CN115077685A - Equipment state detection method, device and system - Google Patents
Equipment state detection method, device and system Download PDFInfo
- Publication number
- CN115077685A CN115077685A CN202210543123.4A CN202210543123A CN115077685A CN 115077685 A CN115077685 A CN 115077685A CN 202210543123 A CN202210543123 A CN 202210543123A CN 115077685 A CN115077685 A CN 115077685A
- Authority
- CN
- China
- Prior art keywords
- target
- characteristic value
- signal
- historical
- domain characteristic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- 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
Abstract
The application discloses a method, a device and a system for detecting equipment states. Wherein, the method comprises the following steps: acquiring a target vibration signal of target equipment; performing singular spectrum analysis on the target vibration signal to determine a target component signal in the target vibration signal; performing time domain analysis and frequency domain analysis on the target component signal to obtain a time domain characteristic value and a frequency domain characteristic value; and inputting the time domain characteristic value and the frequency domain characteristic value into a state diagnosis model, and determining the operation state of the target equipment through the state diagnosis model, wherein the state diagnosis model is a decision tree model obtained based on historical operation data training of the target equipment. The method and the device solve the technical problems that in the related art, when the state of the equipment is detected, the calculation is complex and the efficiency is low due to manual analysis of the detection data.
Description
Technical Field
The application relates to the technical field of power equipment detection, in particular to a method, a device and a system for detecting equipment states.
Background
When GIS equipment has defects, mechanical motion can be generated under the action of factors such as mechanical force of switch operation, alternating electromotive force generated by load current and the like, so that the equipment generates abnormal vibration, the vibration has great harm to the equipment and can also generate potential safety hazard, and accidents can be caused after long-term development. Therefore, the detection of vibration defects and faults of the GIS is enhanced, and the method is an important means for ensuring the safe operation of the GIS.
The characteristic value of GIS in abnormal vibration is different from that of equipment in normal operation, so that the vibration analysis in GIS operation is used as an entry point to diagnose faults. At present, in GIS fault diagnosis based on vibration measurement, most vibration signal to the collection carries out analysis and judgement, when measuring the main equipment, for guaranteeing data acquisition's integrality and reliability, the quantity demand of sensor is great, test data is also more, this has increased GIS state detection's the degree of difficulty and work load undoubtedly, have the limitation, need a large amount of vibration signal data of staff analysis, consuming time and wasting power, and can not discover early trouble, there is the hysteresis quality to troubleshooting trouble, the staff carries out manual processing to all data, also can the greatly increased work repetition, and efficiency is general.
In view of the above problems, no effective solution has been proposed.
Disclosure of Invention
The embodiment of the application provides a method, a device and a system for detecting equipment states, which are used for at least solving the technical problems that when the equipment states are detected in the related art, the calculation is complex and the efficiency is low due to the fact that data are manually analyzed and detected.
According to an aspect of an embodiment of the present application, there is provided a device status detection method, including: acquiring a target vibration signal of target equipment; performing singular spectrum analysis on the target vibration signal to determine a target component signal in the target vibration signal; performing time domain analysis and frequency domain analysis on the target component signal to obtain a time domain characteristic value and a frequency domain characteristic value; and inputting the time domain characteristic value and the frequency domain characteristic value into a state diagnosis model, and determining the operation state of the target equipment through the state diagnosis model, wherein the state diagnosis model is a decision tree model obtained based on historical operation data training of the target equipment.
Optionally, acquiring a vibration signal acquired by a target sensor located on the target device; and performing analog-to-digital conversion on the vibration signal to obtain the target vibration signal.
Optionally, determining a first time sequence corresponding to the target vibration signal, and determining a trajectory matrix of the first time sequence; performing singular value decomposition on the track matrix, and decomposing the track matrix into a first number of elementary matrices; grouping the first number of elementary matrices, summing the elementary matrices in each group to obtain a second number of synthetic matrices, and determining a target synthetic matrix from the second number of synthetic matrices; and converting the target synthesis matrix into a second time sequence in a diagonal average calculation mode, and determining the target component signal based on the second time sequence.
Optionally, performing time domain analysis on the target component signal to obtain the time domain characteristic value, where the time domain characteristic value at least includes one of: a mean value representing a time domain average of the target component signal, a root mean square representing an amplitude and energy of the target component signal, a skewness representing an asymmetric feature of the target component signal, a kurtosis representing a peak of the target component signal, a crest factor representing a shock in the target component signal, a shape factor representing a shape of the target component signal; performing fourier transform on the target component signal to obtain a target frequency spectrum signal, and performing frequency domain analysis on the target frequency spectrum signal to obtain the frequency domain characteristic value, where the frequency domain characteristic value at least includes one of the following: a spectral centroid representing a centroid of the spectral signal, a spectral diffusivity representing a diffusivity of the spectral signal, a spectral skewness representing a symmetry of the spectral signal, a spectral kurtosis representing a transient signal location in the spectral signal, a spectral crest factor representing an indicator of a peak in the spectral signal, a spectral entropy representing an energy distribution of the spectral signal.
Optionally, the training process of the state diagnosis model includes: obtaining the historical operating data of the target device, wherein the historical operating data comprises: historical vibration signals and historical operating states corresponding to the historical vibration signals; performing singular spectrum analysis on the historical vibration signal, determining a historical component signal in the historical vibration signal, and performing time domain analysis and frequency domain analysis on the historical component signal to obtain a historical time domain characteristic value and a historical frequency domain characteristic value; constructing the decision tree model based on a gradient lifting decision tree algorithm, and performing iterative training on the decision tree model based on the historical time domain characteristic value, the historical frequency domain characteristic value and the historical operating state to obtain the state diagnosis model, wherein the decision tree model comprises: root node, intermediate nodes, and leaf nodes.
Optionally, inputting the historical time domain characteristic value and the historical frequency domain characteristic value into the decision tree model, and obtaining the predicted operation state of the equipment through the output of the decision tree model; constructing a target loss function based on the predicted operating state and the historical operating state; and in the iterative training process, optimizing the model parameters of the decision tree model by minimizing a target loss function to obtain the state diagnosis model.
Optionally, after the operating state of the target device is determined by the state diagnostic model, displaying the time domain characteristic value, the frequency domain characteristic value and the operating state of the target device in a display device; and generating fault early warning information when the operating state of the target equipment is determined to be abnormal, wherein the fault early warning information is used for prompting that the target equipment is abnormal and needs to be processed.
According to another aspect of the embodiments of the present application, there is also provided an apparatus for detecting a device status, including: the acquisition module is used for acquiring a target vibration signal of target equipment; the first analysis module is used for performing singular spectrum analysis on the target vibration signal and determining a target component signal in the target vibration signal; the second analysis module is used for carrying out time domain analysis and frequency domain analysis on the target component signal to obtain a time domain characteristic value and a frequency domain characteristic value; and the determining module is used for inputting the time domain characteristic value and the frequency domain characteristic value into a state diagnosis model and determining the operation state of the target equipment through the state diagnosis model, wherein the state diagnosis model is a decision tree model obtained based on historical operation data training of the target equipment.
According to another aspect of the embodiments of the present application, there is also provided an apparatus status detection system, including: the target sensor is positioned on target equipment and used for acquiring a vibration signal of the target equipment; the data acquisition equipment is used for receiving the vibration signal acquired by the target sensor, performing analog-to-digital conversion on the vibration signal to obtain a target vibration signal, and transmitting the target vibration signal to the processor through a communication bus; the processor is used for carrying out singular spectrum analysis on the target vibration signal and determining a target component signal in the target vibration signal; performing time domain analysis and frequency domain analysis on the target component signal to obtain a time domain characteristic value and a frequency domain characteristic value; inputting the time domain characteristic value and the frequency domain characteristic value into a state diagnosis model, and determining the operation state of the target equipment through the state diagnosis model, wherein the state diagnosis model is a decision tree model obtained based on historical operation data training of the target equipment; transmitting the time domain characteristic value, the frequency domain characteristic value and the running state of the target equipment to display equipment through a communication bus; the display device is used for displaying the time domain characteristic value, the frequency domain characteristic value and the running state of the target device; and the communication bus is used for transmitting data.
According to another aspect of the embodiments of the present application, a non-volatile storage medium is further provided, where the non-volatile storage medium includes a stored program, and when the program runs, the device in which the non-volatile storage medium is located is controlled to execute the device state detection method described above.
In the embodiment of the application, after a target vibration signal of target equipment is obtained, singular spectrum analysis is performed on the target vibration signal to determine a target component signal in the target vibration signal, time domain analysis and frequency domain analysis are performed on the target component signal to obtain a time domain characteristic value and a frequency domain characteristic value, the time domain characteristic value and the frequency domain characteristic value are input into a state diagnosis model, the running state of the target equipment is determined through the state diagnosis model, and the state diagnosis model is a decision tree model obtained based on historical running data training of the target equipment. According to the technical scheme, the vibration signals are automatically acquired and processed and analyzed, the analysis efficiency can be improved, meanwhile, the model based on the historical data training of the equipment is utilized for equipment state diagnosis, the accuracy of diagnosis results can be improved, and the technical problems that in the related technology, when the equipment state is detected, the calculation is complex and the efficiency is low due to manual analysis of detection data are solved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. In the drawings:
FIG. 1 is a schematic diagram of a device status detection system according to an embodiment of the present application;
FIG. 2 is a schematic flow chart diagram illustrating a method for device status detection according to an embodiment of the present application;
FIG. 3 is a schematic diagram of signal eigenvalues and eigenvectors according to an embodiment of the present application;
FIG. 4 is a schematic diagram of a target component signal according to an embodiment of the present application;
FIG. 5 is a schematic diagram of a decision tree decision process according to an embodiment of the present application;
FIG. 6 is a schematic diagram of a device status detection process according to an embodiment of the present application;
FIG. 7 is a diagram illustrating a device status detection result according to an embodiment of the present application;
fig. 8 is a schematic structural diagram of an apparatus state detection device according to an embodiment of the present application.
Detailed Description
In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only partial embodiments of the present application, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
It should be noted that the terms "first," "second," and the like in the description and claims of this application and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It should be understood that the data so used may be interchanged under appropriate circumstances such that embodiments of the application described herein may be implemented in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
Example 1
In order to solve the technical problems that in the related art, when the state of equipment is detected, the calculation of detection data is complex and the efficiency is low through manual analysis, the embodiment provided by the application provides an automatic equipment state detection scheme, which can automatically process and analyze vibration signals after the vibration signals are obtained, extract corresponding characteristic values, and determine the running state of the equipment by inputting the characteristic values into a pre-trained state diagnosis model.
An embodiment of the present application first provides an optional device status detection system, as shown in fig. 1, the system at least includes: target sensor 11, data acquisition device 12, processor 13, display device 14 and communication bus 15, wherein:
and the target sensor 11 is positioned on the target equipment and used for acquiring a vibration signal of the target equipment.
The target device may be a GIS device, and since the vibration signal of the GIS housing is an electromechanical vibration signal, the vibration characteristic thereof is considered comprehensively, and the target sensor 11 may select an acceleration sensor with higher sensitivity.
And the data acquisition device 12 is configured to receive the vibration signal acquired by the target sensor, perform analog-to-digital conversion on the vibration signal to obtain a target vibration signal, and transmit the target vibration signal to the processor 13 through the communication bus 15.
The processor 13 is used for performing singular spectrum analysis on the target vibration signal and determining a target component signal in the target vibration signal; performing time domain analysis and frequency domain analysis on the target component signal to obtain a time domain characteristic value and a frequency domain characteristic value; inputting the time domain characteristic value and the frequency domain characteristic value into a state diagnosis model, and determining the operation state of the target equipment through the state diagnosis model, wherein the state diagnosis model is a decision tree model obtained based on historical operation data training of the target equipment; the time domain characteristic value, the frequency domain characteristic value, and the operation state of the target device are transmitted to the display device 14 through the communication bus 15.
And the display device 14 is used for displaying the time domain characteristic value, the frequency domain characteristic value and the operation state of the target device.
The display device 14 may be a computer display screen of an operation and maintenance worker, or an LED large screen of a monitoring room, and may display vibration characteristic parameters, a defect state detection result, and the like of the target device.
And a communication bus 15 for transmitting data.
In addition to the device status detection system described above, the embodiments of the present application also provide an alternative device status detection method, and it should be noted that the steps shown in the flowchart of the drawings may be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that of the flowchart.
Fig. 2 is a schematic flowchart of an alternative device status detection method according to an embodiment of the present application, and as shown in fig. 2, the method at least includes steps S202-S208, where:
step S202, a target vibration signal of the target device is obtained.
In some optional embodiments of the present application, when obtaining a target vibration signal of a target device, a vibration signal collected by a target sensor located on the target device may be obtained first, and then analog-to-digital conversion is performed on the vibration signal to obtain the target vibration signal.
The target device can be a GIS device, and the target sensor can be an acceleration sensor with higher sensitivity. For example, a data acquisition instrument is used to acquire vibration signals in an acceleration sensor on the GIS device, and a/D analog-to-digital conversion is performed on the vibration signals to obtain conversion data, that is, a target vibration signal.
And step S204, performing singular spectrum analysis on the target vibration signal, and determining a target component signal in the target vibration signal.
Singular spectrum analysis is an effective method for researching nonlinear time series data, and constructs a track matrix according to an observed time series, decomposes and reconstructs the track matrix, thereby extracting signals representing different components of an original time series and analyzing the structure of the time series.
In some optional embodiments of the present application, when performing singular spectrum analysis on a target vibration signal, a first time sequence corresponding to the target vibration signal may be determined first, and a trajectory matrix of the first time sequence may be determined; performing singular value decomposition on the track matrix, and decomposing the track matrix into a first number of elementary matrices; grouping the first number of elementary matrices, summing each group of elementary matrices to obtain a second number of synthetic matrices, and determining a target synthetic matrix from the second number of synthetic matrices; and converting the target synthesis matrix into a second time sequence in a diagonal average calculation mode, and determining a target component signal based on the second time sequence.
Specifically, it is assumed that the first time series corresponding to the target vibration signal is Y T ＝(y 1 ，...，y i ，...y T ) And T is the sequence length, and the following algorithm can be referred when performing singular spectrum analysis on the sequence length:
and S1, embedding. An appropriate window length L, which is an integer satisfying 2 ≦ L ≦ T, may be selected first, usually taking
Then one-dimensional first time sequence Y T Conversion into a multidimensional sequence X 1 ，...，X i ，...X K And X i ＝(y i ，...y i+L -1)∈R L K-L +1, thus yielding the corresponding trajectory matrix:
the trajectory matrix X is a Hankel matrix, i.e., all elements along the anti-diagonal i + j const are identical.
And S2, singular value decomposition. Computing matrix C x ＝XX T To obtain its characteristic value lambda 1 ≥λ 2 ≥…≥λ L Not less than 0 and corresponding feature vector U 1 ，U 2 ，...，U L ，
For the singular spectrum of the trajectory matrix X, a first number d ═ min { L, K } is determined, and the trajectory matrix X may be decomposed into d elementary matrices:
wherein the content of the first and second substances,
singular values, U, of the trajectory matrix X i And V i Are all unit orthogonal matrices, U i Referred to as the left matrix, also known as the Temporal Empirical Orthogonal Function (TEOF),
referred to as the right matrix, also known as Temporal Principal Components (TPC).
Fig. 3 shows a schematic diagram of a signal eigenvalue and a signal eigenvector obtained by processing, and it can be seen that the eigenvalue drops very quickly, that is, the trajectory matrix X is greatly influenced, that is, the elementary matrices corresponding to the eigenvalue and the eigenvector arranged in the front are mainly influenced on the target vibration signal, and the subsequent processing mainly considers the elementary matrices.
And S3, grouping. For a first number d of elementary matrices X i Grouping, summing the elementary matrices in each group to obtain a second number p of synthetic matrices, p being set empirically, and then selecting p synthetic matricesAnd determining a target synthesis matrix. For example, divide {1, 2.. d } into p disjoint subsets { I } 1 ，...，I p I ═ I }, let I ═ I 1 ，...，i n H, the composite matrix X corresponding to I I Can be defined as:
the trajectory matrix X may be represented as p composite matrices X I And (3) the sum:
several elementary matrices, such as the first 4 elementary matrices, which have a large influence on the trajectory matrix X may be grouped in the first group and summed to form the target synthesis matrix.
And S4, reconstructing. For the k-th composite matrix of the p composite matrices
Can be reconstructed into a time sequence by a diagonal average calculation mode
The first time series Y T Can be reconstructed into
Wherein the content of the first and second substances,
the value of (a) is the component signal of the target vibration signal.
The method mainly considers dominant component signals with characteristic difference in different states, so that only the target synthesis matrix needs to be reconstructed to obtain a corresponding second time sequence, and values in the second time sequence are the target component signals. Fig. 4 is a schematic diagram of a processed target component signal.
The diagonal average calculation mode is as follows: let A be an L × K matrix in which the elements are a ij I is not less than 1 and not more than L, j is not less than 1 and not more than K, L ═ min { L, K }, K ═ max { L, K }, and T ═ L + K-1, then weight is givenThe kth element in the constructed time series is:
step S206, time domain analysis and frequency domain analysis are carried out on the target component signal to obtain a time domain characteristic value and a frequency domain characteristic value.
In some optional embodiments of the present application, a time domain analysis may be performed on the target component signal to obtain a time domain characteristic value, where the time domain characteristic value at least includes one of the following values:
1) a mean value representing a time domain average of the target component signal, expressed as:
2) the root-mean-square for representing the amplitude and energy of the target component signal is expressed as:
3) skewness for representing an asymmetric feature of a target component signal, expressed as:
4) for representing the kurtosis of the peak of the target component signal, the expression:
5) a crest factor representing an impact in the target component signal, expressed as:
6) a shape factor representing the shape of the target component signal, expressed as:
wherein X (N) is a time domain discrete signal, N is the number of samples of the signal, X sd Is the standard deviation of the samples of the signal, X pcak Is the peak value of the signal, i.e. max [ x (n)]。
Meanwhile, Fourier transform can be carried out on the target component signal to obtain a target frequency spectrum signal, frequency domain analysis is carried out on the target frequency spectrum signal to obtain a frequency domain characteristic value, and the frequency domain characteristic value at least comprises one of the following values:
1) a spectral centroid representing the centroid of the spectral signal, the expression:
2) spectral diffuseness to represent the diffuseness of a spectral signal, the expression:
3) spectral skewness, which is used to represent the symmetry of a spectral signal, is expressed as:
4) for representing the spectral kurtosis of a transient signal location in a spectral signal, the expression:
5) a spectral crest factor representing an indication of a peak in the spectral signal, expressed as:
6) spectral entropy, which is used to represent the energy distribution of a spectral signal, is expressed as:
wherein s (k) is a frequency domain signal corresponding to the time domain discrete signal after Fourier transform, f k For its corresponding frequency size, b 1 、b 2 To analyze the upper and lower limits of the frequency band.
Step S208, inputting the time domain characteristic value and the frequency domain characteristic value into a state diagnosis model, and determining the operation state of the target device through the state diagnosis model, wherein the state diagnosis model is a decision tree model obtained based on historical operation data training of the target device.
In some alternative embodiments of the present application, the state diagnostic model may be obtained by the following training process: acquiring historical operating data of target equipment, wherein the historical operating data comprises: historical vibration signals and historical operating states corresponding to the historical vibration signals; performing singular spectrum analysis on the historical vibration signal, determining a historical component signal in the historical vibration signal, and performing time domain analysis and frequency domain analysis on the historical component signal to obtain a historical time domain characteristic value and a historical frequency domain characteristic value; and constructing a decision tree model based on a gradient lifting decision tree algorithm, and performing iterative training on the decision tree model based on the historical time domain characteristic value, the historical frequency domain characteristic value and the historical running state to obtain a state diagnosis model.
The method comprises the following steps that a Gradient Boosting Decision Tree (GBDT) algorithm is a machine learning method based on a Decision Tree method and a Gradient Boosting method, wherein the Decision Tree method is used for realizing integrated learning of a weak classifier and improving model precision, and the GBDT algorithm is required to be fitted with Gradient values during iterative training, so that a CART regression Tree is mainly adopted for the Decision Tree; the gradient boosting method can boost the weak classifier into a strong learner with stronger fitting ability. Generally, the constructed decision tree model is composed of a plurality of decision trees, and the final classification result is obtained by accumulating the conclusions of all the decision trees.
Decision tree models typically include three types of nodes: root node, intermediate nodes, and leaf nodes. The root node may be a set of all training data or a starting point for classifying and predicting new data; the intermediate node represents one characteristic or attribute of the analysis object and can be split; the leaf nodes represent the final classification labels, i.e., the categories of data. Fig. 5 shows a schematic diagram of a decision tree model for making decisions on vibration features, where each vibration feature is used as a root node, attributes of the vibration feature are analyzed and classified by decision points (i.e., intermediate nodes), and finally, a final data classification result is labeled at leaf nodes to determine a specific class label.
During specific training, the historical time domain characteristic value and the historical frequency domain characteristic value can be input into a decision tree model, and the predicted operation state of the equipment is obtained through the output of the decision tree model; constructing a target loss function based on the predicted running state and the historical running state; in the iterative training process, model parameters of the decision tree model are optimized by minimizing a target loss function, and a state diagnosis model is obtained.
Specifically, the constructed decision tree model can be represented as:
in the formula, T (x, theta) m ) Is represented by (x, theta) m ) The decision tree is a parameter decision tree, and m represents the number of the decision trees.
GBDT algorithm solves the problem of multi-classification by solving a tree T (x, theta) m ) The usual loss function is:
in the formula, y k Where {0, 1} denotes whether the true value belongs to the k-th class, p k (x) The probability that the decision tree prediction sample x belongs to the kth class is expressed by a softmax function as:
GBDT algorithm determines the optimum of the next decision tree parameter by minimizing the penalty function
The calculation formula of (c) is:
after the trained state diagnostic model is obtained, the time domain characteristic value and the frequency domain characteristic value of the target component signal obtained through processing can be input into the state diagnostic model, and the operating state of the target device is determined through the state diagnostic model.
FIG. 6 is a schematic diagram of a complete device state detection process, in which a classification model is trained by performing signal processing and feature extraction on historical vibration data, building a decision tree based on the GBDT algorithm; and then, carrying out signal processing and feature extraction on the data to be detected, and inputting the data to be detected into the trained classification model to obtain a detection result. FIG. 7 shows a schematic of a state prediction result.
According to the method and the device, the historical vibration data of the equipment are referred in the model training process, so that the established state diagnosis model is high in robustness, has certain recognition capability on early faults and can react more sensitively, and the reliability of the final detection result is high.
In some optional embodiments of the present application, after the operating state of the target device is determined by the state diagnostic model, a time domain characteristic value and a frequency domain characteristic value obtained by performing time domain and frequency domain analysis on the target component signal may be displayed on the display device, and the operating state of the target device is displayed at the same time; if the running state of the target equipment is abnormal, fault early warning information can be automatically generated and used for prompting operation and maintenance personnel that the target equipment is abnormal and needs to be processed.
In the embodiment of the application, after a target vibration signal of target equipment is obtained, singular spectrum analysis is performed on the target vibration signal to determine a target component signal in the target vibration signal, time domain analysis and frequency domain analysis are performed on the target component signal to obtain a time domain characteristic value and a frequency domain characteristic value, the time domain characteristic value and the frequency domain characteristic value are input into a state diagnosis model, the running state of the target equipment is determined through the state diagnosis model, and the state diagnosis model is a decision tree model obtained based on historical running data training of the target equipment. According to the technical scheme, the vibration signals are automatically acquired and processed and analyzed, the analysis efficiency can be improved, meanwhile, the model based on the historical data training of the equipment is utilized for equipment state diagnosis, the accuracy of diagnosis results can be improved, and the technical problems that in the related technology, when the equipment state is detected, the calculation is complex and the efficiency is low due to manual analysis of detection data are solved.
Example 2
According to an embodiment of the present application, there is also provided an apparatus state detection device for implementing the apparatus state detection method, as shown in fig. 8, the device at least includes an obtaining module 81, a first analyzing module 82, a second analyzing module 83, and a determining module 84, where:
the obtaining module 81 is configured to obtain a target vibration signal of a target device.
In some optional embodiments of the present application, when the obtaining module obtains a target vibration signal of the target device, the obtaining module may first obtain a vibration signal collected by a target sensor located on the target device, and then perform analog-to-digital conversion on the vibration signal to obtain the target vibration signal. The target equipment can be GIS equipment, the target sensor can be an acceleration sensor with high sensitivity, the acquisition module can acquire vibration signals in the acceleration sensor on the GIS equipment, and A/D analog-to-digital conversion is carried out on the vibration signals to obtain conversion data, namely the target vibration signals.
And the first analysis module 82 is configured to perform singular spectrum analysis on the target vibration signal to determine a target component signal in the target vibration signal.
When performing singular spectrum analysis on the target vibration signal, the first analysis module may first determine a first time sequence corresponding to the target vibration signal, and determine a trajectory matrix of the first time sequence; performing singular value decomposition on the track matrix, and decomposing the track matrix into a first number of elementary matrices; grouping the first number of elementary matrices, summing each group of elementary matrices to obtain a second number of synthetic matrices, and determining a target synthetic matrix from the second number of synthetic matrices; and converting the target synthesis matrix into a second time sequence in a diagonal average calculation mode, and determining a target component signal based on the second time sequence.
And the second analysis module 83 is configured to perform time domain analysis and frequency domain analysis on the target component signal to obtain a time domain characteristic value and a frequency domain characteristic value.
In some optional embodiments of the present application, a time domain analysis may be performed on the target component signal to obtain a time domain characteristic value, where the time domain characteristic value at least includes one of the following values: a mean value representing a time domain average of the target component signal, a root mean square representing an amplitude and energy of the target component signal, a skewness representing an asymmetric feature of the target component signal, a kurtosis representing a peak of the target component signal, a crest factor representing an impact in the target component signal, a shape factor representing a shape of the target component signal.
Meanwhile, the target component signal can be subjected to Fourier transform to obtain a target frequency spectrum signal, and the target frequency spectrum signal is subjected to frequency domain analysis to obtain a frequency domain characteristic value, wherein the frequency domain characteristic value at least comprises one of the following values: the spectral centroid representing the centroid of the spectral signal, the spectral diffusivity representing the diffusivity of the spectral signal, the spectral skewness representing the symmetry of the spectral signal, the spectral kurtosis representing the transient signal position in the spectral signal, the spectral crest factor representing an index of a peak in the spectral signal, and the spectral entropy representing the energy distribution of the spectral signal.
And the determining module 84 is configured to input the time domain characteristic value and the frequency domain characteristic value into a state diagnostic model, and determine the operating state of the target device through the state diagnostic model, where the state diagnostic model is a decision tree model trained based on historical operating data of the target device.
In some alternative embodiments of the present application, the state diagnostic model may be obtained by the following training process: acquiring historical operating data of target equipment, wherein the historical operating data comprises: historical vibration signals and historical operating states corresponding to the historical vibration signals; performing singular spectrum analysis on the historical vibration signal, determining a historical component signal in the historical vibration signal, and performing time domain analysis and frequency domain analysis on the historical component signal to obtain a historical time domain characteristic value and a historical frequency domain characteristic value; and constructing a decision tree model based on a gradient lifting decision tree algorithm, and performing iterative training on the decision tree model based on the historical time domain characteristic value, the historical frequency domain characteristic value and the historical running state to obtain a state diagnosis model.
During specific training, the historical time domain characteristic value and the historical frequency domain characteristic value can be input into a decision tree model, and the predicted operation state of the equipment is obtained through the output of the decision tree model; constructing a target loss function based on the predicted running state and the historical running state; in the iterative training process, model parameters of the decision tree model are optimized by minimizing a target loss function, and a state diagnosis model is obtained.
Optionally, the device state detection apparatus may further include: the display module 85 is configured to display the processed time domain characteristic value, the processed frequency domain characteristic value, and the operating state of the target device; and generating fault early warning information when the operating state of the target equipment is determined to be abnormal, wherein the fault early warning information is used for prompting that the target equipment needs to be processed if the target equipment is abnormal.
It should be noted that, modules in the device status detection apparatus in this embodiment correspond to implementation steps of the device status detection method in embodiment 1 one to one, and because the detailed description is already given in embodiment 1, details that are not included in this embodiment may refer to embodiment 1, and are not described herein again.
Example 3
According to an embodiment of the present application, there is also provided a nonvolatile storage medium including a stored program, wherein, when the program runs, a device in which the nonvolatile storage medium is located is controlled to execute the device state detection method in embodiment 1.
According to an embodiment of the present application, there is also provided a processor configured to execute a program, where the program executes the device state detection method in embodiment 1.
Optionally, when the program runs, the following steps are implemented: acquiring a target vibration signal of target equipment; performing singular spectrum analysis on the target vibration signal to determine a target component signal in the target vibration signal; performing time domain analysis and frequency domain analysis on the target component signal to obtain a time domain characteristic value and a frequency domain characteristic value; and inputting the time domain characteristic value and the frequency domain characteristic value into a state diagnosis model, and determining the running state of the target equipment through the state diagnosis model, wherein the state diagnosis model is a decision tree model obtained based on historical running data training of the target equipment.
The above-mentioned serial numbers of the embodiments of the present application are merely for description and do not represent the merits of the embodiments.
In the embodiments of the present application, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to the related descriptions of other embodiments.
In the embodiments provided in the present application, it should be understood that the disclosed technology can be implemented in other ways. The above-described embodiments of the apparatus are merely illustrative, and for example, a division of a unit may be a division of a logic function, and an actual implementation may have another division, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or may not be executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, units or modules, and may be in an electrical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a separate product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be substantially implemented or contributed to by the prior art, or all or part of the technical solution may be embodied in a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method of the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a removable hard disk, a magnetic or optical disk, and other various media capable of storing program codes.
The foregoing is only a preferred embodiment of the present application and it should be noted that those skilled in the art can make several improvements and modifications without departing from the principle of the present application, and these improvements and modifications should also be considered as the protection scope of the present application.
Claims (10)
1. An apparatus status detection method, comprising:
acquiring a target vibration signal of target equipment;
performing singular spectrum analysis on the target vibration signal to determine a target component signal in the target vibration signal;
performing time domain analysis and frequency domain analysis on the target component signal to obtain a time domain characteristic value and a frequency domain characteristic value;
and inputting the time domain characteristic value and the frequency domain characteristic value into a state diagnosis model, and determining the operation state of the target equipment through the state diagnosis model, wherein the state diagnosis model is a decision tree model obtained based on historical operation data training of the target equipment.
2. The method of claim 1, wherein obtaining a vibration signal of a target device comprises:
acquiring a vibration signal acquired by a target sensor positioned on the target equipment;
and performing analog-to-digital conversion on the vibration signal to obtain the target vibration signal.
3. The method of claim 1, wherein performing a singular spectrum analysis on the target vibration signal to determine a target component signal in the target vibration signal comprises:
determining a first time sequence corresponding to the target vibration signal, and determining a track matrix of the first time sequence;
performing singular value decomposition on the track matrix, and decomposing the track matrix into a first number of elementary matrices;
grouping the first number of elementary matrices, summing the elementary matrices in each group to obtain a second number of synthetic matrices, and determining a target synthetic matrix from the second number of synthetic matrices;
and converting the target synthesis matrix into a second time sequence in a diagonal average calculation mode, and determining the target component signal based on the second time sequence.
4. The method of claim 1, wherein performing time domain analysis and frequency domain analysis on the target component signal to obtain time domain eigenvalues and frequency domain eigenvalues comprises:
performing time domain analysis on the target component signal to obtain the time domain characteristic value, wherein the time domain characteristic value at least comprises one of the following values: a mean value representing a time domain average of the target component signal, a root mean square representing an amplitude and energy of the target component signal, a skewness representing an asymmetric feature of the target component signal, a kurtosis representing a peak of the target component signal, a crest factor representing a shock in the target component signal, a shape factor representing a shape of the target component signal;
performing fourier transform on the target component signal to obtain a target frequency spectrum signal, and performing frequency domain analysis on the target frequency spectrum signal to obtain the frequency domain characteristic value, where the frequency domain characteristic value at least includes one of the following: a spectral centroid representing a centroid of the spectral signal, a spectral diffuseness representing a diffuseness of the spectral signal, a spectral skewness representing a symmetry of the spectral signal, a spectral kurtosis representing a position of a transient signal in the spectral signal, a spectral crest factor representing an indicator of a peak in the spectral signal, a spectral entropy representing an energy distribution of the spectral signal.
5. The method of claim 1, wherein the training process of the state diagnostic model comprises:
obtaining the historical operating data of the target device, wherein the historical operating data comprises: historical vibration signals and historical operating states corresponding to the historical vibration signals;
performing singular spectrum analysis on the historical vibration signal, determining a historical component signal in the historical vibration signal, and performing time domain analysis and frequency domain analysis on the historical component signal to obtain a historical time domain characteristic value and a historical frequency domain characteristic value;
constructing the decision tree model based on a gradient lifting decision tree algorithm, and performing iterative training on the decision tree model based on the historical time domain characteristic value, the historical frequency domain characteristic value and the historical operating state to obtain the state diagnosis model, wherein the decision tree model comprises: root node, intermediate nodes, and leaf nodes.
6. The method of claim 5, wherein iteratively training the decision tree model based on the historical time-domain feature values, the historical frequency-domain feature values, and the historical operating states comprises:
inputting the historical time domain characteristic value and the historical frequency domain characteristic value into the decision tree model, and outputting through the decision tree model to obtain the predicted operation state of the equipment;
constructing a target loss function based on the predicted operating state and the historical operating state;
and in the iterative training process, optimizing the model parameters of the decision tree model by minimizing a target loss function to obtain the state diagnosis model.
7. The method of claim 1, wherein after determining the operational status of the target device via the status diagnostic model, the method further comprises:
displaying the time domain characteristic value, the frequency domain characteristic value and the running state of the target equipment in display equipment;
and generating fault early warning information when the operating state of the target equipment is determined to be abnormal, wherein the fault early warning information is used for prompting that the target equipment is abnormal and needs to be processed.
8. An apparatus state detection device, comprising:
the acquisition module is used for acquiring a target vibration signal of target equipment;
the first analysis module is used for performing singular spectrum analysis on the target vibration signal and determining a target component signal in the target vibration signal;
the second analysis module is used for carrying out time domain analysis and frequency domain analysis on the target component signal to obtain a time domain characteristic value and a frequency domain characteristic value;
and the determining module is used for inputting the time domain characteristic value and the frequency domain characteristic value into a state diagnosis model and determining the operation state of the target equipment through the state diagnosis model, wherein the state diagnosis model is a decision tree model obtained based on historical operation data training of the target equipment.
9. A device condition detection system, comprising:
the target sensor is positioned on target equipment and used for acquiring a vibration signal of the target equipment;
the data acquisition equipment is used for receiving the vibration signal acquired by the target sensor, performing analog-to-digital conversion on the vibration signal to obtain a target vibration signal, and transmitting the target vibration signal to the processor through a communication bus;
the processor is used for carrying out singular spectrum analysis on the target vibration signal and determining a target component signal in the target vibration signal; performing time domain analysis and frequency domain analysis on the target component signal to obtain a time domain characteristic value and a frequency domain characteristic value; inputting the time domain characteristic value and the frequency domain characteristic value into a state diagnosis model, and determining the operation state of the target equipment through the state diagnosis model, wherein the state diagnosis model is a decision tree model obtained based on historical operation data training of the target equipment; transmitting the time domain characteristic value, the frequency domain characteristic value and the running state of the target equipment to display equipment through a communication bus;
the display device is used for displaying the time domain characteristic value, the frequency domain characteristic value and the running state of the target device;
and the communication bus is used for transmitting data.
10. A non-volatile storage medium, comprising a stored program, wherein a device in which the non-volatile storage medium is located is controlled to perform the device status detection method according to any one of claims 1 to 7 when the program runs.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210543123.4A CN115077685A (en) | 2022-05-18 | 2022-05-18 | Equipment state detection method, device and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210543123.4A CN115077685A (en) | 2022-05-18 | 2022-05-18 | Equipment state detection method, device and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115077685A true CN115077685A (en) | 2022-09-20 |
Family
ID=83248951
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210543123.4A Pending CN115077685A (en) | 2022-05-18 | 2022-05-18 | Equipment state detection method, device and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115077685A (en) |
-
2022
- 2022-05-18 CN CN202210543123.4A patent/CN115077685A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11113905B2 (en) | Fault detection system and method for vehicle system prognosis | |
| CN111504676B (en) | Equipment fault diagnosis method, device and system based on multi-source monitoring data fusion | |
| EP3460611B1 (en) | System and method for aircraft fault detection | |
| CN111507376A (en) | Single index abnormality detection method based on fusion of multiple unsupervised methods | |
| CN106153179A (en) | Medium-speed pulverizer vibrating failure diagnosis method | |
| CN115409131B (en) | Production line abnormity detection method based on SPC process control system | |
| US20220004163A1 (en) | Apparatus for predicting equipment damage | |
| CN113391239B (en) | Mutual inductor anomaly monitoring method and system based on edge calculation | |
| CN111678699B (en) | Early fault monitoring and diagnosing method and system for rolling bearing | |
| CN109061387A (en) | A kind of power grid abnormality method of discrimination based on annulus theorem | |
| CN113901977A (en) | Deep learning-based power consumer electricity stealing identification method and system | |
| GB2581390A (en) | Diagnostic system and a method of diagnosing faults | |
| CN110632484A (en) | ELM-based GIS partial discharge defect diagnosis and classification system and method | |
| CN111881159B (en) | Fault detection method and device based on cost-sensitive extreme random forest | |
| CN114723285A (en) | Power grid equipment safety evaluation prediction method | |
| CN109960232A (en) | The method that the selection method of leading auxiliary parameter and plant maintenance diagnose in advance | |
| CN111259949A (en) | Fault identification model construction method, model and identification method for aircraft environmental control system | |
| CN115587309A (en) | Method, device and equipment for extracting key features of short-circuit resistance of transformer | |
| CN112798290B (en) | Abnormal state monitoring method of gas turbine based on spectrum reconstruction error | |
| CN115077685A (en) | Equipment state detection method, device and system | |
| CN116520068A (en) | Diagnostic method, device, equipment and storage medium for electric power data | |
| CN114580472B (en) | Large-scale equipment fault prediction method with repeated cause and effect and attention in industrial internet | |
| CN115169405A (en) | Hotel guest room equipment fault diagnosis method and system based on support vector machine | |
| CN115902557A (en) | Switch cabinet fault diagnosis processing method and device and nonvolatile storage medium | |
| CN115130731A (en) | Method, equipment and medium for detecting fault of wind driven generator |
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 |