CN113092083B - Machine pump fault diagnosis method and device based on fractal dimension and neural network - Google Patents
Machine pump fault diagnosis method and device based on fractal dimension and neural network Download PDFInfo
- Publication number
- CN113092083B CN113092083B CN202110149123.1A CN202110149123A CN113092083B CN 113092083 B CN113092083 B CN 113092083B CN 202110149123 A CN202110149123 A CN 202110149123A CN 113092083 B CN113092083 B CN 113092083B
- Authority
- CN
- China
- Prior art keywords
- fractal dimension
- fractal
- neural network
- matrix
- dimension
- 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
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H17/00—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves, not provided for in the preceding groups
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Testing And Monitoring For Control Systems (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a machine pump fault diagnosis method and a device based on fractal dimension and a neural network, which acquire a first vibration signal by acquiring the vibration signal of a centrifugal pump; performing empirical mode decomposition on the first vibration information to obtain a component set; sequencing a plurality of components in the component set according to the decomposition sequence of the signals to obtain preset condition components; calculating a fractal dimension according to a preset condition component to obtain a fractal dimension set; constructing a fractal matrix according to the fractal dimension set; and inputting the fractal matrix into a convolutional neural network model to obtain an output result, wherein the output result comprises a fault classification result. The technical problems that in the prior art, the fault diagnosis of the machine pump is mainly carried out aiming at certain problems, the signal characteristic extraction cannot be effectively carried out, and the comprehensive fault diagnosis and classification can be carried out are solved. The method achieves the technical effects of performing fault classification on the fractal matrix by using the convolutional neural network, facilitating better extraction of signal characteristics and realizing multiple classification fault diagnosis.
Description
Technical Field
The invention relates to the technical field of machine pump fault diagnosis, in particular to a machine pump fault diagnosis method and device based on fractal dimension and a neural network.
Background
The fault diagnosis of industrial equipment is always a research hotspot, and students at home and abroad have developed a certain research on the fault diagnosis of the pump and analyzed the reason and influencing factors of the vibration of the centrifugal pump; the common diagnosis method is to use pressure pulsation to diagnose cavitation fault problems of the variable-speed centrifugal pump; combining autoregressive spectrum analysis with a hidden Markov model, wherein the model has good diagnosis effect on common centrifugal pump faults; according to the related theory of the Mahalanobis-Taguchi system, fault feature extraction, fault separation and analysis are carried out on the collected vibration signals of the centrifugal pump, so that fault components such as sealing, impeller and filter blockage can be successfully diagnosed; the method utilizes statistical parameters (mean value, standard deviation, kurtosis, skewness and the like) as fault characteristics, utilizes a fuzzy neural network as a classifier, monitors various faults of the rotary machine in real time, and has good diagnosis effects on faults such as cavitation, impeller abrasion, rotor imbalance and the like.
However, in the process of implementing the technical scheme of the invention in the embodiment of the application, the inventor of the application finds that at least the following technical problems exist in the above technology:
In the prior art, the fault diagnosis of the machine pump is mainly aimed at solving the technical problems that signal characteristic extraction cannot be effectively carried out and comprehensive fault diagnosis and classification can be carried out.
Disclosure of Invention
The embodiment of the application solves the technical problems that in the prior art, the fault diagnosis of the pump is mainly carried out aiming at a certain type of problem, the signal characteristic extraction cannot be effectively carried out, and the comprehensive fault diagnosis and classification are carried out by providing the method and the device for diagnosing the fault of the pump based on the fractal dimension and the neural network. The method has the advantages that the fractal dimension is utilized to construct a fractal matrix of the signal, and then the convolutional neural network is utilized to perform fault classification on the fractal matrix, so that signal characteristics can be conveniently and well extracted, a better diagnosis and classification effect is obtained, the accuracy of fault classification and diagnosis is effectively improved, and the technical effects of multiple classification and fault diagnosis are realized.
In view of the above problems, embodiments of the present application provide a method and an apparatus for diagnosing a pump failure based on a fractal dimension and a neural network.
In a first aspect, an embodiment of the present application provides a machine pump fault diagnosis method based on a fractal dimension and a neural network, which is applied to a centrifugal pump, and the method includes: collecting vibration signals of a centrifugal pump to obtain a first vibration signal; performing empirical mode decomposition on the first vibration information to obtain a component set; sequencing a plurality of components in the component set according to the decomposition sequence of the signals to obtain preset condition components; calculating a fractal dimension according to the preset condition component to obtain a fractal dimension set; constructing a fractal matrix according to the fractal dimension set; and inputting the fractal matrix into a convolutional neural network model to obtain an output result, wherein the output result comprises a fault classification result.
On the other hand, the application also provides a machine pump fault diagnosis device based on the fractal dimension and the neural network, and the device comprises:
the first obtaining unit is used for collecting vibration signals of the centrifugal pump and obtaining first vibration signals;
the second obtaining unit is used for carrying out empirical mode decomposition on the first vibration information to obtain a component set;
a third obtaining unit, configured to sort a plurality of components in the component set according to a decomposition order of signals, to obtain a preset condition component;
the first calculation unit is used for calculating the fractal dimension according to the preset condition component to obtain a fractal dimension set;
the first construction unit is used for constructing a fractal matrix according to the fractal dimension set;
the fourth obtaining unit is used for inputting the fractal matrix into a convolutional neural network model to obtain an output result, and the output result comprises a fault classification result.
In a third aspect, the present invention provides a pump failure diagnosis apparatus based on a fractal dimension and a neural network, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the method of the first aspect when executing the program.
One or more technical solutions provided in the embodiments of the present application at least have the following technical effects or advantages:
the embodiment of the application provides a machine pump fault diagnosis method and device based on fractal dimension and a neural network, wherein a first vibration signal is obtained by collecting a vibration signal of a centrifugal pump; performing empirical mode decomposition on the first vibration information to obtain a component set; sequencing a plurality of components in the component set according to the decomposition sequence of the signals to obtain preset condition components; calculating a fractal dimension according to the preset condition component to obtain a fractal dimension set; constructing a fractal matrix according to the fractal dimension set; and inputting the fractal matrix into a convolutional neural network model to obtain an output result, wherein the output result comprises a fault classification result. On the basis of fractal dimension, the running state of equipment is represented by extracting a fractal matrix, and then signal feature extraction is performed by adding a machine learning neural network model, so that fault classification diagnosis is performed more effectively and comprehensively, and the technical problems that in the prior art, the machine pump fault diagnosis is mainly performed aiming at a certain type of problem, the signal feature extraction cannot be performed effectively, and the comprehensive fault diagnosis classification is performed are solved. The method has the advantages that the fractal dimension is utilized to construct a fractal matrix of the signal, and then the convolutional neural network is utilized to perform fault classification on the fractal matrix, so that signal characteristics can be conveniently and well extracted, a better diagnosis and classification effect is obtained, the accuracy of fault classification and diagnosis is effectively improved, and the technical effects of multiple classification and fault diagnosis are realized.
The foregoing description is only an overview of the technical solutions of the present application, and may be implemented according to the content of the specification in order to make the technical means of the present application more clearly understood, and in order to make the above-mentioned and other objects, features and advantages of the present application more clearly understood, the following detailed description of the present application will be given.
Drawings
Fig. 1 is a schematic flow chart of a machine pump fault diagnosis method based on fractal dimension and neural network according to an embodiment of the application;
FIG. 2 is a schematic diagram of a state space according to an embodiment of the present application;
fig. 3 is a schematic structural diagram of a pump fault diagnosis device based on fractal dimension and neural network according to an embodiment of the present application;
fig. 4 is a schematic structural diagram of an exemplary electronic device according to an embodiment of the present application.
Reference numerals illustrate: the first obtaining unit 11, the second obtaining unit 12, the third obtaining unit 13, the first calculating unit 14, the first constructing unit 15, the fourth obtaining unit 16, the bus 300, the receiver 301, the processor 302, the transmitter 303, the memory 304, the bus interface 306.
Detailed Description
The embodiment of the application solves the technical problems that in the prior art, the fault diagnosis of the pump is mainly carried out aiming at a certain type of problem, the signal characteristic extraction cannot be effectively carried out, and the comprehensive fault diagnosis and classification are carried out by providing the method and the device for diagnosing the fault of the pump based on the fractal dimension and the neural network. The method has the advantages that the fractal dimension is utilized to construct a fractal matrix of the signal, and then the convolutional neural network is utilized to perform fault classification on the fractal matrix, so that signal characteristics can be conveniently and well extracted, a better diagnosis and classification effect is obtained, the accuracy of fault classification and diagnosis is effectively improved, and the technical effects of multiple classification and fault diagnosis are realized. Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only some of the embodiments of the present application and not all of the embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.
Summary of the application
The fault diagnosis of industrial equipment is always a research hotspot, and students at home and abroad have developed a certain research on the fault diagnosis of the pump and analyzed the reason and influencing factors of the vibration of the centrifugal pump; the common diagnosis method is to use pressure pulsation to diagnose cavitation fault problems of the variable-speed centrifugal pump; combining autoregressive spectrum analysis with a hidden Markov model, wherein the model has good diagnosis effect on common centrifugal pump faults; according to the related theory of the Mahalanobis-Taguchi system, fault feature extraction, fault separation and analysis are carried out on the collected vibration signals of the centrifugal pump, so that fault components such as sealing, impeller and filter blockage can be successfully diagnosed; the method utilizes statistical parameters (mean value, standard deviation, kurtosis, skewness and the like) as fault characteristics, utilizes a fuzzy neural network as a classifier, monitors various faults of the rotary machine in real time, and has good diagnosis effects on faults such as cavitation, impeller abrasion, rotor imbalance and the like. However, in the prior art, the fault diagnosis of the pump is mainly performed aiming at certain problems, and the signal characteristics cannot be effectively extracted, so that the technical problem of comprehensive fault diagnosis and classification is solved.
Aiming at the technical problems, the technical scheme provided by the application has the following overall thought:
collecting vibration signals of a centrifugal pump to obtain a first vibration signal; performing empirical mode decomposition on the first vibration information to obtain a component set; sequencing a plurality of components in the component set according to the decomposition sequence of the signals to obtain preset condition components; calculating a fractal dimension according to the preset condition component to obtain a fractal dimension set; constructing a fractal matrix according to the fractal dimension set; and inputting the fractal matrix into a convolutional neural network model to obtain an output result, wherein the output result comprises a fault classification result. The method has the advantages that the fractal dimension is utilized to construct a fractal matrix of the signal, and then the convolutional neural network is utilized to perform fault classification on the fractal matrix, so that signal characteristics can be conveniently and well extracted, a better diagnosis and classification effect is obtained, the accuracy of fault classification and diagnosis is effectively improved, and the technical effects of multiple classification and fault diagnosis are realized.
Having described the basic principles of the present application, various non-limiting embodiments of the present application will now be described in detail with reference to the accompanying drawings.
Example 1
As shown in fig. 1, an embodiment of the present application provides a machine pump fault diagnosis method based on a fractal dimension and a neural network, which is applied to a centrifugal pump, and the method includes:
Step S100: collecting vibration signals of a centrifugal pump to obtain a first vibration signal;
specifically, vibration information of the centrifugal pump is collected through the vibration signal collecting device of the centrifugal pump, and the first vibration information is vibration data collected by the centrifugal pump which is processed and analyzed currently. The signal acquisition technology is not described in detail in the prior art.
Step S200: and performing empirical mode decomposition on the first vibration information to obtain a component set.
Specifically, the vibration signal in the first vibration information is decomposed by empirical mode decomposition, i.e., EMD decomposition, to obtain components therein. Empirical mode decomposition (empirical mode decomposition, EMD) is an important component of the Hilbert-yellow change, a new approach to dealing with non-stationary signals, proposed by Norden e.Huang doctor, the Chinese national astronaut, 1998. The time-frequency analysis method based on EMD is suitable for the analysis of nonlinear and non-stationary signals and the analysis of linear and stationary signals, and the analysis of the linear and stationary signals reflects the physical significance of the signals better than other time-frequency analysis methods. Wherein the decomposed components, i.e., the content modal components (Intrinsic Mode Functions, IMF), are the signal components of each layer obtained after the original signal is decomposed by EMD. The EMD presenter Huang E believes that any signal can be split into the sum of several connotative modal components. While the connotation modality component has two constraints: 1) The number of extreme points and the number of zero crossings must be equal or differ by at most not more than one in the whole data segment. 2) At any time, the average value of the upper envelope formed by the local maximum points and the lower envelope formed by the local minimum points is zero, that is, the upper and lower envelopes are locally symmetrical with respect to the time axis.
Step S300: sequencing a plurality of components in the component set according to the decomposition sequence of the signals to obtain preset condition components;
specifically, the corresponding components are obtained according to the EMD decomposition process, the first 5 components after decomposition are taken, specific numerical values are not specifically limited, corresponding adjustment can be performed according to the conditions of the components to be decomposed and specific fault diagnosis targets, and the embodiment of the application takes the first five components as an example to specifically describe. That is, the preset conditions set in the embodiment of the present application are the first five components obtained by decomposition.
Step S400: calculating a fractal dimension according to the preset condition component to obtain a fractal dimension set;
further, the step S400 in the embodiment of the present application includes calculating a fractal dimension according to the preset condition component to obtain a fractal dimension set:
step S410: obtaining a preset algorithm, wherein the preset algorithm comprises a plurality of dimension categories;
step S420: and carrying out dimension calculation on the preset condition components according to the preset algorithm to obtain fractal dimension sets of each component, wherein the fractal dimension sets of each component comprise a plurality of dimension categories.
Specifically, the preset algorithm includes a Box dimension, a Higuchi dimension, a detrusion dimension, and a Petrosian dimension, and the Box dimension, the Higuchi dimension, the detrusion dimension, and the Petrosian dimension are calculated for the determined components respectively to obtain a fractal dimension of each component, wherein the fractal dimension includes the Box dimension, the Higuchi dimension, the detrusion dimension, and the Petrosian dimension of the predetermined conditional component.
Step S500: constructing a fractal matrix according to the fractal dimension set;
further, the step S500 of the embodiment of the present application includes:
step S510: classifying the fractal dimension set according to the numerical variation, and extracting a fractal dimension matrix to obtain N fractal dimension matrices, wherein N is a natural number greater than 1;
step S520: and constructing the fractal matrix according to the fractal dimension matrix.
Further, the embodiment of the application further includes:
step S710: obtaining an equipment space;
step S720: according to the equipment space and the N fractal matrixes, N state spaces are obtained, wherein each state space comprises a first endpoint and a second endpoint, the first endpoint is provided with a first fractal dimension, the second endpoint is provided with a second fractal dimension, the first fractal dimension is the minimum fractal dimension of the state space, and the second fractal dimension is the maximum fractal dimension of the state space.
Further, the embodiment of the application further includes: the N state spaces satisfy the formula S 1 ∩S 2 ∩……∩S N =0, and S 1 、S 2 ……S N Not equal to 0, wherein S 1 Is a first state space, S 2 Is a first state space, S N Is the nth state space.
Specifically, the fractal matrix is extracted according to the change of the fractal dimension value. The fractal dimension is a parameter for measuring the fractal characteristics of the data, and is a measurement parameter for quantifying the fractal characteristics, so that the fractal dimension appears in a numerical form and has the characteristics of simplicity and intuitiveness, and the state of the signal can be distinguished through the change of the numerical value of the fractal dimension, so that the classification of faults is realized. The core idea of extracting the fractal matrix is to represent the running state of the equipment in the form of the fractal matrix, in the fractal fault diagnosis, the state space of any equipment can be represented by a straight line, different states of the equipment can be distinguished by different line segments on the straight line, and the schematic line segment diagrams of the different states in the state space are shown in figure 2. The device has N states, N intervals should be divided in the device space, and the start and end points of each line segment are the minimum and maximum values of the fractal dimension of the state. For example, state 1, with a state space of [ D1, D1 ]']D1 and D1' are the fractal dimension values of state 1 at the endpoints, respectively. Assuming that the state space of the device is S, the state section Sj in the j-th device is S j ＝[Dj,Dj′]Where j=1, 2,3 … N, dj and Dj' are the minimum and maximum values, respectively, of the fractal dimension in this state, where the interval length of state j is |s j |＝[Dj,Dj′]Each status regionThe length of the space is determined by the maximum value and the minimum value of the fractal dimension, the state space is a straight line, the device state is a line segment, therefore, a region in the state space is inevitably provided with a state without correspondence, and the useful state region S u Is S u ＝S1∪S2∪…∪S N The useless status areas are:
the division of the state space is very important using the fractal dimension as a criterion for state discrimination, and if the state space can be well divided, fault diagnosis will be easy. The partitioning of the state space should satisfy the following condition S 1 ∩S 2 ∩……∩S N =0, and S 1 、S 2 ……S N Not 0, i.e. the set of states should not be empty and there should be no intersections between states. However, in practical applications, the fractal dimension of the state interval has a certain fluctuation range, and the fractal dimension characterizes the same state within the range. If there is an intersection of the divided ranges of the fractal dimension, the signal may be misjudged. Or the signal has the characteristics of two states, the characterization state of the fractal dimension can not be determined in the overlapped interval range, and in order to realize fault diagnosis, the condition of interval division needs to be relaxed and some necessary processing needs to be carried out. Besides enlarging the dividing interval of the fractal dimension, the algorithm also uses a plurality of fractal dimension to represent signal state characteristics, and the fractal dimension used in the algorithm is Box dimension, higuchi dimension, detrend dimension and Petrosian dimension respectively so as to determine the certainty of fault diagnosis and improve the accuracy of fault diagnosis results.
Step S600: and inputting the fractal matrix into a convolutional neural network model to obtain an output result, wherein the output result comprises a fault classification result.
Further, the convolutional neural network model includes an input layer, an implicit layer, and an output layer, and the step S600 in this embodiment of the present application includes:
step S610: inputting the fractal matrix into the input layer for data activation;
step S620: inputting the activated fractal matrix into the hidden layer for feature extraction;
step S630: and obtaining a fault classification result through the classifier of the output layer, wherein the fault classification result is the output result.
Further, before the fractal matrix is input into the convolutional neural network model to obtain an output result, the embodiment of the application further includes:
step S640: training the convolutional neural network model, wherein the training the convolutional neural network model comprises:
step S641: initializing the fractal matrix to obtain training data;
step S642: determining input data and target output information according to the training data;
Step S643: processing the training data through the hidden layer and the output layer to obtain actual output information;
step S644: obtaining output offset according to the target output information and the actual output information;
step S645: judging whether the output offset meets the requirement of a preset range;
step S646: when the output offset is not met, obtaining a network error according to the output offset;
step S647: obtaining an error gradient according to the network error;
step S648: obtaining an updating weight according to the error gradient;
step S649: and re-inputting the hidden layer for processing according to the updated weight until the output offset meets the preset range requirement.
Specifically, the constructed fractal matrix is input into a trained convolutional neural network model, a fault classification result is obtained through a Softmax classifier of an output layer of the convolutional neural network model and is output, a final fault diagnosis result is obtained, and the purposes of conveniently and well extracting signal characteristics and obtaining a better fault diagnosis effect are achieved. Therefore, the technical problems that in the prior art, the fault diagnosis of the machine pump is mainly carried out aiming at certain problems, the signal characteristic extraction cannot be effectively carried out, and the comprehensive fault diagnosis and classification are carried out are solved. On the basis of fractal dimension, the running state of the equipment is represented by extracting a fractal matrix, and then signal feature extraction is carried out by adding machine learning, namely a neural network model, so that fault classification diagnosis is more effectively and comprehensively carried out. Experiments show that the fractal dimensions are different under different fault states, the fractal matrix of the signal is constructed by using the fractal dimensions, and then the fractal matrix is subjected to fault classification by using a convolutional neural network, so that a better diagnosis effect is obtained. The convolutional neural network is divided into three structures, namely an input layer, an implicit layer and an output layer, the obtained fractal matrix is input into the convolutional neural network for feature extraction, and a Softmax classifier of the output layer is utilized to obtain a fault classification result. The training of the neural network is divided into two steps, namely forward feature extraction and reverse gradient propagation, and the classification accuracy and loss values of different optimization algorithms in the convolutional neural network training process are different. The loss value represents the difference between the actual network output and the ideal output, and the lower the loss value is, the better the training effect of the network is, and the higher the accuracy is. In order to improve accuracy, different optimization functions are respectively processed, and a conclusion is drawn that the training effect has larger difference, wherein the loss value of the SGD-based optimization algorithm is higher than that of the Adadelta-based optimization algorithm, and the convergence rate of the SGD-based optimization algorithm is slower according to the loss value of the SGD-based optimization algorithm. Based on the RMS, adam and Nadam optimization algorithms, the loss value and accuracy trends are similar, the loss value and accuracy trends are relatively fast and tend to be stable, the final loss value is stabilized at about 0.26, and the accuracy is stabilized at about 90%, so that the training effect can be improved by selecting a proper optimization algorithm, and the effect of accurately performing fault classification diagnosis is achieved.
It should be appreciated that convolutional neural networks (Convolutional Neural Networks, CNN) are a class of feed-forward neural networks (Feedforward Neural Networks) that contain convolutional calculations and have a deep structure, and are one of the representative algorithms of deep learning [1-2]. The convolutional neural network has the capability of representation learning (representation learning), can perform Shift-invariant classification (Shift-invariant classification) on input information according to a hierarchical structure, is therefore also called as a visual perception (visual perception) mechanism construction of a Shift-invariant artificial neural network (Shift-Invariant Artificial Neural Networks, SIANN), can perform supervised learning and unsupervised learning, and has the advantages that the convolutional kernel parameters in hidden layers are shared and the sparsity of interlayer connection enables the convolutional neural network to perform learning on grid-like characteristics, such as pixels and audio, with stable effect and no additional feature engineering (feature engineering) requirements on the data.
The training process is essentially a supervised learning process, and each group of training data is continuously self-corrected and adjusted until the obtained output result is consistent with the target output result, and then the data supervised learning of the group is ended, and the next data supervised learning is carried out; and when the output information of the convolutional neural network reaches the preset accuracy rate/reaches a convergence state, ending the supervised learning process. Through the supervised learning of the convolutional neural network, the convolutional neural network is enabled to process the input information more accurately, and therefore more accurate and proper fault classification diagnosis results are obtained.
Further, the embodiment of the application further includes:
step S810: obtaining the first fractal dimension and the second fractal dimension according to the fractal dimension set until the P fractal dimension, wherein P is a natural number larger than 1;
step S820: generating a first verification code according to the first fractal dimension, wherein the first verification code corresponds to the first fractal dimension one by one;
step S830: generating a second verification code according to the second fractal dimension and the first verification code, and so on, generating a P verification code according to the P fractal dimension and the P-1 verification code;
step S840: and copying and storing all fractal dimensions and verification codes on Q electronic devices, wherein Q is a natural number larger than 1.
Specifically, in order to improve accuracy of a model processing result, prevent input data and training data from being tampered to influence accuracy of a model output result, and improve safety of data storage. Blockchain technology, also known as distributed ledger technology, is an emerging technology that is commonly engaged in "accounting" by several computing devices, together maintaining a complete distributed database. The blockchain technology has the characteristics of decentralization, disclosure transparency, capability of participating in database recording by each computing device and capability of rapidly performing data synchronization among the computing devices, so that the blockchain technology is widely applied in a plurality of fields. Generating a first verification code according to the first fractal dimension, wherein the first verification code corresponds to the first fractal dimension one by one; generating a second verification code according to the second fractal dimension and the first verification code, wherein the second verification code corresponds to the second fractal dimension one by one; and by analogy, generating an Nth verification code according to the Nth fractal dimension and the N-1 th verification code, wherein N is a natural number larger than 1, respectively copying and storing all the fractal dimensions and the verification codes on M pieces of equipment, wherein the first fractal dimension and the first verification code are stored on one piece of equipment as a first storage unit, the second fractal dimension and the second verification code are stored on one piece of equipment as a second storage unit, the Nth fractal dimension and the Nth verification code are stored on one piece of equipment as an Nth storage unit, when the fractal dimension is required to be called, each next node receives data stored by a previous node, and then stores the data after checking through a 'consensus mechanism', and concatenates the data to each storage unit through a hash function, so that screening conditions are not easy to lose and destroy, the fractal dimension is encrypted through logic of a block chain, the fractal security is further ensured, the convolutional neural network model obtained through fractal dimension training is further ensured, the accuracy of a convolutional neural network obtained through fractal dimension training is ensured, and the reliability of the obtained after-fault classification result is ensured.
Example two
Based on the same inventive concept as the machine pump fault diagnosis method based on the fractal dimension and the neural network in the foregoing embodiment, the present invention further provides a machine pump fault diagnosis device based on the fractal dimension and the neural network, as shown in fig. 3, where the device includes:
a first obtaining unit 11, where the first obtaining unit 11 is configured to collect a vibration signal of a centrifugal pump, and obtain a first vibration signal;
a second obtaining unit 12, where the second obtaining unit 12 is configured to perform empirical mode decomposition on the first vibration information to obtain a component set;
a third obtaining unit 13, where the third obtaining unit 13 is configured to sort a plurality of components in the component set according to a decomposition order of signals, so as to obtain a preset condition component;
a first calculating unit 14, where the first calculating unit 14 is configured to calculate a fractal dimension according to the preset condition component, and obtain a fractal dimension set;
a first construction unit 15, where the first construction unit 15 is configured to construct a fractal matrix according to the fractal dimension set;
and a fourth obtaining unit 16, where the fourth obtaining unit 16 is configured to input the fractal matrix into a convolutional neural network model, and obtain an output result, where the output result includes a fault classification result.
Further, the device further comprises:
a fifth obtaining unit, configured to obtain a preset algorithm, where the preset algorithm includes multiple dimension categories;
and the sixth obtaining unit is used for respectively carrying out dimension calculation on the preset condition components according to the preset algorithm to obtain a fractal dimension set of each component, wherein the fractal dimension set of each component comprises a plurality of dimension categories.
Further, the device further comprises:
a seventh obtaining unit, configured to classify the fractal dimension set according to a numerical variation, extract a fractal dimension matrix, and obtain N fractal dimension matrices, where N is a natural number greater than 1;
the second construction unit is used for constructing the fractal dimension matrix according to the fractal dimension matrix.
Further, the device further comprises:
an eighth obtaining unit for obtaining an equipment space;
and a ninth obtaining unit, configured to obtain N state spaces according to the device space and the N fractal matrices, where each state space includes a first endpoint and a second endpoint, the first endpoint has a first fractal dimension, the second endpoint has a second fractal dimension, the first fractal dimension is a minimum fractal dimension of the state space, and the second fractal dimension is a maximum fractal dimension of the state space.
Further, the device further comprises: the N state spaces satisfy the formula S 1 ∩S 2 ∩……∩S N =0, and S 1 、S 2 ……S N Not equal to 0, wherein S 1 Is a first state space, S 2 Is a first state space, S N Is the nth state space.
Further, the device further comprises:
the first execution unit is used for inputting the fractal matrix into the input layer for data activation;
the second execution unit is used for inputting the activated fractal matrix into the hidden layer for feature extraction;
a tenth obtaining unit, configured to obtain a fault classification result through a classifier of the output layer, where the fault classification result is the output result.
Further, the device further comprises:
the first training unit is used for training the convolutional neural network model, wherein the training of the convolutional neural network model comprises the following steps:
an eleventh obtaining unit, configured to initialize the fractal matrix to obtain training data;
the first determining unit is used for determining input data and target output information according to the training data;
A twelfth obtaining unit, configured to process the training data through the hidden layer and the output layer, to obtain actual output information;
a thirteenth obtaining unit configured to obtain an output offset according to the target output information and the actual output information;
the first judging unit is used for judging whether the output offset meets the requirement of a preset range;
a fourteenth obtaining unit configured to obtain a network error according to the output offset when not satisfied;
a fifteenth obtaining unit configured to obtain an error gradient according to the network error;
a sixteenth obtaining unit configured to obtain an update weight according to the error gradient;
and the third execution unit is used for inputting the hidden layer again according to the updated weight for processing until the output offset meets the preset range requirement.
Further, the device further comprises:
a seventeenth obtaining unit, configured to obtain the first fractal dimension, the second fractal dimension, and up to a P-th fractal dimension according to the fractal dimension set, where P is a natural number greater than 1;
The first generation unit is used for generating a first verification code according to the first fractal dimension, and the first verification code corresponds to the first fractal dimension one by one;
the second generation unit is used for generating a second verification code according to the second fractal dimension and the first verification code, and the second generation unit is used for generating a P verification code according to the P fractal dimension and the P-1 verification code;
and the first storage unit is used for copying and storing all fractal dimensions and verification codes on the Q electronic equipment, wherein Q is a natural number larger than 1.
The various variations and specific examples of the method for diagnosing a pump failure based on a fractal dimension and a neural network in the first embodiment of fig. 1 are equally applicable to the device for diagnosing a pump failure based on a fractal dimension and a neural network in the present embodiment, and by the detailed description of the method for diagnosing a pump failure based on a fractal dimension and a neural network, those skilled in the art can clearly know the implementation method of the device for diagnosing a pump failure based on a fractal dimension and a neural network in the present embodiment, so that the description is omitted herein for brevity.
Exemplary electronic device
An electronic device of an embodiment of the present application is described below with reference to fig. 4.
Fig. 4 illustrates a schematic structural diagram of an electronic device according to an embodiment of the present application.
Based on the inventive concept of a machine pump fault diagnosis method based on a fractal dimension and a neural network as in the previous embodiments, the present invention further provides a machine pump fault diagnosis device based on a fractal dimension and a neural network, on which a computer program is stored, which when executed by a processor, implements the steps of any one of the above-described machine pump fault diagnosis methods based on a fractal dimension and a neural network.
Where in FIG. 4, a bus architecture (represented by bus 300), bus 300 may comprise any number of interconnected buses and bridges, with bus 300 linking together various circuits, including one or more processors, represented by processor 302, and memory, represented by memory 304. Bus 300 may also link together various other circuits such as peripheral devices, voltage regulators, power management circuits, etc., as are well known in the art and, therefore, will not be described further herein. Bus interface 306 provides an interface between bus 300 and receiver 301 and transmitter 303. The receiver 301 and the transmitter 303 may be the same element, i.e. a transceiver, providing a means for communicating with various other systems over a transmission medium.
The processor 302 is responsible for managing the bus 300 and general processing, while the memory 304 may be used to store data used by the processor 302 in performing operations.
One or more technical solutions provided in the embodiments of the present application at least have the following technical effects or advantages:
the embodiment of the application provides a machine pump fault diagnosis method and device based on fractal dimension and a neural network, wherein a first vibration signal is obtained by collecting a vibration signal of a centrifugal pump; performing empirical mode decomposition on the first vibration information to obtain a component set; sequencing a plurality of components in the component set according to the decomposition sequence of the signals to obtain preset condition components; calculating a fractal dimension according to the preset condition component to obtain a fractal dimension set; constructing a fractal matrix according to the fractal dimension set; and inputting the fractal matrix into a convolutional neural network model to obtain an output result, wherein the output result comprises a fault classification result. On the basis of fractal dimension, the running state of equipment is represented by extracting a fractal matrix, and then signal feature extraction is performed by adding a machine learning neural network model, so that fault classification diagnosis is performed more effectively and comprehensively, and the technical problems that in the prior art, the machine pump fault diagnosis is mainly performed aiming at a certain type of problem, the signal feature extraction cannot be performed effectively, and the comprehensive fault diagnosis classification is performed are solved. The method has the advantages that the fractal dimension is utilized to construct a fractal matrix of the signal, and then the convolutional neural network is utilized to perform fault classification on the fractal matrix, so that signal characteristics can be conveniently and well extracted, a better diagnosis and classification effect is obtained, the accuracy of fault classification and diagnosis is effectively improved, and the technical effects of multiple classification and fault diagnosis are realized.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a system for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks. While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (5)
1. A machine pump fault diagnosis method based on fractal dimension and neural network, which is applied to a centrifugal pump, wherein the method comprises the following steps:
collecting vibration signals of a centrifugal pump to obtain a first vibration signal;
performing empirical mode decomposition on the first vibration information to obtain a component set;
sequencing a plurality of components in the component set according to the decomposition sequence of the signals to obtain preset condition components;
calculating a fractal dimension according to the preset condition component to obtain a fractal dimension set;
constructing a fractal matrix according to the fractal dimension set;
inputting the fractal matrix into a convolutional neural network model to obtain an output result, wherein the output result comprises a fault classification result;
calculating a fractal dimension according to the preset condition component to obtain a fractal dimension set, wherein the method comprises the following steps:
obtaining a preset algorithm, wherein the preset algorithm comprises a plurality of dimension categories;
Carrying out dimension calculation on the preset condition components according to the preset algorithm to obtain fractal dimension sets of each component, wherein the fractal dimension sets of each component comprise a plurality of dimension categories;
constructing a fractal matrix according to the fractal dimension set, wherein the fractal matrix comprises:
classifying the fractal dimension set according to the numerical variation, and extracting a fractal dimension matrix to obtain N fractal dimension matrices, wherein N is a natural number greater than 1; constructing the fractal dimension matrix according to the fractal dimension matrix;
the convolutional neural network model comprises an input layer, an implicit layer and an output layer, wherein the fractal matrix is input into the convolutional neural network model to obtain an output result, and the convolutional neural network model comprises the following components:
inputting the fractal matrix into the input layer for data activation;
inputting the activated fractal matrix into the hidden layer for feature extraction;
obtaining a fault classification result through a classifier of the output layer, wherein the fault classification result is the output result;
the step of inputting the fractal matrix into a convolutional neural network model, before obtaining an output result, comprises the following steps:
training the convolutional neural network model, wherein the training the convolutional neural network model comprises:
Initializing the fractal matrix to obtain training data;
determining input data and target output information according to the training data;
processing the training data through the hidden layer and the output layer to obtain actual output information;
obtaining output offset according to the target output information and the actual output information;
judging whether the output offset meets the requirement of a preset range;
when the output offset is not met, obtaining a network error according to the output offset;
obtaining an error gradient according to the network error;
obtaining an updating weight according to the error gradient;
re-inputting the hidden layer according to the updated weight for processing until the output offset meets the preset range requirement;
obtaining the first fractal dimension and the second fractal dimension according to the fractal dimension set until the P fractal dimension, wherein P is a natural number larger than 1;
generating a first verification code according to the first fractal dimension, wherein the first verification code corresponds to the first fractal dimension one by one;
generating a second verification code according to the second fractal dimension and the first verification code, and so on, generating a P verification code according to the P fractal dimension and the P-1 verification code;
And copying and storing all fractal dimensions and verification codes on Q electronic devices, wherein Q is a natural number larger than 1.
2. The method of claim 1, wherein the method comprises:
obtaining an equipment space;
according to the equipment space and the N fractal matrixes, N state spaces are obtained, wherein each state space comprises a first endpoint and a second endpoint, the first endpoint is provided with a first fractal dimension, the second endpoint is provided with a second fractal dimension, the first fractal dimension is the minimum fractal dimension of the state space, and the second fractal dimension is the maximum fractal dimension of the state space.
3. The method of claim 2, wherein the method comprises:
the N state spaces satisfy the formula s1N S2N … … N sn=0, and S1, S2 … … sn+.0, where S1 is a first state space, S2 is a second state space, and SN is an nth state space.
4. A machine pump fault diagnosis device based on fractal dimension and neural network, adapted for use in the steps of any one of claims 1-3, wherein said device comprises:
the first obtaining unit is used for collecting vibration signals of the centrifugal pump and obtaining first vibration signals;
The second obtaining unit is used for carrying out empirical mode decomposition on the first vibration information to obtain a component set;
a third obtaining unit, configured to sort a plurality of components in the component set according to a decomposition order of signals, to obtain a preset condition component;
the first calculation unit is used for calculating the fractal dimension according to the preset condition component to obtain a fractal dimension set;
the first construction unit is used for constructing a fractal matrix according to the fractal dimension set;
the fourth obtaining unit is used for inputting the fractal matrix into a convolutional neural network model to obtain an output result, and the output result comprises a fault classification result;
the first execution unit is used for inputting the fractal matrix into the input layer for data activation;
the second execution unit is used for inputting the activated fractal matrix into the hidden layer for feature extraction;
a tenth obtaining unit, configured to obtain a fault classification result through a classifier of the output layer, where the fault classification result is the output result;
The first training unit is used for training the convolutional neural network model, wherein the training of the convolutional neural network model comprises the following steps:
an eleventh obtaining unit, configured to initialize the fractal matrix to obtain training data;
the first determining unit is used for determining input data and target output information according to the training data;
a twelfth obtaining unit, configured to process the training data through the hidden layer and the output layer, to obtain actual output information;
a thirteenth obtaining unit configured to obtain an output offset according to the target output information and the actual output information;
the first judging unit is used for judging whether the output offset meets the requirement of a preset range;
a fourteenth obtaining unit configured to obtain a network error according to the output offset when not satisfied;
a fifteenth obtaining unit configured to obtain an error gradient according to the network error;
a sixteenth obtaining unit configured to obtain an update weight according to the error gradient;
The third execution unit is used for inputting the hidden layer again according to the updated weight for processing until the output offset meets the preset range requirement;
a seventeenth obtaining unit, configured to obtain the first fractal dimension, the second fractal dimension, and up to a P-th fractal dimension according to the fractal dimension set, where P is a natural number greater than 1;
the first generation unit is used for generating a first verification code according to the first fractal dimension, and the first verification code corresponds to the first fractal dimension one by one;
the second generation unit is used for generating a second verification code according to the second fractal dimension and the first verification code, and the second generation unit is used for generating a P verification code according to the P fractal dimension and the P-1 verification code;
and the first storage unit is used for copying and storing all fractal dimensions and verification codes on the Q electronic equipment, wherein Q is a natural number larger than 1.
5. A machine pump fault diagnosis device based on fractal dimension and neural network, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the method of any one of claims 1-3 when the processor executes the program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110149123.1A CN113092083B (en) | 2021-02-03 | 2021-02-03 | Machine pump fault diagnosis method and device based on fractal dimension and neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110149123.1A CN113092083B (en) | 2021-02-03 | 2021-02-03 | Machine pump fault diagnosis method and device based on fractal dimension and neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113092083A CN113092083A (en) | 2021-07-09 |
| CN113092083B true CN113092083B (en) | 2023-06-13 |
Family
ID=76663794
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110149123.1A Active CN113092083B (en) | 2021-02-03 | 2021-02-03 | Machine pump fault diagnosis method and device based on fractal dimension and neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113092083B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114406409B (en) * | 2022-03-30 | 2022-07-12 | 中国船级社 | Method, device and equipment for determining fault state of welding machine |
| CN115238389B (en) * | 2022-07-27 | 2023-05-16 | 中国航发沈阳发动机研究所 | Engine case maneuvering load analysis method of one-dimensional model of aeroengine |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109707615A (en) * | 2019-02-26 | 2019-05-03 | 东北石油大学 | Reciprocating compressor method for diagnosing faults based on fine multi-fractal |
| CN110188867A (en) * | 2019-06-17 | 2019-08-30 | 浙江浙能嘉华发电有限公司 | Steam turbine hostdown diagnostic method based on integrated empirical mode decomposition and convolutional neural networks |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101900789B (en) * | 2010-07-07 | 2012-05-02 | 湖南大学 | Tolerance analog circuit fault diagnosing method based on wavelet transform and fractal dimension |
| CN108303255A (en) * | 2018-01-09 | 2018-07-20 | 内蒙古科技大学 | Low-speed heave-load device Fault Diagnosis of Roller Bearings, equipment and medium |
| CN108544303A (en) * | 2018-03-30 | 2018-09-18 | 上海交通大学 | A kind of main shaft of numerical control machine tool fault diagnosis method and system |
| CN110147739A (en) * | 2019-04-29 | 2019-08-20 | 河海大学 | A kind of Reactor Fault recognition methods based on Multifractal Analysis |
-
2021
- 2021-02-03 CN CN202110149123.1A patent/CN113092083B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109707615A (en) * | 2019-02-26 | 2019-05-03 | 东北石油大学 | Reciprocating compressor method for diagnosing faults based on fine multi-fractal |
| CN110188867A (en) * | 2019-06-17 | 2019-08-30 | 浙江浙能嘉华发电有限公司 | Steam turbine hostdown diagnostic method based on integrated empirical mode decomposition and convolutional neural networks |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113092083A (en) | 2021-07-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Malhotra et al. | TimeNet: Pre-trained deep recurrent neural network for time series classification | |
| CN111694879B (en) | Multielement time sequence abnormal mode prediction method and data acquisition monitoring device | |
| Monahan | Nonlinear principal component analysis by neural networks: Theory and application to the Lorenz system | |
| US9542255B2 (en) | Troubleshooting based on log similarity | |
| CN113092083B (en) | Machine pump fault diagnosis method and device based on fractal dimension and neural network | |
| US10558207B1 (en) | Event monitoring system | |
| CN108985380B (en) | Point switch fault identification method based on cluster integration | |
| JP6661839B1 (en) | Time series data diagnosis device, additional learning method, and program | |
| Serir et al. | Evidential evolving Gustafson–Kessel algorithm for online data streams partitioning using belief function theory | |
| CN103513983A (en) | Method and system for predictive alert threshold determination tool | |
| Zheng et al. | Fault diagnosis system of bridge crane equipment based on fault tree and Bayesian network | |
| CN113868006B (en) | Time sequence detection method and device, electronic equipment and computer storage medium | |
| CN104679818A (en) | Video keyframe extracting method and video keyframe extracting system | |
| Cismondi et al. | Computational intelligence methods for processing misaligned, unevenly sampled time series containing missing data | |
| CN107133632A (en) | A kind of wind power equipment fault diagnosis method and system | |
| CN110011990A (en) | Intranet security threatens intelligent analysis method | |
| Elmi et al. | Computing skyline from evidential data | |
| CN114492555A (en) | Method and device for identifying reasons of abnormal operation of conveying pipeline | |
| CN116502162A (en) | Abnormal computing power federal detection method, system and medium in edge computing power network | |
| CN108573282A (en) | Target identification method based on the BN parameter learnings under small data set | |
| Zhou et al. | A generalized approach to possibilistic clustering algorithms | |
| CN112434930A (en) | Fault diagnosis method, system and equipment in drilling process | |
| US20170293608A1 (en) | Unusual score generators for a neuro-linguistic behavioral recognition system | |
| CN107220710B (en) | Calculation method of BN model parameters under condition of scarce sample data set | |
| CN114155578A (en) | Portrait clustering method, device, electronic equipment and storage medium |
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 |