CN113469300B - Equipment state detection method and related device - Google Patents
Equipment state detection method and related device Download PDFInfo
- Publication number
- CN113469300B CN113469300B CN202111035618.8A CN202111035618A CN113469300B CN 113469300 B CN113469300 B CN 113469300B CN 202111035618 A CN202111035618 A CN 202111035618A CN 113469300 B CN113469300 B CN 113469300B
- Authority
- CN
- China
- Prior art keywords
- state
- feature
- sequence
- reconstructed
- reconstruction
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06N3/045—Combinations of networks
-
- 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
Abstract
In the equipment state detection method and the related device, the data processing equipment combines the self-encoder with the plurality of cyclic neural networks, respectively reconstructs a plurality of state feature sets of the equipment to be detected through the self-encoder, and determines the health state of the equipment to be detected according to the difference between the plurality of state feature sets and the respective reconstructed feature sets; the characteristics to be reconstructed of the self-encoder are obtained by respectively carrying out characteristic mining on the plurality of coding sequences with different characteristic scales by the plurality of cyclic neural networks, so that the information which is relevant to the health state of the equipment and is contained in the equipment state sequence can be fully mined, and the aim of improving the detection precision is fulfilled.
Description
Technical Field
The application relates to the field of machine learning, in particular to a device state detection method and a related device.
Background
As the health state of the industrial equipment has important significance for the safety production and the economic benefit of enterprises, the health state monitoring method of the industrial equipment based on data driving is provided. According to the method, a device state sequence of the industrial device is collected by a sensor, modeling of a black box is carried out based on a deep neural network, end-to-end training is carried out, and the health state information of the industrial device, which is contained in mass data, is mined from a space dimension and a time dimension.
However, the inventors have studied to find that the related art does not fully utilize the features in the local perception field of view of the data to be analyzed.
Disclosure of Invention
In order to overcome at least one of the deficiencies in the prior art, an embodiment of the present application provides an apparatus status detection method and a related apparatus, including:
in a first aspect, the present application provides a device status detection method applied to a data processing device, where the data processing device is configured with a first detection model, where the first detection model includes a first encoder, a first decoder, and a plurality of recurrent neural networks, and the method includes:
acquiring a plurality of state feature sets of equipment to be detected, wherein the state feature sets have a preset sequence, and each state feature set has respective state information of a plurality of parts of the equipment to be detected;
respectively extracting features of the plurality of state feature sets from a space dimension through the first encoder to obtain a plurality of encoding sequences with different feature scales, wherein the plurality of encoding sequences respectively correspond to different recurrent neural networks, each encoding sequence comprises a plurality of encoding features with the preset sequence, and the plurality of encoding features respectively correspond to different state feature sets;
for each coding sequence, performing feature extraction on the coding sequence from a time dimension through the corresponding recurrent neural network to obtain a sequence to be reconstructed, wherein the sequence to be reconstructed comprises a plurality of features to be reconstructed, and the plurality of features to be reconstructed respectively correspond to different state feature sets;
inputting all the features to be reconstructed of the state feature set into the first decoder for reconstruction aiming at each state feature set to obtain a reconstruction feature set of the state feature set;
and determining the health state of the equipment to be detected according to the difference between the plurality of state feature sets and the respective reconstruction feature set.
In a second aspect, an embodiment of the present application provides a device status detection apparatus, applied to a data processing device, where the data processing device is configured with a first detection model, where the first detection model includes a first encoder, a first decoder, and a plurality of recurrent neural networks, and the device status detection apparatus includes:
the device comprises a characteristic acquisition module, a state detection module and a state detection module, wherein the characteristic acquisition module is used for acquiring a plurality of state characteristic sets of equipment to be detected, the plurality of state characteristic sets have a preset sequence, and each state characteristic set has respective state information of a plurality of parts of the equipment to be detected;
the characteristic extraction module is used for respectively extracting characteristics of the plurality of state characteristic sets from a space dimension through the first encoder to obtain a plurality of coding sequences with different characteristic scales, wherein the plurality of coding sequences respectively correspond to different recurrent neural networks, each coding sequence comprises a plurality of coding characteristics with the preset sequence, and the plurality of coding characteristics respectively correspond to different state characteristic sets;
the feature extraction module is further configured to, for each coding sequence, perform feature extraction on the coding sequence from a time dimension through the corresponding recurrent neural network to obtain a sequence to be reconstructed, where the sequence to be reconstructed includes a plurality of features to be reconstructed, and the plurality of features to be reconstructed respectively correspond to different status feature sets;
the characteristic reconstruction module is used for inputting all characteristics to be reconstructed of the state characteristic set to the first decoder for reconstruction aiming at each state characteristic set to obtain a reconstruction characteristic set of the state characteristic set;
and the state detection module is used for determining the health state of the equipment to be detected according to the difference between the plurality of state feature sets and the respective reconstruction feature set.
In a third aspect, an embodiment of the present application provides a data processing apparatus, where the data processing apparatus includes a processor and a memory, where the memory stores a computer program, and the computer program, when executed by the processor, implements the apparatus state detection method.
In a fourth aspect, an embodiment of the present application provides a computer storage medium, where a computer program is stored, and when the computer program is executed by a processor, the method for detecting a device state is implemented.
Compared with the prior art, the method has the following beneficial effects:
in the equipment state detection method and the related device for the implementation, the data processing equipment combines the self-encoder with a plurality of cyclic neural networks, respectively reconstructs a plurality of state feature sets of the equipment to be detected through the self-encoder, and determines the health state of the equipment to be detected according to the difference between the plurality of state feature sets and the respective reconstructed feature sets; the characteristics to be reconstructed of the self-encoder are obtained by respectively carrying out characteristic mining on the plurality of coding sequences with different characteristic scales by the plurality of cyclic neural networks, so that the information which is relevant to the health state of the equipment and is contained in the equipment state sequence can be fully mined, and the aim of improving the detection precision is fulfilled.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.
Fig. 1 is a schematic structural diagram of a data processing apparatus according to an embodiment of the present application;
fig. 2 is a schematic flowchart of an apparatus status detection method according to an embodiment of the present disclosure;
fig. 3 is a schematic structural diagram of a first detection model provided in an embodiment of the present application;
fig. 4 is a second schematic flowchart of an apparatus status detection method according to an embodiment of the present application;
fig. 5 is a schematic structural diagram of a second detection model provided in the embodiment of the present application;
fig. 6 is a schematic structural diagram of an apparatus state detection device according to an embodiment of the present application.
Icon: 120-a memory; 130-a processor; 140-a communication device; 201-a feature acquisition module; 202-a feature extraction module; 203-a feature reconstruction module; 204-state detection module.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, 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 some embodiments of the present application, but not all embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. 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: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
In the description of the present application, it is noted that the terms "first", "second", "third", and the like are used merely for distinguishing between descriptions and are not intended to indicate or imply relative importance. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
It should be understood that the operations of the flow diagrams may be performed out of order, and steps without logical context may be performed in reverse order or simultaneously. One skilled in the art, under the guidance of this application, may add one or more other operations to, or remove one or more operations from, the flowchart.
In the related technology, a device state sequence acquired by a sensor of industrial equipment is utilized, modeling of a black box is carried out based on a deep neural network, end-to-end training is carried out, and the health state information of the industrial equipment, which is contained in mass data, is mined from a space dimension and a time dimension.
However, in the process of extracting features based on the device state sequence, the deep neural network focuses more on feature information in the global visual field range, and ignores feature information in the local perception visual field range, so that the information related to the health state of the device, which is hidden in the device state sequence, is not fully explored.
In view of this, the present application provides a device status detection method applied to a data processing device. In the method, the data processing equipment combines a self-encoder with a plurality of cyclic neural networks, reconstructs a plurality of state feature sets of equipment to be detected respectively through the self-encoder, and determines the health state of the equipment to be detected according to the difference between the plurality of state feature sets and the respective reconstructed feature sets; the characteristics to be reconstructed of the self-encoder are obtained by respectively carrying out characteristic mining on the plurality of coding sequences with different characteristic scales through the plurality of recurrent neural networks, so that the information which is relevant to the health state of the equipment and is stored in the equipment state sequence can be fully mined.
The equipment state detection method is suitable for different types of equipment to be detected. For example, the device to be detected can be an industrial robot, a machine tool, a gate control device, a motor car, an unmanned aerial vehicle and the like.
And the data processing equipment can select different types of equipment according to the types of the equipment to be detected. In some embodiments, the data processing device may be a server communicatively connected to the device to be detected. The type of the server may be, but is not limited to, a Web server, an FTP (File Transfer Protocol) server, a data processing server, and the like. In addition, the server may be a single server or a server group. The set of servers can be centralized or distributed (e.g., the servers can be a distributed system). In some embodiments, the server may be local or remote to the user terminal. In some embodiments, the server may be implemented on a cloud platform; by way of example only, the Cloud platform may include a private Cloud, a public Cloud, a hybrid Cloud, a Community Cloud, a distributed Cloud, a cross-Cloud (Inter-Cloud), a Multi-Cloud (Multi-Cloud), and the like, or any combination thereof. In some embodiments, the server may be implemented on an electronic device having one or more components.
In some other embodiments, the data processing device may also be a user terminal communicatively connected to the device to be detected. The specific type of the user terminal may be, but is not limited to, a mobile terminal, a tablet computer, a laptop computer, or a built-in device in a motor vehicle, etc., or any combination thereof. In some embodiments, the mobile terminal may include a smart home device, a wearable device, a smart mobile device, a virtual reality device, an augmented reality device, or the like, or any combination thereof. In some embodiments, the smart home devices may include smart lighting devices, control devices for smart electrical devices, smart monitoring devices, smart televisions, smart cameras, or walkie-talkies, or the like, or any combination thereof. In some embodiments, the wearable device may include a smart bracelet, a smart lace, smart glass, a smart helmet, a smart watch, a smart garment, a smart backpack, a smart accessory, and the like, or any combination thereof. In some embodiments, the smart mobile device may include a smartphone, a Personal Digital Assistant (PDA), a gaming device, a navigation device, or a Point of Sale (POS) device, or the like, or any combination thereof.
The embodiment also provides a structural schematic diagram of the data processing device. As shown in fig. 1, the data processing apparatus includes a memory 120, a processor 130, and a communication device 140.
The memory 120, processor 130, and communication device 140 are electrically connected to each other directly or indirectly to enable data transmission or interaction. For example, the components may be electrically connected to each other via one or more communication buses or signal lines.
The Memory 120 may be, but is not limited to, a Random Access Memory (RAM), a Read Only Memory (ROM), a Programmable Read-Only Memory (PROM), an Erasable Read-Only Memory (EPROM), an electrically Erasable Read-Only Memory (EEPROM), and the like. The memory 120 is used for storing a program, and the processor 130 executes the program after receiving the execution instruction.
The communication device 140 is used for transmitting and receiving data through a network. The Network may include a wired Network, a Wireless Network, a fiber optic Network, a telecommunication Network, an intranet, the internet, a Local Area Network (LAN), a Wide Area Network (WAN), a Wireless Local Area Network (WLAN), a Metropolitan Area Network (MAN), a Wide Area Network (WAN), a Public Switched Telephone Network (PSTN), a bluetooth Network, a ZigBee Network, or a Near Field Communication (NFC) Network, or any combination thereof. In some embodiments, the network may include one or more network access points. For example, the network may include wired or wireless network access points, such as base stations and/or network switching nodes, through which one or more components of the service request processing system may connect to the network to exchange data and/or information.
The processor 130 may be an integrated circuit chip having signal processing capabilities, and may include one or more processing cores (e.g., a single-core processor or a multi-core processor). Merely by way of example, the Processor may include a Central Processing Unit (CPU), an Application Specific Integrated Circuit (ASIC), an Application Specific Instruction Set Processor (ASIP), a Graphics Processing Unit (GPU), a Physical Processing Unit (PPU), a Digital Signal Processor (DSP), a Field Programmable Gate Array (FPGA), a Programmable Logic Device (PLD), a controller, a microcontroller Unit, a Reduced Instruction Set computer (Reduced Instruction Set computer), a microprocessor, or the like, or any combination thereof.
Based on the above-mentioned related description about the data processing apparatus, the apparatus state detection method performed in the data processing apparatus is explained in detail below. In this embodiment, a neural network model in which an auto-encoder is combined with a plurality of recurrent neural networks is referred to as a first detection model, and the auto-encoder includes a first encoder and a first decoder. Fig. 2 is a schematic flow chart of the method for detecting the status of the device, and the steps of the method are described in detail below with reference to fig. 2. As shown in fig. 2, the method includes:
step S101, a plurality of state feature sets of the device to be detected are obtained.
The plurality of status feature sets have a preset sequence, and each status feature set has respective status information of a plurality of components of the equipment to be detected.
In this example, it is considered that the multiple components need to be matched with each other when the device to be detected works, and therefore, the state data of the multiple components have a specific linkage relationship, and the linkage relationship is embodied in a data distribution manner. Based on the discovery, the linkage relation contained in the state data corresponding to the multiple components is discovered through the first detection model, and the linkage relation is used for determining the health condition of the device to be detected.
As an implementation, the data processing apparatus acquires a first state sequence of the device under test, in which the first state sequence includes respective state data of the plurality of components of the device under test, considering that the state data of the plurality of components have a high-dimensional feature space, which requires a high computational complexity.
The data processing device splits the first state sequence into a plurality of data segments; acquiring a covariance matrix of the data segments aiming at each data segment; and finally, taking the covariance matrix of all the data segments as a plurality of state feature sets of the equipment to be detected.
The industrial robot described above is exemplified as the device to be tested. It is assumed that the industrial robot comprises 6 degrees of freedom, one for each drive motor. The data processing device periodically and synchronously acquires the states of the 6 driving motors through the sensor to obtain the respective state data of the 6 driving motors, which can be expressed as:
……
in the formula (I), the compound is shown in the specification,
representing the state sequence of the first drive motor,
showing the first drive motor in
Status data that is collected at a moment; the representation manner of the state sequence corresponding to other driving motors is the same as that of the first driving motor, and the description thereof is omitted in this embodiment.
The type of the state data collected in this example may include at least one of current, rotation speed, rotation angle, and the like when the driving motor operates; the types of the state data may be the same or partially the same between the drive motors. For example, the type of the status data of the first driving motor may include current, rotation speed, rotation angle; and the type of the state data of the 2 nd driven motor may include current, rotation speed. Therefore, those skilled in the art can adapt the method according to the needs, and the embodiments of the present application are not limited specifically.
Then, for each driving motor, the data processing device may intercept the state sequence of the driving motor through a time window to obtain a plurality of sequence segments of the driving motor. It should be understood that, compared with the method of performing average segmentation on the state sequence, the method of performing truncation on the state sequence in the embodiment performs truncation on the state sequence in a sliding window manner, so that the truncated sequence segments are relatively continuous, data distribution is prevented from being affected by truncation scale, and more sequence segments can be obtained.
In addition, in order to keep the dimensions of the sequence segments the same between different driving motors, a sliding window with the same dimension may be selected for each driving motor in the embodiment. Finally, for the sequence segments of all the drive motors, the data processing device classifies the sequence segments of the same sequence position into one class to obtain a plurality of data segments.
Assuming that the scale of the sliding window is 3 state data and the step length of the sliding window is 2 state data, the first data segment can be represented as:
……
the second piece of data may be represented as:
……
by analogy, other data segments can be obtained, and the embodiment does not need to be complained. In the formula (I), the compound is shown in the specification,
a first sequence segment representing a first drive motor,
a second sequence segment representing a first drive motor;
a first sequence segment representing a second drive motor,
a second sequence segment representing a second drive motor; the meanings of other sequence segments can be obtained by analogy, and the description is omitted in this embodiment.
Therefore, each of the data segments includes the respective status data of the plurality of driving motors. In order to facilitate the first detection model to find out the health state information contained in the first detection model, the data processing device uses a covariance matrix between sequence segments in the data segments as a state feature set. It should be understood that the covariance matrix carries the state information of each drive motor, including the autocorrelation between individual drive motor state data and the correlation between multiple drive motors.
Step S102, respectively extracting the features of the plurality of state feature sets from the space dimension through a first encoder to obtain a plurality of coding sequences with different feature scales.
The coding sequences respectively correspond to different recurrent neural networks, each coding sequence comprises a plurality of coding features with a preset sequence, and the coding features respectively correspond to different state feature sets.
As a possible implementation, the first encoder comprises a plurality of convolutional layers. For each state feature set, the data processing equipment sequentially performs feature extraction on the state feature set through a plurality of convolution layers to obtain a plurality of coding features of the state feature set; the plurality of coding features of the state feature set have different feature scales respectively and are obtained from different convolutional layers respectively.
And finally, the data processing equipment classifies the coding features of the plurality of state feature sets according to feature scales to obtain a plurality of coding sequences.
The exemplary description continues with the industrial robot example described above. As shown in fig. 3, the encoder comprises 4 convolutional layers, and it is assumed that the number of state feature sets of the industrial robot is 5. For convenience of description, the 5 status feature sets are respectively represented as status feature set a, status feature set B, status feature set C, status feature set D, and status feature set E. For each set of status features, the data processing device inputs it to a plurality of convolutional layers connected in series in turn to obtain 4 coding features of the set of status features.
Taking the state feature set a as an example, in the process that the data processing device processes the state feature set a through a plurality of convolutional layers of the encoder, not only the coding features output by each convolutional layer are input to the next adjacent convolutional layer, but also the coding features output by each convolutional layer are copied to construct a plurality of coding sequences.
Therefore, the data processing device can obtain 4 different feature scale coding features through the state feature set a, that is, the above 5 state feature sets can obtain 5 × 4=20 coding features in total.
Because different feature scales correspond to different convolutional layers respectively, the data processing equipment classifies the 20 coding features according to the feature scales, and divides the coding features from the same convolutional layer into one class; thus, 4 encoding sets are obtained, each comprising 5 encoding features, each corresponding to a different set of status features.
And the preset sequence is considered to exist among different state feature sets, and the preset sequence corresponds to the interception sequence of the data segments and also reflects the acquisition sequence of the state data. Therefore, for each code set, the data processing device sorts the code features in the order of the state feature set corresponding to the code features to obtain corresponding code sequences, that is, 4 code sequences can be obtained by 4 code sets.
And S103, performing feature extraction on the coding sequences from a time dimension through a corresponding recurrent neural network aiming at each coding sequence to obtain a sequence to be reconstructed.
The sequence to be reconstructed comprises a plurality of features to be reconstructed, and the plurality of features to be reconstructed respectively correspond to different state feature sets.
Illustratively, the plurality of recurrent neural networks may be 4 LSTM (Long Short-term Memory) networks in fig. 3. And for each coding sequence, the data processing equipment sequentially inputs 5 coding features in the coding sequence into a corresponding LSTM network according to a preset sequence to obtain 5 features to be reconstructed. Wherein, the 5 characteristics to be reconstructed correspond to different status characteristic sets of the industrial robot respectively.
And step S104, inputting all the characteristics to be reconstructed of the state characteristic set to a first decoder for reconstruction aiming at each state characteristic set, and obtaining a reconstruction characteristic set of the state characteristic set.
In one implementation, the first decoder includes a plurality of deconvolution layers corresponding to the plurality of recurrent neural networks, respectively. The data processing equipment respectively inputs all the characteristics to be reconstructed of the state characteristic set into corresponding deconvolution layers for reconstruction according to the corresponding relation between the plurality of cyclic neural networks and the plurality of deconvolution layers, and the reconstruction characteristic set of the state characteristic set is obtained.
Illustratively, continuing with the first decoder in fig. 3, there are 4 deconvolution layers in series, one for each different LSTM network. Taking the state feature set a of the industrial robot as an example, the 4 LSTM networks respectively output one feature to be reconstructed of the state feature set a, that is, a total of 4 features to be reconstructed are obtained. The data processing equipment respectively inputs 4 characteristics to be reconstructed into corresponding deconvolution layers for reconstruction so as to obtain a reconstruction characteristic set corresponding to the state characteristic set A. Similarly, the status feature set B, the status feature set C, the status feature set D, and the status feature set E can all obtain their respective reconstruction feature sets.
And in order to keep the characteristics from losing, the data processing equipment splices the two and inputs the spliced two into the next deconvolution layer. As shown in fig. 3, the input features of the deconvolution layer include the output features of the adjacent last deconvolution layer and the feature to be reconstructed corresponding to the LSTM network output.
And S105, determining the health state of the equipment to be detected according to the difference between the plurality of state feature sets and the respective reconstruction feature set.
It should be noted that the first detection model is obtained by training a sample state feature set of the device to be detected during normal operation in advance, and therefore, assuming that the device to be detected is not abnormal, the first detection model can respectively reconstruct the plurality of state features, so that the difference between the plurality of state feature sets and the respective reconstructed feature set does not exceed a set first threshold; on the contrary, if the equipment to be detected is abnormal, the first detection model is difficult to respectively reconstruct the plurality of state characteristics; the differences between the plurality of status feature sets and the respective reconstruction feature set are made to be greater than a set first threshold.
Thus, as one implementation, the data processing apparatus obtains a first mean square error between a plurality of sets of state features and respective sets of reconstruction features; and if the first mean square error is larger than the first threshold value, the equipment to be detected is abnormal.
Based on the design, the data processing equipment combines the self-encoder with a plurality of cyclic neural networks, reconstructs a plurality of state feature sets of the equipment to be detected respectively through the self-encoder, and determines the health state of the equipment to be detected according to the difference between the plurality of state feature sets and the respective reconstructed feature sets; the characteristics to be reconstructed of the self-encoder are obtained by respectively carrying out characteristic mining on the plurality of coding sequences with different characteristic scales by the plurality of cyclic neural networks, so that the information which is relevant to the health state of the equipment and is contained in the equipment state sequence can be fully mined, and the aim of improving the detection precision is fulfilled.
In addition, it is worth to be noted that most of the state data collected from the industrial equipment belongs to high-frequency time series data, and the cost of the labeled data is very high, so that, for the characteristics of the state data of the industrial equipment, the first detection model provided by the embodiment can detect the health state of the industrial equipment in a self-supervision manner.
In this embodiment, not only the state of the entire device to be detected is detected, but also each component is detected. The following is a detailed description of any one target component selected from a plurality of components, by way of example. In this embodiment, for example, to detect the state of the target component, the data processing apparatus is further configured with a second detection model for the target component, where the model includes a second encoder and a second decoder.
As shown in fig. 4, based on the second detection model, the device status detection method further includes:
in step S106, a second state sequence of the target component is acquired.
Wherein the second state sequence includes state data of the target component. Illustratively, continuing with the example of the industrial robot, assuming that one of the driving motors of the industrial robot is used as the target component, the data of the working state of the driving motor may include any one of current, rotation speed and rotation angle. Assuming that the type of the state data is current, the data processing device periodically collects the current of the driving motor during working, and intercepts the collected current data by adopting a sliding window to obtain a second state sequence of the driving motor.
In order to better analyze the distribution of the state data corresponding to the target component, in this embodiment, the state data is obtained by normalizing the original state data of the target component.
As an alternative to normalization, the data can be normalized using the z-score (zero-mean normalization) method. In the method, the original shape data is scaled to make the scaled data fall in the interval with the mean value of 0 and the standard deviation of 1. The expression for the z-score method is as follows:
in the formula (I), wherein
Is shown as
The data of the original state is stored in a memory,
represents the total amount of raw state data,
is an average value of the raw state data,
is the standard deviation of the raw state data,
indicating normalized status data.
It should be noted that after z-score standardization is performed on the original state data, the distribution request of the data itself is not changed, but the state data distribution intervals after standardization processing are basically consistent and mainly distributed in [ -2, 2] intervals; so that the following algorithms can focus more on analyzing the distribution of the state data itself.
And step S107, performing feature extraction on the second state sequence through a second encoder to obtain a first feature to be reconstructed.
In this embodiment, as shown in fig. 5, in order to extract the feature information of the state data in the time dimension, the second encoder may select an LSTM network and a PCA (Principal Component Analysis) model, and the second decoder may select the LSTM network.
The working principle of the LSTM network is to input the time sequence of the state number of the target component in turn, and then to update the hidden state of the target component, which is expressed as
At the last step of a time series (denoted as
) The LSTM network contains all the information of the previous sequence (also called context vector), i.e.
Therefore, the method can be used for exploring characteristic information in data from a time dimension.
Step S108, acquiring a second feature to be reconstructed of the second state sequence.
The second feature to be reconstructed includes one or a combination of a time domain feature, a frequency domain feature and a time-frequency domain feature of the second state sequence.
Time domain characteristics:
may include a valid value
Square root amplitude
Peak to peak value
Crest factor
Margin index
Deviation index
Kurtosis index
Form factor of
Pulse factor of
And entropy of information
. Splicing the 10 parameters together can obtain a time domain feature vector
The corresponding expression is:
in the following, the above 10 parameters will be explained in detail by taking the second state sequence of the industrial robot as an example, again in connection with the above industrial robot. Since the second state sequence of the industrial robot comprises current data while the industrial robot is working, the meaning of the parameters is as follows:
the effective value, i.e. the root mean square value of the current data, mainly describes the effective power of the current.
And the square root amplitude is used for describing the total amplitude of the current vibration and reflecting the real level of the current vibration.
And the peak-to-peak value represents the difference between the maximum value and the minimum value of the current data in the second state sequence and is mainly used for describing the span range of the current.
Crest factor, which is used to describe the surge condition existing in the current.
And the margin index is used for describing the abrasion condition of the current corresponding to the machine and is sensitive to impact type faults.
The skewness index and the statistical average of the third moment of the vibration signal are mainly used for describing the magnitude of asymmetry of the current.
The kurtosis index and the fourth moment statistical average of the current data are mainly used for describing the impact magnitude of the current.
And the form factor is used for representing the original shape property of the current data waveform and is independent of the amplitude.
The pulse factor, which is used to represent the impact of the current, may be complementary to the kurtosis index, although it is less sensitive than the kurtosis index.
And the information entropy is used for describing the uncertainty degree of the current.
Frequency domain characteristics:
according to the Barseval theorem (Parseval's theorem): whether a real signal or a complex signal (i.e., the state data of the target component in this embodiment), the integral of the square of the signal amplitude is equal to the square of the modulus of the signal where the energy of the signal is equal to the spectral density of the signal. The corresponding expression can be expressed as:
wherein the content of the first and second substances,
which is indicative of the energy of the signal,
a threshold value at which the signal is represented,
representing a signal frequency threshold. Therefore, in this embodiment, the data processing device first obtains the spectrogram of the second state sequence by a Fast Fourier Transform (FFT) method, and then integrates the spectrogram by taking the frequency axis of the spectrogram as the time axis. Thus, the frequency domain characteristics of the second state sequence are obtained
。
Time-frequency domain characteristics:
the data processing equipment performs time-frequency analysis on the second state sequence by using an EMD (Empirical Mode Decomposition) method and an STFT (short-time Fourier Transform) method to obtain the time-frequency domain characteristics.
Firstly, the data processing equipment obtains the correlation between the second state sequence and the target component fault through an EMD methodnIMF (Intrinsic Mode Function); then, the screened sample was subjected to EMDnAnd (4) respectively taking 4 characteristic values of energy, variance, skewness index and kurtosis index by each IMF, and finally obtaining 2 characteristic values of standard deviation of instantaneous frequency and signal-to-noise ratio of the instantaneous frequency by using an STFT method.
Splicing the characteristic values to obtain 4n+ 2D time-frequency domain feature vector
The corresponding expression is:
in the formula (I), the compound is shown in the specification,
to representnThe energy of the individual IMFs is,
to representnThe variance of the individual IMFs is determined,
identificationnThe deviation index of the IMF is determined,
to representnThe kurtosis index of each IMF,
the standard deviation of the instantaneous frequency is such that,
representing the signal-to-noise ratio of the instantaneous frequency.
Step S109, inputting the splicing feature of the first feature to be reconstructed and the second feature to be reconstructed into a second decoder for reconstruction, so as to obtain a reconstruction sequence of the second state sequence.
Fig. 5 is a schematic diagram of a possible structure of the second detection model. The data processing equipment inputs the second state sequence into an LSTM coding network and a PCA model to obtain a first feature to be reconstructed; taking the time domain characteristics, the frequency domain characteristics and the time-frequency domain characteristics as second characteristics to be reconstructed; and inputting the first feature to be reconstructed and the second feature to be reconstructed into the LSTM decoding network to obtain a reconstruction sequence of the second state sequence.
Step S110, determining the health status of the target component according to the difference between the second state sequence and the reconstructed sequence.
It should be noted that, similar to the first detection model, the second detection model is also obtained by training the sample state sequence of the target component during normal operation in advance, and therefore, assuming that the target device is not abnormal, the second detection model can reconstruct the second state sequence, so that a second mean square error between the second state sequence and the reconstructed sequence is not greater than a second threshold; on the contrary, if the target component is abnormal, the second detection model is difficult to reconstruct the second state sequence; such that a second mean squared error between the second sequence of states and the reconstructed sequence is greater than a second threshold.
In addition, the first detection model and the second detection model are complementary to each other, and when any one detection model detects an abnormality, the equipment to be detected is regarded as abnormal.
In this embodiment, in consideration of the influence of the error factor, a threshold range of the second threshold is provided to improve the fault tolerance, and the expression is as follows:
in the formula (I), the compound is shown in the specification,
a sequence of states of the sample is represented,
a reconstructed sequence representing a sequence of sample states,
the mean variance of the two is shown,
the standard deviation between the two is shown,
representing a range of the second threshold. When the mean square error between the second state sequence and the reconstruction sequence is within the range, the target component is normal, otherwise, the target component is abnormal.
And when the abnormality of the target component is detected, calculating an abnormality-at-occurrence score by the following expression
For measuring the severity of the abnormality:
in the formula (I), the compound is shown in the specification,
representing the mean square error between the second state sequence and the reconstructed sequence,
is a function of the sign when
When less than 0, the value of the function is 1, when
When equal to 0, the value of the function is 0, when
Above 0, the value of this function is 1.
Based on the same inventive concept as the device status detection method, the embodiment also provides a related device of the method, which comprises:
the present embodiment further provides an apparatus state detection device, which is applied to a data processing apparatus, wherein the apparatus state detection device includes at least one functional module that can be stored in a memory in a software form. As shown in fig. 6, the device status detection apparatus may include, functionally divided:
the characteristic acquisition module 201 is configured to acquire a plurality of status feature sets of the device to be detected, where the plurality of status feature sets have a preset sequence, and each status feature set has respective status information of a plurality of components of the device to be detected.
In this embodiment, the feature obtaining module 201 is configured to implement step S101 in fig. 2, and for a detailed description of the feature obtaining module 201, refer to a detailed description of step S101.
The feature extraction module 202 is configured to perform feature extraction on the plurality of status feature sets from a spatial dimension through the first encoder, so as to obtain a plurality of coding sequences with different feature scales, where the plurality of coding sequences correspond to different recurrent neural networks, each coding sequence includes a plurality of coding features in a preset order, and the plurality of coding features correspond to different status feature sets.
The feature extraction module 202 is further configured to, for each coding sequence, perform feature extraction on the coding sequence from a time dimension through a corresponding recurrent neural network to obtain a sequence to be reconstructed, where the sequence to be reconstructed includes a plurality of features to be reconstructed, and the plurality of features to be reconstructed correspond to different state feature sets respectively.
In this embodiment, the feature extraction module 202 is configured to implement steps S102 to S103 in fig. 2, and for a detailed description of the feature extraction module 202, refer to a detailed description of steps S102 to S103.
And the feature reconstruction module 203 is configured to, for each state feature set, input all features to be reconstructed of the state feature set to the first decoder for reconstruction, and obtain a reconstruction feature set of the state feature set.
In this embodiment, the feature reconstruction module 203 is configured to implement step S104 in fig. 2, and for a detailed description of the feature reconstruction module 203, refer to a detailed description of step S104.
And the state detection module 204 is configured to determine a health state of the device to be detected according to differences between the plurality of state feature sets and the respective reconstructed feature sets.
In this embodiment, the state detection module 204 is configured to implement step S105 in fig. 2, and for a detailed description of the state detection module 204, refer to a detailed description of step S105.
It should be noted that the device state detection apparatus may further include other software functional modules, which are used to implement other steps or sub-steps of the device state detection method; of course, the feature obtaining module 201, the feature extracting module 202, the feature reconstructing module 203, and the state detecting module 204 may also be used to implement other steps or sub-steps of the device state detecting method; those skilled in the art can make appropriate adjustments according to different module division standards, and the embodiment is not limited specifically.
The embodiment also provides a data processing device, which comprises a processor and a memory, wherein the memory stores a computer program, and the computer program is executed by the processor to realize the device state detection method.
The embodiment also provides a computer storage medium, which stores a computer program, and when the computer program is executed by a processor, the method for detecting the device state is realized.
To sum up, in the device state detection method and the related apparatus provided in the embodiment of the present application, the data processing device combines the self-encoder with the plurality of recurrent neural networks, reconstructs a plurality of state feature sets of the device to be detected respectively through the self-encoder, and determines the health state of the device to be detected according to the differences between the plurality of state feature sets and the respective reconstructed feature sets; the characteristics to be reconstructed of the self-encoder are obtained by respectively carrying out characteristic mining on the plurality of coding sequences with different characteristic scales by the plurality of cyclic neural networks, so that the information which is relevant to the health state of the equipment and is contained in the equipment state sequence can be fully mined, and the aim of improving the detection precision is fulfilled.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The apparatus embodiments described above are merely illustrative, and for example, the flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
In addition, functional modules in the embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.
The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including 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 according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above description is only for various embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of changes or substitutions within the technical scope of the present application, and all such changes or substitutions are included in the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (10)
1. A device status detection method applied to a data processing device configured with a first detection model including a first encoder, a first decoder, and a plurality of recurrent neural networks, the method comprising:
acquiring a plurality of state feature sets of equipment to be detected, wherein the state feature sets have a preset sequence, and each state feature set has respective state information of a plurality of parts of the equipment to be detected;
respectively extracting features of the plurality of state feature sets from a space dimension through the first encoder to obtain a plurality of encoding sequences with different feature scales, wherein the plurality of encoding sequences respectively correspond to different recurrent neural networks, each encoding sequence comprises a plurality of encoding features with the preset sequence, and the plurality of encoding features respectively correspond to different state feature sets;
for each coding sequence, performing feature extraction on the coding sequence from a time dimension through the corresponding recurrent neural network to obtain a sequence to be reconstructed, wherein the sequence to be reconstructed comprises a plurality of features to be reconstructed, and the plurality of features to be reconstructed respectively correspond to different state feature sets;
inputting all the features to be reconstructed of the state feature set into the first decoder for reconstruction aiming at each state feature set to obtain a reconstruction feature set of the state feature set;
and determining the health state of the equipment to be detected according to the difference between the plurality of state feature sets and the respective reconstruction feature set.
2. The apparatus status detecting method according to claim 1, wherein the first encoder includes a plurality of convolutional layers, and the obtaining of the plurality of encoded sequences with different feature scales by the first encoder through feature extraction of the plurality of status feature sets from a spatial dimension comprises:
for each state feature set, sequentially performing feature extraction on the state feature set through the plurality of convolutional layers to obtain a plurality of coding features of the state feature set, wherein the coding features of the state feature set have different feature scales and are obtained from the different convolutional layers respectively;
and classifying the coding features of the plurality of state feature sets according to feature scales to obtain the plurality of coding sequences.
3. The device state detection method according to claim 1, wherein the first decoder includes a plurality of deconvolution layers respectively corresponding to the plurality of recurrent neural networks, and the obtaining of the reconstruction feature set of the state feature set by inputting all features to be reconstructed of the state feature set to the first decoder for reconstruction for each of the state feature sets includes:
and according to the corresponding relation between the plurality of cyclic neural networks and the plurality of deconvolution layers, respectively inputting all the characteristics to be reconstructed of the state characteristic set into the corresponding deconvolution layers for reconstruction, and obtaining the reconstruction characteristic set of the state characteristic set.
4. The apparatus state detection method according to claim 1, wherein the acquiring a plurality of state feature sets of the apparatus to be detected comprises:
acquiring a first state sequence of the device to be detected, wherein the first state sequence comprises respective state data of a plurality of parts of the device to be detected;
splitting the first state sequence into a plurality of data segments;
for each data segment, acquiring a covariance matrix of the data segment;
and taking the covariance matrixes of all the data segments as a plurality of state feature sets of the equipment to be detected.
5. The device state detection method of claim 1, wherein determining the health state of the device under test based on the differences between the plurality of state feature sets and the respective reconstructed feature sets comprises:
acquiring a first mean square error between the plurality of state feature sets and respective reconstruction feature sets;
and if the first mean square error is larger than a first threshold value, the equipment to be detected is abnormal.
6. The device state detection method of claim 1, wherein the data processing device is further configured with a second detection model of a target component, the second detection model comprising a second encoder and a second decoder, the method further comprising:
obtaining a second state sequence of the target component, wherein the second state sequence comprises state data of the target component;
performing feature extraction on the second state sequence through the second encoder to obtain a first feature to be reconstructed;
acquiring a second feature to be reconstructed of the second state sequence, wherein the second feature to be reconstructed comprises one or a combination of a time domain feature, a frequency domain feature and a time-frequency domain feature of the second state sequence, the frequency domain feature is obtained by integrating a spectrogram of the second state sequence, and the time-frequency domain feature is obtained by performing time-frequency analysis on the second state sequence by an EMD (empirical mode decomposition) and STFT (standard time-frequency transformation) method;
inputting the splicing feature of the first feature to be reconstructed and the second feature to be reconstructed into the second decoder for reconstruction, so as to obtain a reconstruction sequence of the second state sequence;
determining a health state of the target component based on a difference between the second state sequence and the reconstructed sequence.
7. The device state detection method of claim 6, wherein the determining the health state of the target component from the difference between the second state sequence and the reconstruction sequence comprises:
obtaining a second mean square error between the second state sequence and the reconstructed sequence;
and if the second mean square error is larger than a second threshold value, the target component has an abnormality.
8. A device status detection apparatus applied to a data processing device provided with a first detection model including a first encoder, a first decoder, and a plurality of recurrent neural networks, the device status detection apparatus comprising:
the device comprises a characteristic acquisition module, a state detection module and a state detection module, wherein the characteristic acquisition module is used for acquiring a plurality of state characteristic sets of equipment to be detected, the plurality of state characteristic sets have a preset sequence, and each state characteristic set has respective state information of a plurality of parts of the equipment to be detected;
the characteristic extraction module is used for respectively extracting characteristics of the plurality of state characteristic sets from a space dimension through the first encoder to obtain a plurality of coding sequences with different characteristic scales, wherein the plurality of coding sequences respectively correspond to different recurrent neural networks, each coding sequence comprises a plurality of coding characteristics with the preset sequence, and the plurality of coding characteristics respectively correspond to different state characteristic sets;
the feature extraction module is further configured to, for each coding sequence, perform feature extraction on the coding sequence from a time dimension through the corresponding recurrent neural network to obtain a sequence to be reconstructed, where the sequence to be reconstructed includes a plurality of features to be reconstructed, and the plurality of features to be reconstructed respectively correspond to different status feature sets;
the characteristic reconstruction module is used for inputting all characteristics to be reconstructed of the state characteristic set to the first decoder for reconstruction aiming at each state characteristic set to obtain a reconstruction characteristic set of the state characteristic set;
and the state detection module is used for determining the health state of the equipment to be detected according to the difference between the plurality of state feature sets and the respective reconstruction feature set.
9. A data processing device comprising a processor and a memory, the memory storing a computer program which, when executed by the processor, implements the device status detection method of any one of claims 1 to 7.
10. A computer storage medium, characterized in that the computer storage medium stores a computer program which, when executed by a processor, implements the device status detection method of any one of claims 1 to 7.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111035618.8A CN113469300B (en) | 2021-09-06 | 2021-09-06 | Equipment state detection method and related device |
| PCT/CN2021/131934 WO2023029229A1 (en) | 2021-09-06 | 2021-11-19 | Device state detection method and related apparatus |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111035618.8A CN113469300B (en) | 2021-09-06 | 2021-09-06 | Equipment state detection method and related device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113469300A CN113469300A (en) | 2021-10-01 |
| CN113469300B true CN113469300B (en) | 2021-12-07 |
Family
ID=77867475
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111035618.8A Active CN113469300B (en) | 2021-09-06 | 2021-09-06 | Equipment state detection method and related device |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN113469300B (en) |
| WO (1) | WO2023029229A1 (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113657628A (en) * | 2021-08-20 | 2021-11-16 | 武汉霖汐科技有限公司 | Industrial equipment monitoring method and system, electronic equipment and storage medium |
| CN113469300B (en) * | 2021-09-06 | 2021-12-07 | 北京航空航天大学杭州创新研究院 | Equipment state detection method and related device |
| CN114298443B (en) * | 2022-03-03 | 2022-06-14 | 北京航空航天大学杭州创新研究院 | Industrial equipment predictive maintenance method and device based on health state index and electronic equipment |
| CN115270165B (en) * | 2022-09-28 | 2022-12-13 | 南通佳茂霖智能科技有限公司 | Information management method for engineering survey |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11024321B2 (en) * | 2018-11-30 | 2021-06-01 | Google Llc | Speech coding using auto-regressive generative neural networks |
| US20210065059A1 (en) * | 2019-08-27 | 2021-03-04 | Nec Laboratories America, Inc. | Monitoring computing system status by implementing a deep unsupervised binary coding network |
| CN111738781B (en) * | 2020-07-31 | 2021-01-05 | 支付宝(杭州)信息技术有限公司 | Deep learning model and method for predicting flow data thereof |
| CN112179691B (en) * | 2020-09-04 | 2021-07-13 | 西安交通大学 | Mechanical equipment running state abnormity detection system and method based on counterstudy strategy |
| CN112461537B (en) * | 2020-10-16 | 2022-06-17 | 浙江工业大学 | Wind power gear box state monitoring method based on long-time and short-time neural network and automatic coding machine |
| CN112462261B (en) * | 2020-11-25 | 2022-09-20 | 创新奇智(广州)科技有限公司 | Motor abnormality detection method and device, electronic equipment and storage medium |
| CN112633476B (en) * | 2020-12-24 | 2023-09-22 | 东北大学 | Engine state monitoring method, system, equipment and medium |
| CN113067754B (en) * | 2021-04-13 | 2022-04-26 | 南京航空航天大学 | Semi-supervised time series anomaly detection method and system |
| CN113469300B (en) * | 2021-09-06 | 2021-12-07 | 北京航空航天大学杭州创新研究院 | Equipment state detection method and related device |
-
2021
- 2021-09-06 CN CN202111035618.8A patent/CN113469300B/en active Active
- 2021-11-19 WO PCT/CN2021/131934 patent/WO2023029229A1/en unknown
Also Published As
| Publication number | Publication date |
|---|---|
| WO2023029229A1 (en) | 2023-03-09 |
| CN113469300A (en) | 2021-10-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113469300B (en) | Equipment state detection method and related device | |
| Cornish et al. | Bayeswave: Bayesian inference for gravitational wave bursts and instrument glitches | |
| JP6636845B2 (en) | Hyperspectral demixing using forbidden compressed projection | |
| US20110191076A1 (en) | Error detection method and system | |
| CA2874991A1 (en) | Methods and apparatuses for defect diagnosis in a mechanical system | |
| CN109580004A (en) | A kind of temperature checking method and device | |
| CN108760305B (en) | Bearing fault detection method, device and equipment | |
| CN112257263B (en) | Equipment residual life prediction system based on self-attention mechanism | |
| Spiridonakos et al. | An FS-TAR based method for vibration-response-based fault diagnosis in stochastic time-varying structures: experimental application to a pick-and-place mechanism | |
| CN113486868B (en) | Motor fault diagnosis method and system | |
| CN110263767A (en) | In conjunction with the intelligent Rotating Shaft Fault method of compressed data acquisition and deep learning | |
| Maiorino et al. | Multifractal characterization of protein contact networks | |
| CN115015683B (en) | Cable production performance test method, device, equipment and storage medium | |
| He et al. | Separation and analyzing of harmonics and inter‐harmonics based on single channel independent component analysis | |
| US20170286841A1 (en) | Monitoring device and monitoring method thereof, monitoring system, and recording medium in which computer program is stored | |
| CN108562853A (en) | Method of Motor Fault Diagnosis based on error correcting output codes support vector machines and system | |
| CN113396369A (en) | Abnormality detection device | |
| KR20180103133A (en) | Simplified texture comparison engine | |
| CN117036732B (en) | Electromechanical equipment detection system, method and equipment based on fusion model | |
| CN116541771A (en) | Unbalanced sample bearing fault diagnosis method based on multi-scale feature fusion | |
| US11228820B2 (en) | Generating and/or encoding rotational data for a mechanical element over a digital network | |
| CN116090353A (en) | Product remaining life prediction method and device, electronic equipment and storage medium | |
| CN107067034B (en) | Method and system for rapidly identifying infrared spectrum data classification | |
| CN110159495B (en) | Fan tower early warning method, device and system | |
| CN107025382B (en) | A kind of engineering system health analysis system and method based on critical phase transformation theory |
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 |