is 1,2, …, s, epsilon l-1 The output of the proxy layer is then performed,
in order to output the vector, the vector is,
represents a weight vector;
4) in the hidden layer, based on the image information of each fault type, a fuzzy C-means clustering algorithm is adopted to obtain a corresponding initial fuzzy rule, and the problem of regular optimization is solved
Obtaining the weight vector value w of the fuzzy rule in each fuzzy inference module l ＝[λ 1 I+(y l ) T (y l )] -1 (y l ) T z l Where I is the identity matrix, λ 1 Represents the regularization coefficient, z l ＝(z 1 ,z 2 ,…,z l ) For the desired output residual vector, z 1 Y is the desired output, z when l 2, …, s l ＝z l-1 -ε l-2 ；
5) In the output layer, the output layer is connected with the hidden layer, and the output result is obtained by accumulating all intermediate weighted values layer by layer
Wherein
Representing the type of failure of the final output.
The invention also provides a central air conditioner fault diagnosis system based on image and depth blurring, which comprises the following steps:
a data processing module for performing the method of step (1);
a map conversion module for performing the method of step (2);
a residual depth blurring module for performing the method of step (3);
and (3) a fault diagnosis module for executing the method of the step (4).
Compared with the prior art, the invention has the following advantages:
1. the method uses a kernel slow characteristic analysis algorithm to extract slowly changing characteristics from dynamic air conditioner operation data, and sorts characteristic variables according to the slowly changing degree of the characteristics, so that fault characteristics are enhanced;
2. the data after the characteristic enhancement is converted into the image by the digital image conversion method, and the neighborhood information and the spatial correlation characteristic among characteristic variables are fully mined;
3. the sliding window method can generate abundant image data sets, and provides necessary conditions for accurately establishing a fault diagnosis model;
4. the residual depth fuzzy model can quickly and accurately identify images for fault diagnosis, and the constructed fault diagnosis model is made to be interpretable.
Drawings
The above and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:
FIG. 1 is a block diagram of the present invention.
Fig. 2 is a schematic diagram of a two-dimensional grayscale image structure converted from numerical data.
FIG. 3 is a schematic diagram of an example structure for use with the sliding window approach.
Fig. 4 is a schematic structural diagram of the residual depth blur model.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention.
As shown in fig. 1 to 4, in order to effectively solve the problems in the background art, the fault diagnosis model of the central air conditioner is constructed by adopting a strategy of combining a kernel slow feature analysis algorithm, a number-map conversion algorithm and a residual depth fuzzy model.
The kernel slow feature analysis algorithm can derive slowly varying output variables from the input data and rank the feature variables according to their slowly varying degrees. The number map conversion method can convert the obtained numerical data into a corresponding two-dimensional gray image so as to mine neighborhood information and spatial correlation of the characteristic variables. The residual error depth fuzzy model has better interpretable characteristics, and the output judgment of the fault diagnosis result can be relied on.
In view of the above situation, the invention provides a central air conditioner fault diagnosis method and system based on image and depth blurring. The system mainly comprises four modules: the device comprises a data processing module, a digital-image conversion module, a residual error depth fuzzy module and a fault diagnosis module. The data processing module is used for extracting and sequencing features of the data set; then, a data set after the characteristic extraction is converted into a corresponding two-dimensional gray image by a digital image conversion module, and a rich image data set is generated by using a sliding window method; the residual depth blurring module trains a residual depth blurring model for fault diagnosis by using a two-dimensional grayscale image data set. The real-time fault diagnosis module of the central air conditioner can acquire data through a plurality of sensors and input the data into the fault diagnosis model to judge whether faults exist or not.
The embodiment provides a central air conditioner fault diagnosis method based on image and depth blurring, which comprises the following steps:
step (1): data processing
Carrying out data acquisition and labeling on data of normal operation and various fault operations of the central air conditioner; preprocessing the acquired data; establishing a kernel slow characteristic analysis model for carrying out characteristic extraction and characteristic sequencing on data to obtain a kernel slow characteristic data set;
step (2): number map conversion
Converting the kernel slow characteristic data set after characteristic extraction into a corresponding two-dimensional gray image, and generating a rich image data set by using a sliding window method;
and (3): building and training residual error depth fuzzy model
Establishing a residual depth fuzzy model; training a residual error depth fuzzy model for fault diagnosis through the acquired two-dimensional gray image data set;
and (4): central air-conditioning fault diagnosis
Measuring and collecting data when the central air conditioner operates to obtain newly collected data, executing the first step and the second step on the newly collected data, inputting the processed data into the established residual error depth fuzzy model, and judging whether the fault exists or not.
As explained in further detail below:
1. data processing module
The module is used for executing the method of the step (1); the module is used for extracting the characteristics and arranging the characteristics of data by establishing a kernel slow characteristic analysis model so as to solve the problems of time dynamic characteristics and nonlinearity of the central air conditioner. The module comprises the following specific steps:
1.1 in the step (1), the specific method for data acquisition and labeling is as follows: the method comprises the following steps of (1) collecting data of normal operation and various fault operations by installing sensors at a plurality of positions of the central air conditioner (installing sensors at core positions of a return air pipeline, a fresh air valve, an air return valve, a return fan and the like of the central air conditioner); data acquisition is carried out by taking one day as a unit, and the sampling time interval is one minute; representing the acquired raw data set as
It contains data for N days, and there are T time samples and P characteristic variables in the data for each day. Thus, data for a day may be represented as
Wherein T is an element of [1,2]The time range is represented by a time range,
the value corresponding to the p-th characteristic variable in the continuous T time sequences; and after the data acquisition is finished, marking the data with a corresponding operating condition label.
1.2 in the step (1), the specific method for preprocessing the data comprises the following steps:
the raw data set X was normalized per day using the Z-score algorithm as follows:
wherein mean (x) p (t)) is x p Mean value of (t), std (x) p (t)) is x p (t) standard deviation;
defining the data normalized by a certain day as
Wherein
Therefore, the data of each day is standardized to obtain a data set
1.3 in the step (1), the specific method for extracting the features comprises the following steps: determining optimal parameters by using the standardized normal operation data, and establishing a kernel slow characteristic analysis model; first, the data is mapped using an implicit nonlinear mapping function φ (-) to
Mapping to high dimensionality
The slow feature is then obtained by solving the following optimization problem:
wherein the output data y j (t) is x from the input data φ (t) the jth slow feature extracted,
is y j (t) a first derivative of the time variable t,<.>is defined as
K is a sampling interval;
further, the data after feature extraction is represented as
Extracting slowly-changing features from all fault data sets based on the trained kernel slow feature analysis model to obtain N-day kernel slow feature data sets which are recorded as
2. Digital-image conversion module
The module is used for executing the method of the step (2); the module is used for converting numerical data into corresponding two-dimensional gray images by taking each day as a unit according to a data set provided by the data processing module, so that neighborhood information and spatial correlation characteristics among characteristic variables are enhanced. The module comprises the following steps:
2.1 in the step (2), the specific method for converting the numerical diagram is as follows:
carrying out Min-Max normalization processing on data of each day in a kernel slow feature data set Y after feature extraction, and multiplying an output result by 255, wherein the formula is as follows:
wherein, min (y) η (t)) and max (y) η (t)) are each y η (t) (η ═ 1, 2.., R) minimum and maximum values of the vector, and the acquired N-day dataset was recorded as
Converting the data set Z obtained in the step into a corresponding two-dimensional gray image; arranging the feature variables according to the slow change degree of the feature variables in the image; specifically, the feature variable which changes slowest is converted into a pixel in a first column of the image, then the feature variable which changes second slowest is converted into a pixel in a second column of the image, and so on, to obtain a converted two-dimensional gray image, and fig. 2 is an example of the converted two-dimensional gray image; wherein the conversion of data to an image is performed by converting the input data to an unsigned 8 integer type, followed by using a data format conversion instruction; the larger the numerical value is, the deeper the gray scale is;
expanding the generated image data set by using a sliding window method; setting the lag parameter to L, the first image is generated from the data of the M-th line to the (M + L-1) line; if the sliding distance of the sliding window is set as q, intercepting from the (M + q) line data to the (M + L + q-1) line data, and generating a second image by analogy; if the data in a day has R feature variables and T samples, repeating the above operation can obtain (T-L)/q +1 images, the size of which is L × R, and fig. 3 is an example of the sliding window method.
3. Residual depth blurring module
The module is used for executing the method of the step (3); the module is used for training a residual error depth fuzzy model for fault diagnosis through the acquired two-dimensional gray image data set, and the model structure is shown in FIG. 4.
The residual depth fuzzy model in the step (3) comprises an input layer, a hidden layer and an output layer; the model is realized by stacking fuzzy inference modules from bottom to top layer by layer; the detailed structure and principle of the model are as follows:
1) the input layer initially acquires the input image information of each fault type and transmits the information to the next layer (hidden layer) through the nodes of equivalent mapping;
2) in the hidden layer, the fuzzy inference modules are stacked layer by layer, and the hidden layer has s layers; the 1 st hidden layer is provided with s fuzzy inference modules, the 2 nd hidden layer is provided with (s-1) fuzzy inference modules, and the analogy is repeated, and only one fuzzy inference module is arranged in the s th hidden layer;
3) in the hidden layer, the output variable y of all fuzzy inference modules of the previous layer l And the residual error of the expected result is weighted and then is used as the input quantity of the fuzzy inference module of the next layer, and the output of the ith layer is as follows:
where l is 1,2, …, s, epsilon l-1 The output of the proxy layer is then performed,
in order to output the vector, the vector is,
4) in the hidden layer, based on the image information of each fault type, a fuzzy C-means clustering algorithm is adopted to obtain a corresponding initial fuzzy rule, and the problem of regular optimization is solved
Obtaining the weight vector value w of the fuzzy rule in each fuzzy inference module l ＝[λ 1 I+(y l ) T (y l )] -1 (y l ) T z l Where I is the identity matrix, λ 1 Represents the regularization coefficient, z l ＝(z 1 ,z 2 ,…,z l ) For the desired output residual vector, z 1 Y is the desired output, z when l 2, …, s l ＝z l-1 -ε l-2 ；
5) In the output layer, the output layer is connected with the hidden layer, and the output result is obtained by accumulating all intermediate weighted values layer by layer
Wherein
Representing the type of failure of the final output.
4. Fault diagnosis module
The module is used for executing the method of the step (4); the module is used for fault diagnosis of the central air conditioner. And measuring and collecting data during the operation of the air conditioner through a plurality of sensors, and inputting the data into the established fault diagnosis model. First, data is converted into a two-dimensional grayscale image, and then fault diagnosis is performed using a residual depth blur model.
From the above examples, it can be seen that: (1) the method uses a kernel slow characteristic analysis algorithm to extract slowly changing characteristics from dynamic air conditioner operation data, and sorts characteristic variables according to the slowly changing degree of the characteristics, so that fault characteristics are enhanced; (2) the data after the characteristic enhancement is converted into the image by the digital image conversion method, and the neighborhood information and the spatial correlation characteristic among characteristic variables are fully mined; (3) the sliding window method can generate abundant image data sets, and provides necessary conditions for accurately establishing a fault diagnosis model; (4) the residual depth fuzzy model can quickly and accurately identify the image for fault diagnosis, and the constructed fault diagnosis model has interpretability.
While embodiments of the invention have been shown and described, it will be understood by those of ordinary skill in the art that: various changes, modifications, substitutions and alterations can be made to the embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.
Claims (7)
1. A central air conditioner fault diagnosis method based on image and depth blurring is characterized by comprising the following steps:
step (1): data processing
Carrying out data acquisition and marking on data of normal operation and various fault operations of the central air conditioner; preprocessing the acquired data; establishing a kernel slow characteristic analysis model for carrying out characteristic extraction and characteristic sequencing on data to obtain a kernel slow characteristic data set;
step (2): number-to-picture conversion
Converting the kernel slow characteristic data set after characteristic extraction into a corresponding two-dimensional gray image, and generating a rich image data set by using a sliding window method;
and (3): building and training residual depth fuzzy model
Establishing a residual depth fuzzy model; training a residual error depth fuzzy model for fault diagnosis through the acquired two-dimensional gray image data set;
and (4): central air conditioner fault diagnosis
Measuring and collecting data when the central air conditioner operates to obtain newly collected data, executing the first step and the second step on the newly collected data, inputting the processed data into the established residual error depth fuzzy model, and judging whether the fault exists or not.
2. The central air-conditioning fault diagnosis method based on image and depth blurring as claimed in claim 1, wherein in the step (1), the specific method for data acquisition and labeling is as follows: the method comprises the steps that sensors are arranged at a plurality of positions of a central air conditioner, and data of normal operation and various fault operations are collected; data acquisition is carried out by taking one day as a unit, and sampling time intervals are one minute; representing the acquired raw data set as
It contains N days of data, and each day of data has T time samples and PAnd (4) characteristic variables. Thus, data for a day may be represented as
Wherein T is an element of [1,2]The time range is represented by a time range,
the value corresponding to the p-th characteristic variable in the continuous T time sequences; and after the data acquisition is finished, marking the data with a corresponding operating condition label.
3. The central air-conditioning fault diagnosis method based on image and depth blurring as claimed in claim 1, wherein in the step (1), a specific method for data preprocessing is as follows:
the raw data set X was normalized per day using the Z-score algorithm as follows:
wherein mean (x) p (t)) is x p Mean value of (t), std (x) p (t)) is x p (t) standard deviation;
defining the data normalized by a certain day as
Wherein
Therefore, the data of each day are standardized to obtain a data set
4. The central air-conditioning fault diagnosis method based on image and depth blurring as claimed in claim 1The method is characterized in that in the step (1), the specific method for extracting the features comprises the following steps: determining optimal parameters by using the standardized normal operation data, and establishing a kernel slow characteristic analysis model; first, the data is mapped using an implicit nonlinear mapping function φ (-) to
Mapping to high dimensionality
The slow feature is then obtained by solving the following optimization problem:
wherein the output data y j (t) is x from the input data φ (t) the jth slow feature extracted,
is y j (t) a first derivative of the time variable t,<.>is defined as
K is a sampling interval;
further, the data after feature extraction is represented as
On the basis of the trained kernel slow characteristic analysis model, slowly-changing characteristics are extracted from all fault data sets to obtain N-day kernel slow characteristic data sets which are recorded as
5. The central air-conditioning fault diagnosis method based on image and depth blurring as claimed in claim 1, wherein in the step (2), a specific method for converting the number map is as follows:
carrying out Min-Max normalization processing on data of each day in a kernel slow feature data set Y after feature extraction, and multiplying an output result by 255, wherein the formula is as follows:
wherein, min (y) η (t)) and max (y) η (t)) are each y η (t) (η ═ 1, 2.., R) minimum and maximum values of the vector, and the acquired N-day dataset was recorded as
Converting the data set Z obtained in the step into a corresponding two-dimensional gray image; arranging the feature variables according to the slow change degree of the feature variables in the image; specifically, the characteristic variable which changes slowest is converted into a pixel in a first column of the image, then the characteristic variable which changes second slowest is converted into a pixel in a second column of the image, and so on, so that a converted two-dimensional gray image is obtained; wherein the conversion of data to an image is performed by converting the input data to an unsigned 8 integer type, followed by using a data format conversion instruction; the larger the numerical value is, the deeper the gray scale is;
expanding the generated image data set by using a sliding window method; setting the lag parameter to L, the first image is generated by data of the M line to the (M + L-1) line; if the sliding distance of the sliding window is set as q, intercepting is started from the (M + q) row of data, and a second image is generated after the (M + L + q-1) row of data is finished, and so on; if the data in a day has R characteristic variables and T samples, repeating the above operation can obtain (T-L)/q +1 images, and the size of each image is L multiplied by R.
6. The central air-conditioning fault diagnosis method based on image and depth blur of claim 1, characterized in that the residual depth blur model in step (3) comprises an input layer, a hidden layer and an output layer; the model is realized by stacking fuzzy inference modules from bottom to top layer by layer; the detailed structure and principle of the model are as follows:
1) the input layer preliminarily acquires the input image information of each fault type and transmits the information to the hidden layer through the nodes of the equivalent mapping;
2) in the hidden layer, the fuzzy inference modules are stacked layer by layer, and have s layers in total; the 1 st hidden layer is provided with s fuzzy inference modules, the 2 nd hidden layer is provided with (s-1) fuzzy inference modules, and by analogy, the s th hidden layer is provided with only one fuzzy inference module;
3) in the hidden layer, the output variable y of all fuzzy inference modules of the previous layer l And the residual error of the expected result is weighted and then is used as the input quantity of the fuzzy inference module of the next layer, and the output of the ith layer is as follows:
where l is 1,2, …, s, epsilon l-1 The output of the proxy layer is then performed,
in order to output the vector, the vector is output,
represents a weight vector;
4) in the hidden layer, based on the image information of each fault type, a fuzzy C-means clustering algorithm is adopted to obtain a corresponding initial fuzzy rule, and the problem of regular optimization is solved
Obtaining the weight vector value w of the fuzzy rule in each fuzzy inference module l ＝[λ 1 I+(y l ) T (y l )] -1 (y l ) T z l Where I is the identity matrix, λ 1 Representing regularization coefficients，z l ＝(z 1 ,z 2 ,…,z l ) For the desired output residual vector, z 1 Y is the desired output, z when l 2, …, s l ＝z l-1 -ε l-2 ；
5) In the output layer, the output layer is connected with the hidden layer, and the output result is obtained by accumulating all intermediate weighted values layer by layer
Wherein
Representing the type of failure of the final output.
7. An image and depth blurring-based central air-conditioning fault diagnosis system, which is used for implementing the steps of an image and depth blurring-based central air-conditioning fault diagnosis method of any one of claims 1-6 when being executed, and comprises the following steps:
a data processing module for performing the method of step (1);
a map conversion module for performing the method of step (2);
a residual depth blurring module for performing the method of step (3);
and (3) a fault diagnosis module for executing the method of the step (4).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210571075.XA CN114941890B (en) | 2022-05-24 | 2022-05-24 | Central air conditioner fault diagnosis method and system based on image and depth blur |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210571075.XA CN114941890B (en) | 2022-05-24 | 2022-05-24 | Central air conditioner fault diagnosis method and system based on image and depth blur |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114941890A true CN114941890A (en) | 2022-08-26 |
| CN114941890B CN114941890B (en) | 2024-04-30 |
Family
ID=82908875
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210571075.XA Active CN114941890B (en) | 2022-05-24 | 2022-05-24 | Central air conditioner fault diagnosis method and system based on image and depth blur |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114941890B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117556202A (en) * | 2023-11-09 | 2024-02-13 | 南通大学 | Industrial process micro fault detection method based on probability correlation slow feature analysis |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080219336A1 (en) * | 2007-03-05 | 2008-09-11 | General Electric Company | System and method for fault detection and localization in time series and spatial data |
| CN110880024A (en) * | 2019-12-05 | 2020-03-13 | 山东建筑大学 | Nonlinear process fault identification method and system based on discrimination kernel slow characteristic analysis |
| US20200285900A1 (en) * | 2019-03-06 | 2020-09-10 | Wuhan University | Power electronic circuit fault diagnosis method based on optimizing deep belief network |
| CN112214006A (en) * | 2020-10-13 | 2021-01-12 | 山东建筑大学 | Intermittent process fault detection method and system considering two-dimensional dynamic characteristics |
| CN113869339A (en) * | 2021-05-18 | 2021-12-31 | 华能沁北发电有限责任公司 | Deep learning classification model for fault diagnosis and fault diagnosis method |
| CN114021275A (en) * | 2021-10-29 | 2022-02-08 | 上海海事大学 | Rolling bearing fault diagnosis method based on deep convolution fuzzy system |
-
2022
- 2022-05-24 CN CN202210571075.XA patent/CN114941890B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080219336A1 (en) * | 2007-03-05 | 2008-09-11 | General Electric Company | System and method for fault detection and localization in time series and spatial data |
| US20200285900A1 (en) * | 2019-03-06 | 2020-09-10 | Wuhan University | Power electronic circuit fault diagnosis method based on optimizing deep belief network |
| CN110880024A (en) * | 2019-12-05 | 2020-03-13 | 山东建筑大学 | Nonlinear process fault identification method and system based on discrimination kernel slow characteristic analysis |
| CN112214006A (en) * | 2020-10-13 | 2021-01-12 | 山东建筑大学 | Intermittent process fault detection method and system considering two-dimensional dynamic characteristics |
| CN113869339A (en) * | 2021-05-18 | 2021-12-31 | 华能沁北发电有限责任公司 | Deep learning classification model for fault diagnosis and fault diagnosis method |
| CN114021275A (en) * | 2021-10-29 | 2022-02-08 | 上海海事大学 | Rolling bearing fault diagnosis method based on deep convolution fuzzy system |
Non-Patent Citations (1)
| Title |
|---|
| 张汉元等: "基于改进核慢特征分析的间歇过程故障检测", 山东建筑大学学报, vol. 35, no. 01, 15 February 2020 (2020-02-15), pages 42 - 49 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117556202A (en) * | 2023-11-09 | 2024-02-13 | 南通大学 | Industrial process micro fault detection method based on probability correlation slow feature analysis |
| CN117556202B (en) * | 2023-11-09 | 2024-06-11 | 南通大学 | Industrial process micro fault detection method based on probability correlation slow feature analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114941890B (en) | 2024-04-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113191215B (en) | Rolling bearing fault diagnosis method integrating attention mechanism and twin network structure | |
| CN110287849B (en) | Lightweight depth network image target detection method suitable for raspberry pi | |
| CN108319962B (en) | Tool wear monitoring method based on convolutional neural network | |
| CN106895975B (en) | Bearing fault diagnosis method based on Stacked SAE deep neural network | |
| CN111914883B (en) | Spindle bearing state evaluation method and device based on deep fusion network | |
| CN114092832B (en) | High-resolution remote sensing image classification method based on parallel hybrid convolutional network | |
| CN107688856B (en) | Indoor robot scene active identification method based on deep reinforcement learning | |
| CN106096655B (en) | A kind of remote sensing image airplane detection method based on convolutional neural networks | |
| CN114358124B (en) | New fault diagnosis method for rotary machinery based on deep countermeasure convolutional neural network | |
| CN108509701B (en) | Rotary machine fault direct intelligent diagnosis method based on vibration signal | |
| CN112232328A (en) | Remote sensing image building area extraction method and device based on convolutional neural network | |
| CN110675421B (en) | Depth image collaborative segmentation method based on few labeling frames | |
| CN114429152A (en) | Rolling bearing fault diagnosis method based on dynamic index antagonism self-adaption | |
| CN113392881A (en) | Rotary machine fault diagnosis method based on transfer learning | |
| CN114169377A (en) | G-MSCNN-based fault diagnosis method for rolling bearing in noisy environment | |
| CN114548190A (en) | Wind turbine fault diagnosis method based on self-adaptive residual error neural network | |
| CN116467577A (en) | Fan gear box fault diagnosis method based on MTF and improved dense connection network | |
| CN117009916A (en) | Actuator fault diagnosis method based on multi-sensor information fusion and transfer learning | |
| CN115204272A (en) | Industrial system fault diagnosis method and equipment based on multi-sampling rate data | |
| CN115290326A (en) | Rolling bearing fault intelligent diagnosis method | |
| CN114359702A (en) | Method and system for identifying building violation of remote sensing image of homestead based on Transformer | |
| CN114941890A (en) | Central air conditioner fault diagnosis method and system based on image and depth blurring | |
| CN115508073A (en) | Prototype adaptation mechanical equipment fault diagnosis method based on multi-scale attention | |
| CN114548153B (en) | Planetary gear box fault diagnosis method based on residual error-capsule network | |
| CN111488974A (en) | Deep learning neural network-based ocean wind energy downscaling method |
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 |