CN113138005A - Vehicle random load pattern recognition system based on feature extraction and neural network - Google Patents
Vehicle random load pattern recognition system based on feature extraction and neural network Download PDFInfo
- Publication number
- CN113138005A CN113138005A CN202110524020.9A CN202110524020A CN113138005A CN 113138005 A CN113138005 A CN 113138005A CN 202110524020 A CN202110524020 A CN 202110524020A CN 113138005 A CN113138005 A CN 113138005A
- Authority
- CN
- China
- Prior art keywords
- data
- signal
- neural network
- module
- load
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01G—WEIGHING
- G01G19/00—Weighing apparatus or methods adapted for special purposes not provided for in the preceding groups
- G01G19/08—Weighing apparatus or methods adapted for special purposes not provided for in the preceding groups for incorporation in vehicles
- G01G19/12—Weighing apparatus or methods adapted for special purposes not provided for in the preceding groups for incorporation in vehicles having electrical weight-sensitive devices
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Testing Or Calibration Of Command Recording Devices (AREA)
Abstract
The invention discloses a vehicle random load pattern recognition system based on feature extraction and a neural network, which can accurately recognize the load state of mechanical equipment, improve the intelligent degree of a health monitoring system and ensure the safe and healthy operation of the mechanical equipment. The technical scheme is as follows: the system comprises a hardware platform capable of collecting samples and strain signals of mechanical equipment and carrying out data processing wireless transmission, and a software platform capable of preprocessing load data and identifying different random load types according to machine learning algorithms such as modern signal processing technologies, neural networks and the like, wherein a node network of the hardware platform is combined with the designed software and hardware equipment to form a set of system for identifying the random load modes of the vehicle, so that the random load types under different road conditions can be accurately identified, and the system has the capabilities of data wireless transmission and distributed strain detection.
Description
Technical Field
The invention relates to a technology for recognizing a vehicle random load pattern, in particular to a system for recognizing the vehicle random load pattern based on signal feature extraction and a neural network algorithm.
Background
Mechanical equipment is developing towards high speed, high accuracy and intellectualization, and the requirements of the working environment on the mechanical equipment are more severe, so that the mechanical equipment needs an accurate and intelligent health monitoring system to carry out real-time monitoring and early warning on the mechanical equipment in order to ensure the normal and healthy operation of the mechanical equipment. The health monitoring system can collect mass data to reflect the load condition of machinery and further evaluate the health condition and the service life of mechanical equipment, so that the health monitoring of the mechanical equipment enters a big data era. The mass of large mechanical data has the characteristics of large data volume, various contents and high effectiveness, but the current computing capability is limited, so how to extract characteristic data capable of accurately reflecting the operation condition of mechanical equipment from the mass data and grasp the operation state of the mechanical equipment under different working conditions has very important significance for improving the reliability, the safety and the intellectualization of the mechanical equipment.
The on-line monitoring technology is developing from a wired sensing monitoring network to a wireless sensing network with the characteristics of low power consumption, low cost, distribution and self-organization. The wireless sensor network technology comprises a sensor, a data acquisition and transmission module, a data processing and diagnosis module.
The sensor is the basis of wireless sensor network technology, and when a mechanical structure is monitored, the fatigue degree of the structure and the possible occurrence area of cracks can be found through strain measurement. The metal resistance type strain sensor has the characteristics of low manufacturing cost and relatively simple information reading system, and can be applied to a wireless sensing network to measure strain. However, the metal resistance type strain sensor works on the principle that the geometric shape of a sensing layer is changed, so that the sensitivity of the device is low, and the reliability of sensing data is influenced. The low sensitivity and high power consumption make the metal resistance strain sensor unable to meet the requirements of wireless sensing network.
Modern mechanical equipment has higher and higher requirements on intellectualization, and the mechanical equipment is required to react in time under the loads of different working environments. This is actually a typical pattern recognition problem, the key of which is the extraction and selection of features. Although the traditional neural network can identify signals, the extraction and selection of the features are relatively single, and accurate classification cannot be obtained. Therefore, the traditional neural network is directly applied to the health monitoring of the mechanical equipment, the requirement of the accuracy of the loading state of the mechanical equipment still cannot be met, and how to improve the traditional neural network is becoming reluctant.
Disclosure of Invention
The following presents a simplified summary of one or more aspects in order to provide a basic understanding of such aspects. This summary is not an extensive overview of all contemplated aspects, and is intended to neither identify key or critical elements of all aspects nor delineate the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form as a prelude to the more detailed description that is presented later.
The invention aims to solve the problems and provides a vehicle random load pattern recognition system based on feature extraction and a neural network, which can accurately recognize the load state of mechanical equipment, improve the intelligent degree of a health monitoring system and ensure the safe and healthy operation of the mechanical equipment.
The technical scheme of the invention is as follows: the invention discloses a vehicle random load pattern recognition system based on feature extraction and a neural network, which comprises a hardware platform, wherein the hardware platform is used for measuring a sample and a mechanical equipment strain signal and carrying out data acquisition processing and wireless transmission, the hardware platform comprises a sensor, a signal regulating circuit, a microprocessor, a data transmission chip and a power supply module, the signal regulating circuit is respectively and electrically connected with the sensor and the microprocessor, the microprocessor is respectively and electrically connected with the signal regulating circuit and the data transmission chip, and the system comprises:
the sensor is used for acquiring load data of the mechanical equipment under different working conditions;
the signal regulating circuit is used for applying voltage to the two ends of the sensor, amplifying and filtering the signal generated by the sensor through the signal regulating circuit, then performing analog-to-digital conversion to obtain a digital signal, and transmitting the digital signal to the microprocessor;
the microprocessor is used for awakening the power supply module through timing interruption and then carrying out power supply control on the power supply module, converting a digital signal obtained by the signal regulating circuit into stress data after filtering and preprocessing, processing the stress data to obtain a service life prediction result, and respectively sending the preprocessed stress data and the obtained service life prediction result to an external computer terminal and an external portable terminal in real time through a data transmission chip;
the data transmission chip is used for sending the stress data processed by the microprocessor to an external computer end for pattern recognition processing, and sending the service life prediction result processed by the microprocessor to an external portable terminal for display
And the power supply module is used for supplying power to each device in the hardware platform.
According to one embodiment of the vehicle random load pattern recognition system based on feature extraction and neural network, the sensor is a full-bridge flexible sensor based on porous conductive materials prepared by laser direct writing of a polyimide carbide film.
According to one embodiment of the system for recognizing the random load pattern of the vehicle based on the feature extraction and the neural network, the microprocessor obtains a life prediction result according to a digital signal transmitted by the signal regulating circuit based on a life prediction algorithm for fatigue damage life accumulation.
According to an embodiment of the system for recognizing the random load pattern of the vehicle based on the feature extraction and the neural network, the system further comprises a software platform, wherein the software platform is used for recognizing and classifying the random load pattern by adopting a modern signal processing technology and a neural network algorithm, and comprises a preprocessing module, a feature extraction module, a feature evaluation module and a neural network training module, wherein:
the preprocessing module is used for converting the voltage digital signals obtained by the signal regulating circuit into strain data, converting the strain data into stress data and inputting the stress data into the subsequent module for processing;
the characteristic extraction module is connected with the preprocessing module and used for carrying out signal characteristic processing on the data set of the stress data to respectively obtain a wavelet decomposition frequency band signal and an eigenmode component data set, and extracting dimensionless characteristic parameters in the wavelet decomposition frequency band signal and the eigenmode component data to obtain a characteristic parameter data set;
the characteristic evaluation module is connected with the characteristic extraction module and is used for evaluating the characteristic sensitivity of the characteristic parameter data and sequencing the characteristic parameter data according to the evaluated sensitivity;
the neural network training module is connected with the characteristic evaluation module and used for carrying out neural network training for matching the characteristics extracted from the stress data with the load mode by establishing a deep neural network model, establishing the relation between a random load spectrum and the load mode and realizing mode identification.
According to an embodiment of the system for random load pattern recognition of a vehicle based on feature extraction and neural networks of the present invention, the feature extraction module is further configured to:
decomposing the original load spectrum by using a wavelet packet decomposition algorithm, repeatedly decomposing the original load signal into continuous low-frequency and high-frequency components to obtain a wavelet decomposition frequency band signal, and decomposing the original load spectrum data by using an empirical mode decomposition algorithm to decompose the original load signal into a plurality of eigen-mode components.
According to one embodiment of the system for recognizing the random load pattern of the vehicle based on the feature extraction and the neural network, the feature evaluation module is used for carrying out feature evaluation on the sensitivity of the features through an evaluation method based on the distance between the features.
According to one embodiment of the system for recognizing the random load pattern of the vehicle based on the feature extraction and the neural network, the neural network training module divides an extracted feature parameter data set into a training set and a testing set according to a proportion, the training set data is divided into training data and verification data, the neural network is trained by using the training sets of the random load spectrum feature data of different road conditions, and then the testing samples of the random load spectrum feature parameters corresponding to the road conditions are input into the trained neural network for testing.
Compared with the prior art, the invention has the following beneficial effects: the system comprises a hardware platform capable of collecting samples and strain signals of mechanical equipment and performing data processing wireless transmission, and a software platform capable of preprocessing load data and identifying different random load types according to machine learning algorithms such as modern signal processing technologies, neural networks and the like, wherein a node network of the hardware platform is combined with the designed software and hardware equipment to form a system for identifying the random load modes of the vehicle, so that the system can accurately identify the random load types under different road conditions, and has the capabilities of data wireless transmission and distributed strain detection.
Drawings
The above features and advantages of the present disclosure will be better understood upon reading the detailed description of embodiments of the disclosure in conjunction with the following drawings. In the drawings, components are not necessarily drawn to scale, and components having similar relative characteristics or features may have the same or similar reference numerals.
Fig. 1 is a connection structure diagram of a hardware device in an embodiment of the vehicle random load pattern recognition system based on feature extraction and neural network of the present invention.
FIG. 2 is a schematic diagram of a software platform in an embodiment of the present invention for a system for random load pattern recognition of a vehicle based on feature extraction and neural networks.
FIG. 3 is a flowchart of the software platform process in an embodiment of the present invention for a system for identifying a random loading pattern of a vehicle based on feature extraction and neural networks.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. It is noted that the aspects described below in connection with the figures and the specific embodiments are only exemplary and should not be construed as imposing any limitation on the scope of the present invention.
The vehicle random load pattern recognition system based on feature extraction and neural network is a platform system combining software and hardware, wherein a hardware platform can measure a sample and a mechanical equipment strain signal and carry out data acquisition processing and wireless transmission, and a software platform can preprocess data and carry out random load pattern recognition and classification according to wavelet packet decomposition, empirical mode decomposition and other modern signal processing technologies and neural network algorithms. The vehicle random load pattern recognition system formed by combining the two can accurately recognize the random load types under different road conditions, and has the capabilities of data wireless transmission and distributed strain detection.
Fig. 1 shows functions implemented by respective hardware devices in a hardware platform and electrical connection relationships between the hardware devices. Referring to fig. 1, the hardware platform mainly includes a sensor, a signal conditioning circuit, a microprocessor MPU, a data transmission chip, and a power module.
The sensor is used for collecting load data of mechanical equipment under different working conditions. In this embodiment, the sensor may be a full-bridge flexible sensor based on a laser direct-write carbonized polyimide film to prepare a porous conductive material. The full-bridge sensor has the characteristics of high sensitivity strain and high tensile rate, and the polyimide film is irradiated by using ultraviolet laser with the wavelength of 405nm as a laser source in a laser direct writing preparation mode under the common environmental condition to generate the graphite carbon sensor and the sensor array. The polyimide film having a thickness of 0.05mm was used without any further treatment. The surface dust of the polyimide film is cleaned firstly, and then the film is spread on a glass plate flatly and is firmly stuck by using an adhesive tape. In the laser irradiation process, a laser beam is scanned on a surface of a Polyimide (PI) film, and various designed patterns can be easily produced by controlling a scanning path of a laser head. For example, a laser direct writing method can be used to carbonize a uniaxial strain gauge and a full-bridge strain gauge on a polyimide film at a laser power of 3W and a scanning speed of 7000 mm/min. The sensitivity and the tensile rate of the strain gauge prepared under different powers are different, the parameter with the optimal performance is selected, the sensitivity can reach 10.7 when the tensile strain reaches 3.3%, and compared with the traditional metal strain gauge, the sensitivity of the laser direct-writing carbonized PI strain gauge is greatly improved.
The signal conditioning circuit is respectively electrically connected with the sensor and the microprocessor and used for applying voltage to two ends of the sensor, amplifying and filtering signals generated by the sensor through the signal conditioning circuit, then performing analog-to-digital conversion to obtain digital signals, and transmitting the digital signals to the microprocessor MPU through an I2C interface. The signal conditioning circuit is, for example, a PGA305 conditioning circuit.
The microprocessor is respectively and electrically connected with the signal regulating circuit and the data transmission chip and is used for awakening the normal operation of the whole hardware platform through timing interruption (awakening the normal operation of the whole hardware platform through a threshold value acquired by the sensor), controlling the power supply of the power supply module after being awakened regularly, converting a digital signal obtained by the signal regulating circuit into stress data after filtering and preprocessing, processing a stress data service life prediction algorithm to obtain a service life prediction result, and respectively transmitting the preprocessed stress data and the service life prediction result obtained by calculation to an external computer end and an external portable terminal (such as an external mobile phone) in real time through the data transmission chip. The specific treatment is as follows: the service life prediction algorithm is an engineering algorithm for extracting a stress spectrum in original stress data by using a rain flow counting method and counting different stresses; and deducing the service life of the equipment material under the stress by using a Basquin formula for each stress value through comparing S-N curves of the material, and then calculating the residual life of the equipment material according to Miner criterion. The service life prediction algorithm and the real-time monitoring and displaying of the stress data are realized in the microprocessor, and then the preprocessed random load spectrum data are sent to the PC terminal for subsequent feature extraction, sensitivity evaluation and mode identification processing.
The data transmission chip is electrically connected with the microprocessor and used for sending the stress data processed by the microprocessor to an external computer end for pattern recognition processing, and sending a service life prediction result processed by the microprocessor to an external portable terminal for display. The data transmission chip is, for example, a CC2640R2F wireless transmission bluetooth chip.
The power module is used for supplying power to each device in the hardware platform. The power module is, for example, a CR2032 lithium manganese battery and a circuit board for supporting the battery.
As shown in fig. 2, the software platform in the system of the present embodiment includes the following modules: the device comprises a preprocessing module, a feature extraction module, a feature evaluation module and a neural network training module. The preprocessing module is used for processing in a microprocessor MPU (micro processing unit), and comprises filtering, converting digital voltage into stress data; the feature extraction module, the feature evaluation module and the neural network training module run at the PC end due to the computing power requirement. Fig. 3 shows a processing flow of the software platform, please refer to fig. 2 and fig. 3 simultaneously.
The preprocessing module is used for converting the voltage digital signals obtained by the signal regulating circuit into strain data, converting the strain data into stress data and storing the stress data, wherein the physical quantity of the stress and the strain in the conversion process is regarded as a linear elastic relation, namely sigma is E epsilon, wherein epsilon is the strain data, sigma is the stress data, and E is the elastic modulus of the material. In addition, the preprocessing module can also send the stress data stored in a period of time to an upper computer for subsequent processing at regular time through a data transmission chip.
The characteristic extraction module is connected with the preprocessing module and used for processing the data set of the stress data by adopting a modern signal characteristic processing technology including wavelet packet decomposition and empirical mode decomposition algorithms to respectively obtain a wavelet decomposition frequency band signal and an eigen mode component data set, and extracting dimensionless characteristic parameters in the wavelet decomposition frequency band signal and the eigen mode component data to obtain a characteristic parameter data set.
The specific processing procedures of the wavelet packet decomposition and the empirical mode decomposition are as follows:
the method comprises the steps of decomposing an original load spectrum by using a wavelet packet decomposition algorithm, and repeatedly decomposing an original load signal into continuous low-frequency and high-frequency components. In wavelet packet analysis, the wavelet details of each layer, in addition to decomposing only the wavelet approximation component in conventional wavelet analysis, are further decomposed into its own approximation and detail components. The method is a more precise signal analysis method, and improves the time domain resolution of the signal. Assuming an upper limit on the frequency of the signal of fnAfter the three-layer wavelet packet decomposition, the signal is decomposed into 8 mutually connected and independent frequency bands, the signal component of each frequency band subspace corresponds to the time-frequency local information of the original signal on the frequency band, and the integrity of the signal can be ensured. After reconstruction of the decomposed signal, the original signal is decomposed into time domain signals at 8 frequency bands, each band ranging from fnAnd/8, the wavelet packet decomposition can separate different frequency components appearing in the time domain, and effectively decompose non-stationary signals.
Meanwhile, the original load spectrum data is decomposed by using an Empirical Mode Decomposition (EMD) algorithm, the EMD can decompose a complex signal into a finite number of eigen mode functions (IMFs), and the instantaneous frequency of any point of the eigen mode function is meaningful. In the extraction of eigenmode components, each decomposition extracts a detail signal (fundamental mode component) and a low-frequency component having a frequency lower than that of the detail. The signal is therefore represented by n fundamental mode components and a residual as:
wherein c isi(t) is the fundamental mode component, whose frequency decreases in order from 1 to n, rn(t) is a non-oscillating monotonic sequence.
For example, as shown in fig. 3, time domain dimensionless indexes such as skewness, kurtosis, peak index, waveform index, pulse index and margin index are extracted from the original load spectrum data, 8 band wavelet packet decomposition signals and the first 6 eigenmode components in their time domains, and corresponding frequency domain dimensionless indexes are extracted from the frequency domains, so as to obtain a feature parameter data set with 102 features.
The characteristic evaluation module is connected with the characteristic extraction module and used for carrying out characteristic sensitivity evaluation on the characteristic parameter data and sequencing the characteristic parameter data according to the evaluated sensitivity. For example, the feature evaluation module performs feature evaluation on the sensitivity of the features by an evaluation method based on the distance between the features.
Some characteristic parameters in the extracted characteristic parameters are sensitive to different road condition loads and are independent of each other, so that the accuracy of classification and identification can be improved; there are many characteristic parameters with irrelevant and redundant information, these characteristics have no value to load pattern recognition and classification, too many input variables will increase training time of neural network recognition and classification, even reduce accuracy of load pattern recognition. Therefore, it is necessary to evaluate the sensitivity of the extracted features and select the feature parameters with high sensitivity.
Feature selection, which may also be referred to as feature subset selection, for a given N sample data set D of M-dimensional features, and a pattern recognition classification variable C, the objective is to select M features from the data set D of M features to best characterize the C classification variables. And evaluating the characteristic sensitivity based on the distance between the characteristics according to the following principle: the smaller the intra-class feature distance of the same class, the larger the inter-class feature distance of different classes, the higher the feature sensitivity of the feature is considered. The method comprises the following steps:
calculating the in-class distance of the jth feature of the c-th class
In the above formula: mcRepresents the number of samples of class c; j represents the number of features; c represents the number of categories; q. q.sm,c,j、q1,c,jRespectively represent the mth sum of class cThe characteristic value of the jth feature of the ith sample.
Calculate the average of the intra-class distances of the jth feature C class:
computing class c McAverage of jth feature of individual samples:
then, the average of the inter-class distances of the jth characteristic C class is calculated:
in the above formula: u. ofe,j、uc,jMean values of jth features of the e-th and c-th classes, respectively.
Calculating an evaluation factor for the jth feature:
αjthe size of (d) reflects the ease with which the jth feature classifies the C classes. Alpha is alphajThe larger the j-th feature is, the more sensitive the j-th feature is, the better the evaluation principle can be met, and the C classes can be classified more easily.
In the embodiment, the sensitive features are selected as input parameters of the deep neural network, and the influence of other features on the training and classification of the neural network is eliminated, so that the accuracy of random load pattern recognition is improved, and the training burden of the network is also reduced. The first 20 characteristic indexes with the sensitivity from large to small are selected as the input characteristics of the deep neural network, and the training speed, the accuracy and the applicability of the neural network are prevented from being influenced by the input of invalid characteristics. The deep neural network is a multilayer structure comprising an input layer, six hidden layers and an output layer, the input layer is 20 selected sensitive features, and the output layer is a load mode of a road condition. Firstly, establishing a characteristic parameter data set, dividing the extracted characteristic parameter data set into a training set and a testing set according to the proportion of 3:1, and dividing the training set data into training data and verification data. And training the network by using the random load spectrum characteristic data training set of different road conditions, and then inputting the test sample of the random load spectrum characteristic parameters of the corresponding road conditions into the trained network for testing.
The neural network training module performs neural network training for matching the features extracted from the stress data with the load mode by establishing a deep neural network model, and establishes a relation between a random load spectrum and the load mode to realize mode identification.
While, for purposes of simplicity of explanation, the methodologies are shown and described as a series of acts, it is to be understood and appreciated that the methodologies are not limited by the order of acts, as some acts may, in accordance with one or more embodiments, occur in different orders and/or concurrently with other acts from that shown and described herein or not shown and described herein, as would be understood by one skilled in the art.
Those of skill would further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
The various illustrative logical blocks, modules, and circuits described in connection with the embodiments disclosed herein may be implemented or performed with a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. A processor may also be implemented as a combination of computing devices, e.g., a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor such the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium may be integral to the processor. The processor and the storage medium may reside in an ASIC. The ASIC may reside in a user terminal. In the alternative, the processor and the storage medium may reside as discrete components in a user terminal.
In one or more exemplary embodiments, the functions described may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software as a computer program product, the functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A storage media may be any available media that can be accessed by a computer. By way of example, and not limitation, such computer-readable media can comprise RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer. Any connection is properly termed a computer-readable medium. For example, if the software is transmitted from a web site, server, or other remote source using a coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of medium. Disk (disk) and disc (disc), as used herein, includes Compact Disc (CD), laser disc, optical disc, Digital Versatile Disc (DVD), floppy disk and blu-ray disc where disks (disks) usually reproduce data magnetically, while discs (discs) reproduce data optically with lasers. Combinations of the above should also be included within the scope of computer-readable media.
The previous description of the disclosure is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other variations without departing from the spirit or scope of the disclosure. Thus, the disclosure is not intended to be limited to the examples and designs described herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (7)
1. The utility model provides a vehicle random load pattern recognition system based on feature extraction and neural network, its characterized in that, the system includes the hardware platform, the hardware platform is used for measuring sample, mechanical equipment strain signal and carries out data acquisition processing and wireless transmission, wherein the hardware platform includes sensor, signal conditioning circuit, microprocessor, data transmission chip, power module, signal conditioning circuit is connected with sensor and microprocessor electricity respectively, microprocessor is connected with signal conditioning circuit and data transmission chip electricity respectively, wherein:
the sensor is used for acquiring load data of the mechanical equipment under different working conditions;
the signal regulating circuit is used for applying voltage to the two ends of the sensor, amplifying and filtering the signal generated by the sensor through the signal regulating circuit, then performing analog-to-digital conversion to obtain a digital signal, and transmitting the digital signal to the microprocessor;
the microprocessor is used for awakening the power supply module through timing interruption and then carrying out power supply control on the power supply module, converting a digital signal obtained by the signal regulating circuit into stress data after filtering and preprocessing, processing the stress data to obtain a service life prediction result, and respectively sending the preprocessed stress data and the obtained service life prediction result to an external computer terminal and an external portable terminal in real time through a data transmission chip;
the data transmission chip is used for sending the stress data processed by the microprocessor to an external computer end for pattern recognition processing, and sending a service life prediction result obtained by the processing of the microprocessor to an external portable terminal for display;
and the power supply module is used for supplying power to each device in the hardware platform.
2. The system for recognizing the random load pattern of the vehicle based on the feature extraction and the neural network as claimed in claim 1, wherein the sensor is a full-bridge flexible sensor based on a porous conductive material prepared by laser direct writing of a polyimide carbide film.
3. The system of claim 1, wherein the microprocessor obtains the life prediction result according to the digital signal transmitted from the signal conditioning circuit based on the life prediction algorithm of fatigue damage life accumulation.
4. The system of claim 1, further comprising a software platform for recognizing and classifying the random loading pattern by using modern signal processing technology and neural network algorithm, wherein the software platform comprises a preprocessing module, a feature extraction module, a feature evaluation module and a neural network training module, wherein:
the preprocessing module is used for converting the voltage digital signals obtained by the signal regulating circuit into strain data, converting the strain data into stress data and inputting the stress data into the subsequent module for processing;
the characteristic extraction module is connected with the preprocessing module and used for carrying out signal characteristic processing on the data set of the stress data to respectively obtain a wavelet decomposition frequency band signal and an eigenmode component data set, and extracting dimensionless characteristic parameters in the wavelet decomposition frequency band signal and the eigenmode component data to obtain a characteristic parameter data set;
the characteristic evaluation module is connected with the characteristic extraction module and is used for evaluating the characteristic sensitivity of the characteristic parameter data and sequencing the characteristic parameter data according to the evaluated sensitivity;
the neural network training module is connected with the characteristic evaluation module and used for carrying out neural network training for matching the characteristics extracted from the stress data with the load mode by establishing a deep neural network model, establishing the relation between a random load spectrum and the load mode and realizing mode identification.
5. The feature extraction and neural network based vehicle random load pattern recognition system of claim 4, wherein the feature extraction module is further configured to:
decomposing the original load spectrum by using a wavelet packet decomposition algorithm, repeatedly decomposing the original load signal into continuous low-frequency and high-frequency components to obtain a wavelet decomposition frequency band signal, and decomposing the original load spectrum data by using an empirical mode decomposition algorithm to decompose the original load signal into a plurality of eigen-mode components.
6. The system of claim 4, wherein the feature evaluation module is configured to evaluate the sensitivity of the features by an evaluation method based on the distance between the features.
7. The system of claim 4, wherein the neural network training module divides the extracted feature parameter data set into a training set and a testing set in proportion, the training set data is divided into training data and verification data, the neural network is trained by using the training sets of the random load spectrum feature data of different road conditions, and then the testing samples of the random load spectrum feature parameters corresponding to the road conditions are input into the trained neural network for testing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110524020.9A CN113138005B (en) | 2021-05-13 | 2021-05-13 | Vehicle random load pattern recognition system based on feature extraction and neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110524020.9A CN113138005B (en) | 2021-05-13 | 2021-05-13 | Vehicle random load pattern recognition system based on feature extraction and neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113138005A true CN113138005A (en) | 2021-07-20 |
| CN113138005B CN113138005B (en) | 2023-07-11 |
Family
ID=76817756
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110524020.9A Active CN113138005B (en) | 2021-05-13 | 2021-05-13 | Vehicle random load pattern recognition system based on feature extraction and neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113138005B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113970436A (en) * | 2021-11-12 | 2022-01-25 | 北京亿华通科技股份有限公司 | Fatigue performance testing method for fuel cell cantilever support |
| CN113984175A (en) * | 2021-10-26 | 2022-01-28 | 东北大学秦皇岛分校 | Vehicle-mounted recalibration method based on artificial neural network and cloud service system |
| CN114993434A (en) * | 2022-05-20 | 2022-09-02 | 湖南科技大学 | Vehicle weight identification system and method based on EMD energy entropy of acoustic signal |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH07146936A (en) * | 1993-11-24 | 1995-06-06 | Babcock Hitachi Kk | Pattern recognizing method by hierarchical neural network |
| CN101408951A (en) * | 2008-11-18 | 2009-04-15 | 太原科技大学 | Method for obtaining equivalent load spectrum and estimating weariness residual longevity of bridge crane based on neural network |
| CN107990918A (en) * | 2017-10-20 | 2018-05-04 | 苏州大学 | The method that high sensitive piezoresistive transducer is prepared by multilevel hierarchy design |
| CN111241693A (en) * | 2020-01-16 | 2020-06-05 | 华东理工大学 | Life prediction system based on fatigue accumulation hypothesis |
| CN111679654A (en) * | 2020-06-17 | 2020-09-18 | 衢州职业技术学院 | Electric drive system vibration signal diagnosis method and device based on neural network |
-
2021
- 2021-05-13 CN CN202110524020.9A patent/CN113138005B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH07146936A (en) * | 1993-11-24 | 1995-06-06 | Babcock Hitachi Kk | Pattern recognizing method by hierarchical neural network |
| CN101408951A (en) * | 2008-11-18 | 2009-04-15 | 太原科技大学 | Method for obtaining equivalent load spectrum and estimating weariness residual longevity of bridge crane based on neural network |
| CN107990918A (en) * | 2017-10-20 | 2018-05-04 | 苏州大学 | The method that high sensitive piezoresistive transducer is prepared by multilevel hierarchy design |
| CN111241693A (en) * | 2020-01-16 | 2020-06-05 | 华东理工大学 | Life prediction system based on fatigue accumulation hypothesis |
| CN111679654A (en) * | 2020-06-17 | 2020-09-18 | 衢州职业技术学院 | Electric drive system vibration signal diagnosis method and device based on neural network |
Non-Patent Citations (1)
| Title |
|---|
| 谢金鑫: "高速动车组典型运用工况识别技术与方法研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113984175A (en) * | 2021-10-26 | 2022-01-28 | 东北大学秦皇岛分校 | Vehicle-mounted recalibration method based on artificial neural network and cloud service system |
| CN113970436A (en) * | 2021-11-12 | 2022-01-25 | 北京亿华通科技股份有限公司 | Fatigue performance testing method for fuel cell cantilever support |
| CN113970436B (en) * | 2021-11-12 | 2023-11-14 | 北京亿华通科技股份有限公司 | Fatigue performance testing method for fuel cell cantilever bracket |
| CN114993434A (en) * | 2022-05-20 | 2022-09-02 | 湖南科技大学 | Vehicle weight identification system and method based on EMD energy entropy of acoustic signal |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113138005B (en) | 2023-07-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113138005B (en) | Vehicle random load pattern recognition system based on feature extraction and neural network | |
| JP5145417B2 (en) | Signal determination method, signal determination apparatus, program, and signal determination system | |
| Wu et al. | Modeling and analysis of tool wear prediction based on SVD and BiLSTM | |
| CN114563150B (en) | Bridge health online detection module generation method, detection method, tool box and device | |
| CN112115643B (en) | Smart train service life non-invasive prediction method | |
| CN117037427B (en) | Geological disaster networking monitoring and early warning system | |
| Temko et al. | An SVM-based system and its performance for detection of seizures in neonates | |
| Huang et al. | Automatic classification of magnetic tiles internal defects based on acoustic resonance analysis | |
| CN116681186B (en) | Power quality analysis method and device based on intelligent terminal | |
| Xie et al. | Internal defect inspection in magnetic tile by using acoustic resonance technology | |
| CN117172601A (en) | Non-invasive load monitoring method based on residual total convolution neural network | |
| CN116698986A (en) | Sound absorbing material quality control system and method based on tuned band-pass filter | |
| CN117074852A (en) | Power distribution network electric energy monitoring and early warning management method and system | |
| CN115455684A (en) | Oil-immersed transformer non-contact state evaluation method and system based on noise analysis | |
| CN115931141A (en) | Temperature identification method of infrared temperature measurement map based on improved ANN algorithm | |
| CN114881938A (en) | Grain size detection method and system based on wavelet analysis and neural network | |
| CN113973403A (en) | Temperature-induced strain field redistribution intelligent sensing method based on structure discrete measuring point topology | |
| CN112880726A (en) | Sensor fault diagnosis method and device based on variational modal decomposition sample entropy | |
| CN111816404A (en) | Demagnetization method and system | |
| CN117898685B (en) | Pressure detection method and device based on different emotion states | |
| CN117292243B (en) | Method, equipment and medium for predicting magnetocardiogram signal space-time image based on deep learning | |
| CN117783792B (en) | Valve side sleeve insulation state detection method and system based on multiparameter real-time monitoring | |
| CN117949539B (en) | Engineering building strength detection system and detection method thereof | |
| CN117074291B (en) | Non-contact texture friction prediction method | |
| Harikumar et al. | A comparison of elman and radial basis function (rbf) neural networks in optimization of fuzzy outputs for epilepsy risk levels classification from eeg signals |
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 |