CN114371649A - Spray flow regulation and control system and method based on convolutional neural network - Google Patents
Spray flow regulation and control system and method based on convolutional neural network Download PDFInfo
- Publication number
- CN114371649A CN114371649A CN202210023228.7A CN202210023228A CN114371649A CN 114371649 A CN114371649 A CN 114371649A CN 202210023228 A CN202210023228 A CN 202210023228A CN 114371649 A CN114371649 A CN 114371649A
- Authority
- CN
- China
- Prior art keywords
- nozzle
- neural network
- convolutional neural
- spraying
- regulation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 58
- 239000007921 spray Substances 0.000 title claims abstract description 51
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000005507 spraying Methods 0.000 claims abstract description 59
- 230000005236 sound signal Effects 0.000 claims abstract description 25
- 230000002159 abnormal effect Effects 0.000 claims abstract description 12
- 238000004519 manufacturing process Methods 0.000 claims abstract description 6
- 238000012549 training Methods 0.000 claims abstract description 5
- 230000001276 controlling effect Effects 0.000 claims description 12
- 230000001105 regulatory effect Effects 0.000 claims description 12
- 238000000605 extraction Methods 0.000 claims description 10
- 238000005299 abrasion Methods 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 9
- 238000011176 pooling Methods 0.000 claims description 9
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 238000005259 measurement Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 230000003068 static effect Effects 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 2
- 230000004907 flux Effects 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 5
- 239000000575 pesticide Substances 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 239000000284 extract Substances 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 241000196324 Embryophyta Species 0.000 description 2
- 239000003814 drug Substances 0.000 description 2
- 229940079593 drug Drugs 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 241000238631 Hexapoda Species 0.000 description 1
- 241000607479 Yersinia pestis Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000009530 blood pressure measurement Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 210000005069 ears Anatomy 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000009432 framing Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 239000003595 mist Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/04—Programme control other than numerical control, i.e. in sequence controllers or logic controllers
- G05B19/042—Programme control other than numerical control, i.e. in sequence controllers or logic controllers using digital processors
- G05B19/0423—Input/output
-
- A—HUMAN NECESSITIES
- A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
- A01M—CATCHING, TRAPPING OR SCARING OF ANIMALS; APPARATUS FOR THE DESTRUCTION OF NOXIOUS ANIMALS OR NOXIOUS PLANTS
- A01M7/00—Special adaptations or arrangements of liquid-spraying apparatus for purposes covered by this subclass
- A01M7/0089—Regulating or controlling systems
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B05—SPRAYING OR ATOMISING IN GENERAL; APPLYING FLUENT MATERIALS TO SURFACES, IN GENERAL
- B05B—SPRAYING APPARATUS; ATOMISING APPARATUS; NOZZLES
- B05B12/00—Arrangements for controlling delivery; Arrangements for controlling the spray area
- B05B12/08—Arrangements for controlling delivery; Arrangements for controlling the spray area responsive to condition of liquid or other fluent material to be discharged, of ambient medium or of target ; responsive to condition of spray devices or of supply means, e.g. pipes, pumps or their drive means
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01F—MEASURING VOLUME, VOLUME FLOW, MASS FLOW OR LIQUID LEVEL; METERING BY VOLUME
- G01F1/00—Measuring the volume flow or mass flow of fluid or fluent solid material wherein the fluid passes through a meter in a continuous flow
- G01F1/66—Measuring the volume flow or mass flow of fluid or fluent solid material wherein the fluid passes through a meter in a continuous flow by measuring frequency, phase shift or propagation time of electromagnetic or other waves, e.g. using ultrasonic flowmeters
- G01F1/666—Measuring the volume flow or mass flow of fluid or fluent solid material wherein the fluid passes through a meter in a continuous flow by measuring frequency, phase shift or propagation time of electromagnetic or other waves, e.g. using ultrasonic flowmeters by detecting noise and sounds generated by the flowing fluid
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/20—Pc systems
- G05B2219/24—Pc safety
- G05B2219/24215—Scada supervisory control and data acquisition
Landscapes
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Fluid Mechanics (AREA)
- Electromagnetism (AREA)
- Insects & Arthropods (AREA)
- Pest Control & Pesticides (AREA)
- Wood Science & Technology (AREA)
- Zoology (AREA)
- Environmental Sciences (AREA)
- Special Spraying Apparatus (AREA)
Abstract
A spray flow regulation and control system and method based on a convolutional neural network are disclosed, wherein the system is composed of an acoustic signal measuring unit, an operation control unit and a flow regulation and control unit. The method comprises the following steps: acquiring nozzle sound signals with different pressures and different working durations, extracting MFCC combination characteristics, training convolutional neural network parameters, and establishing a nozzle working state recognition model; setting optimal spraying parameters according to the growth condition of crops, collecting sound signals of a nozzle during working in real time, and inputting the sound signals into an identification model; when the working state of the nozzle is abnormal, the parameters of the spraying system are adjusted, and an alarm is given after the limit is exceeded, so that production technicians are reminded to replace the nozzle. The flow regulation and control of the invention directly adjusts the rotating speed and the starting and stopping of the motor of the diaphragm pump through the motor driving board to realize the compensation and the existence of the nozzle flow, converts one-dimensional audio information into two-dimensional image information with time dimension, accelerates the identification speed and precision, and has good flow prediction effect and high precision.
Description
Technical Field
The invention relates to the field of accurate spraying and deep learning voice classification, and particularly provides a spraying flow control system and method based on a convolutional neural network.
Background
With the improvement of national environmental protection requirements and the development of spraying technology, precise pesticide application is the current development trend. The core of the accurate pesticide application technology is to acquire the difference of diseases, insect pests and weeds in small field areas, and apply the pesticide according to needs by adopting an efficient spraying technology and a variable pesticide application technology.
At present, research related to accurate pesticide application mainly focuses on two aspects of field information acquisition and variable spraying devices. Miao Yubin et al proposed a system for weed area calculation and variable spraying by using machine vision and ultrasonic ranging; qili jun et al designed an automatic variable speed swing type greenhouse spraying device, which realizes accurate spraying by a spraying mechanism with movable position during application. However, in the actual drug application process, the error is large, and the controllability and accuracy of the drug dosage are poor.
Disclosure of Invention
The invention aims to provide a spray flow regulation and control system and method based on a convolutional neural network by monitoring the working state of a nozzle, so that early warning and regulation and control of abnormal spray flow conditions and real-time monitoring and feedback of spray flow are realized, production technicians are reminded to replace damaged nozzles in time, and the reliability and stability of a spray system are improved. The invention takes a key terminal part nozzle of a spraying system as a monitoring object, obtains the characteristics of the nozzle under different working conditions by an audio characteristic extraction method, further realizes the identification of the working state of the nozzle by a convolution neural network method, and finally realizes the feedback control of the spraying flow by adjusting the rotating speed of a motor of a diaphragm pump.
A spray flow regulation and control system based on a convolutional neural network comprises an acoustic signal measuring unit, an operation control unit and a flow regulation and control unit. The acoustic signal measuring unit is used for receiving an acoustic signal emitted by the nozzle during spraying and coding analog audio into digital audio; the operation control unit is used for carrying out audio feature extraction and convolutional neural network identification of the working state of the nozzle on the nozzle sound signal and sending out a corresponding control instruction according to the spraying state; the flow regulating and controlling unit is used for meeting the requirements of crops on spraying and regulating and controlling and alarming when the spraying is abnormal. The regulation and control system collects acoustic signals of the nozzle under different pressures and different working durations through the microphone, and fully extracts acoustic characteristics of the nozzle under common spray pressure and different abrasion degrees. The sound signals collected when the spraying system works are input into the trained convolutional neural network in the operation control unit, and the working state of the nozzle is output. Through the rotational speed of regulation and control diaphragm pump, compensate nozzle wearing and tearing to when wearing and tearing surpass the allowable degree and the nozzle blocks up, drive alarm module reports to the police.
Further, the acoustic signal measurement unit is composed of a microphone and an audio codec; the microphone is a capacitive waterproof microphone and is used for receiving an acoustic signal emitted by the nozzle during spraying; the audio codec is a low-power-consumption and low-voltage stereo audio codec, and is used for encoding analog audio received by the microphone into digital audio and further sending the digital audio to the operation control unit.
Furthermore, the operation control unit comprises an embedded high-performance AI computing module and a singlechip; the embedded high-performance AI computing module is a microcomputer with a highly integrated CPU and GPU and is used for carrying out audio feature extraction and convolutional neural network identification of the working state of the nozzle on the nozzle acoustic signal; the single chip microcomputer is an embedded microcontroller and is used for receiving the control information of the embedded high-performance AI computing module and further driving the flow control unit.
Furthermore, the flow regulation and control unit consists of a spraying module, an early warning module and a regulation and control module; the spraying module mainly comprises a diaphragm pump, a pressure gauge and a nozzle and is used for generating stable spraying under certain pressure; the diaphragm pump is a pressure switch type high-pressure electric diaphragm pump; the pressure gauge is a pointer type pressure gauge meeting the common spray pressure measurement; the nozzle is a proper type nozzle with a spray angle and flow rate meeting requirements; the early warning module mainly comprises a buzzer and is used for warning when the flow state is abnormal; the regulation and control module mainly comprises a motor drive plate and is used for regulating the rotating speed of the diaphragm pump and controlling starting and stopping of the diaphragm pump.
Furthermore, the sound signal measuring unit is communicated with the high-performance AI computing module in an I2C and I2S mode, and the single chip microcomputer is communicated with the high-performance AI computing module through a UART serial port
A spray flow regulation and control method based on a convolutional neural network comprises the following steps: acquiring nozzle sound signals with different pressures and different working durations, extracting MFCC combination characteristics, training convolutional neural network parameters, and establishing a nozzle working state recognition model; setting optimal spraying parameters according to the growth condition of crops, collecting sound signals of a nozzle during working in real time, and inputting the sound signals into an identification model; when the working state of the nozzle is abnormal, the parameters of the spraying system are adjusted, and an alarm is given after the limit is exceeded, so that production technicians are reminded to replace the nozzle. The regulation and control method fully extracts the acoustic characteristics of the nozzle with different abrasion degrees under the common spray pressure by acquiring the acoustic signals of the nozzle under different pressures and different working durations. The acoustic signal acquired when the spraying system works is input into the trained convolutional neural network to acquire the working state of the nozzle, and the method has high recognition speed and precision. The abrasion of the nozzle is compensated by regulating and controlling the parameters of the spraying system, and an alarm is given when the abrasion exceeds an allowable degree and the nozzle is blocked.
Further, the MFCC combination feature is composed of a 13-dimensional MFCC feature and first-order and second-order differences thereof, and is a 39-dimensional audio feature used for representing the sounding characteristics of the nozzle. The MFCC is a short name of Mel-Frequency Cepstral coeffients (Mel-Frequency Cepstral coeffients), is a cepstrum parameter extracted from a nozzle sound signal in a nonlinear Mel scale Frequency domain, reflects the static characteristic of the nozzle sound signal, and the first-order and second-order differences of the nozzle sound signal are used for reflecting the dynamic characteristic.
Further, the convolutional neural network is composed of 2 convolutional layers, 2 pooling layers, 1 fully-connected layer and 1 depth layer. The convolution layer Conv2D _1 is composed of 16 convolution kernels of 3 × 3 with a step size of 1. The convolution layer Conv2D _2 is composed of 32 2 × 2 convolution kernels, and has a step size of 2. Each convolutional layer is followed by a 2 x2 pooling layer with a step size of 2.
Further, convolutional layers and depth layers of a convolutional neural network use different activation functions. The two convolutional layers use a linear rectification function (ReLU) to increase the nonlinearity of the network; the depth layer connected to the output uses Mean Squared Error (MSE) as a loss function to more accurately predict the flow value of the nozzle acoustic signal response.
Further, in the convolutional neural network, the parameter number of the depth layer connected to the output is 1, and is used for converting 960 parameters of the fully-connected layer into a final flow prediction result.
Compared with the existing spray flow regulation system and method, the invention has the following beneficial effects:
1) compared with the conventional method for regulating and controlling the flow by adopting an electromagnetic valve and the like, the flow regulation and control method provided by the invention has the advantages that the rotation speed and the starting and stopping of the motor of the diaphragm pump are directly regulated by the motor driving plate, so that the compensation and the existence of the flow of the nozzle are realized.
2) In the identification of the working state of the nozzle, the invention converts one-dimensional audio information into two-dimensional image information with time dimension, thereby accelerating the speed and the precision of the identification.
3) The parameter number of the depth layer connected with the output of the convolutional neural network adopted by the invention is 1, 960 parameters of the full-connection layer are converted into a final flow prediction result, and the convolutional neural network has good prediction effect and high precision.
Drawings
Fig. 1 is a functional diagram of an apparatus of a spray flow control system based on a convolutional neural network according to an embodiment of the present invention.
Fig. 2 is a control flow chart of a spray flow rate control method based on a convolutional neural network according to an embodiment of the present invention.
Fig. 3 is a schematic diagram illustrating specific components of MFCC combination features in a convolutional neural network-based spray flow rate control method according to an embodiment of the present invention.
Fig. 4 is a schematic diagram of the structure and parameters of a convolutional neural network according to an embodiment of the present invention.
Detailed Description
The technical scheme of the invention is further explained in detail by combining the drawings in the specification.
A system and a method for controlling a spray flow based on a convolutional neural network according to an embodiment of the present invention are described below with reference to the accompanying drawings.
Fig. 1 is a functional diagram of an apparatus of a spray flow control system based on a convolutional neural network according to an embodiment of the present invention. As shown in fig. 1, the spray flow rate control system based on the convolutional neural network comprises: the device comprises an acoustic signal measuring unit, an operation control unit and a flow regulating and controlling unit.
The acoustic signal measuring unit is used for receiving an acoustic signal emitted by the nozzle during spraying and encoding analog audio into digital audio; the operation control unit is used for carrying out audio feature extraction and convolutional neural network identification of the working state of the nozzle on the nozzle acoustic signal and sending a corresponding control instruction according to the spraying state; the flow regulation and control unit is used for meeting the requirements of crops on the generation of spraying and the regulation and control and alarm when the spraying is abnormal.
Specifically, the acoustic signal measurement unit is constituted by a microphone and an audio codec; the microphone is a capacitive waterproof microphone and is used for receiving an acoustic signal sent by the nozzle during spraying, and the type of the microphone adopted in one embodiment of the invention is CME-1538-100 LB; the audio codec is a low-power-consumption and low-voltage stereo audio codec, and is used for encoding analog audio received by a microphone into digital audio and further sending the digital audio to an operation control unit, and the type of the audio codec adopted in one embodiment of the invention is TLV320AIC 3204.
Specifically, the operation control unit comprises an embedded high-performance AI computing module and a singlechip; the embedded high-performance AI computing module is a microcomputer with a highly integrated CPU and GPU, and is used for performing audio feature extraction and convolutional neural network recognition of the working state of a nozzle on a nozzle acoustic signal, and an NVIDIA Jetson TX2 development kit is adopted in one embodiment of the invention and comprises a core board, a carrier board, a radiator, a power supply and other components; the single chip microcomputer is an embedded microcontroller and is used for receiving control information of the embedded high-performance AI computing module and further driving the flow control unit, and an embodiment of the invention adopts the single chip microcomputer of STM32F103 type.
Specifically, the flow regulation and control unit consists of a spraying module, an early warning module and a regulation and control module; the spraying module mainly comprises a diaphragm pump, a pressure gauge and a nozzle and is used for generating stable spraying under certain pressure; the diaphragm pump is a pressure switch type high-pressure electric diaphragm pump, and the diaphragm pump adopted in one embodiment of the invention is XTL-3210 HB-24-100; the pressure gauge is a pointer type pressure gauge meeting the measurement of common spray pressure, and the type of the pressure gauge adopted in one embodiment of the invention is Y-100 red flags; the nozzle is a proper type nozzle with the spray angle and the flow rate meeting the requirements, and the type of the nozzle adopted in one embodiment of the invention is LICHENG VP 11001; the early warning module mainly comprises an active buzzer and is used for warning when the flow state is abnormal; the control module mainly comprises a motor driving board and is used for controlling the adjustment and the start-stop of the rotating speed of the diaphragm pump, and the model of the motor driving board adopted in one embodiment of the invention is L298N.
When the spray flow regulation and control system based on the convolutional neural network works, a microphone firstly collects a nozzle sound signal, the nozzle sound signal is transmitted to an embedded high-performance AI computing module through an audio codec to be identified, a control instruction is sent to a single chip microcomputer according to a computing result, a buzzer is driven to give an alarm or a motor driving plate is used for regulating and controlling the rotating speed of a diaphragm pump motor, and early warning and regulation and control of the spray system flow are realized through judgment of the state of a nozzle.
When the system works, the audio codec and the high-performance AI computing module are communicated in an I2C and I2S mode, the single chip microcomputer and the high-performance AI computing module are communicated through a UART serial port, and the single chip microcomputer drives the buzzer or the motor driving board through the serial port.
The spray flow control system based on the convolutional neural network collects acoustic signals of the nozzle under different pressures and different working durations through the microphone, and fully extracts acoustic characteristics of the nozzle under common spray pressure and different abrasion degrees. The sound signals collected when the spraying system works are input into the trained convolutional neural network in the operation control unit, and the working state of the nozzle is output. Through the rotational speed of regulation and control diaphragm pump, compensate nozzle wearing and tearing to when wearing and tearing surpass the allowable degree and the nozzle blocks up, drive alarm module reports to the police.
Fig. 2 is a control flow chart of a spray flow rate control method based on a convolutional neural network according to an embodiment of the present invention. As shown in fig. 2, the spray flow rate control method based on the convolutional neural network includes the following steps:
in step S01, a nozzle operating condition recognition model is established by collecting nozzle acoustic signals with different pressures and different operating durations, extracting MFCC combination characteristics, training convolutional neural network parameters. The acoustic signals of the nozzle with different working time periods are obtained, so that the acoustic characteristics of the nozzle under common spray pressure and different abrasion degrees are extracted.
As shown in FIG. 3, the MFCC composite features consist of 13-dimensional MFCC features and their first and second order differences, which are 39-dimensional audio features used to characterize the nozzle sound production. The MFCC is a short name of Mel-Frequency Cepstral coeffients (Mel-Frequency Cepstral coeffients), is a cepstrum parameter extracted from a nozzle sound signal in a nonlinear Mel scale Frequency domain, reflects the static characteristic of the nozzle sound signal, and the first-order and second-order differences of the nozzle sound signal are used for reflecting the dynamic characteristic.
The extraction of MFCC features mainly comprises the following steps:
1) intercepting a section of 1s of nozzle acoustic signal segment for identifying flow characteristics; 2) dividing frames, wherein the default length of the frame is 2048, the default length of the frame is 512, the audio vector of X1 is processed into an N M matrix, and zero padding is carried out at the end when the audio vector cannot be divided completely to form a complete frame; 3) adopting a Hanning window to carry out windowing operation so as to eliminate discontinuity between frames after framing; 4) performing fast Fourier transform, absolute value taking and square opening operation frame by frame to obtain an energy spectrogram; 5) constructing 128 Mel filters, performing dot product operation with the energy spectrogram, and converting the energy spectrogram into Mel frequency spectrum closer to human ear mechanism; 6) in order to simulate the logarithmic characteristic of human ears, taking the logarithm of a Mel frequency spectrogram; 7) the data distribution is changed by discrete cosine transform, and most of signal data is concentrated to a low frequency region. One embodiment of the present invention takes the first 13 values as MFCC feature values.
One embodiment of the invention employs the audio feature extraction toolkit Python _ speed _ features in Python for MFCC combination feature extraction.
The spray flow control method based on the convolutional neural network takes the trained convolutional neural network as a recognizer and a classifier of the nozzle acoustic signal, and the current spray state is recognized by inputting the extracted MFCC combined features into the network.
Fig. 4 is a structural and parameter diagram of a convolutional neural network in a convolutional neural network-based spray flow rate control method according to an embodiment of the present invention.
As shown in fig. 4, the convolutional neural network inputs the two-dimensional MFCC combined features with the size of 39 × 44 into the network, converts the two-dimensional MFCC combined features into one-dimensional data with the size of 960 through convolution, pooling and Flatten operations, and outputs the one-dimensional data with the size of 1 through a density operation, that is, the predicted value of the current spray flow.
The convolutional neural network consists of 2 convolutional layers, 2 pooling layers, 1 full-link layer and 1 depth layer. The convolution layer Conv2D _1 is composed of 16 convolution kernels of 3 × 3 with a step size of 1. The convolution layer Conv2D _2 is composed of 32 2 × 2 convolution kernels, and has a step size of 2. Each convolutional layer is followed by a 2 x2 pooling layer with a step size of 2. Referring to fig. 4, the convolutional neural network includes, in order from an Input to an Output, a convolutional layer Conv2D _1, a pooling layer MaxPooling2D _1, a convolutional layer Conv2D _1, a pooling layer MaxPooling2D _2, a full-link layer Flatten, and a depth layer density.
Different activation functions are used for the convolutional layers and the depth layers of the convolutional neural network. The two convolutional layers use a linear rectification function (ReLU) to increase the nonlinearity of the network; the depth layer connected to the output uses Mean Squared Error (MSE) as a loss function to more accurately predict the flow value of the nozzle acoustic signal response.
Wherein, yiAnd
respectively a predicted value and a true value of the spray flow. And y isiTo convolve the output of the neural network in actual operation,
the output of the convolutional neural network is correspondingly obtained when the network is trained.
In step S02, the optimum spray parameters are set according to the growth conditions of the crops, and the sound signals of the nozzles during operation are collected in real time and input to the recognition model.
The optimal spraying parameters refer to parameters for ensuring the optimal spraying effect, such as the type and the spraying angle of a nozzle, the pressure during spraying, the corresponding flow value, the installation height of the nozzle and the like.
When the acoustic signal of the nozzle is collected in real time, the installation position of the microphone relative to the nozzle is subjected to certain experimental research. For the flat fan nozzle used in one embodiment of the invention, the microphone is arranged on the side of the fan-shaped spray close to the nozzle, so that a more reliable sound signal can be acquired on the premise of avoiding the microphone from being wetted by the spray. When gathering the nozzle acoustic signal in real time, for the ambient noise when reducing the acoustic signal collection, at the microphone except that the direction at nozzle place, installed a protection casing to improve the rate of accuracy of spraying state discernment.
In step S03, when the nozzle is in an abnormal operating condition, the parameters of the spraying system are adjusted, and an alarm is given after the limit is exceeded, thereby reminding the production technician to replace the nozzle.
Abnormal nozzle operating conditions include both nozzle clogging and nozzle wear. When the nozzle is blocked, the spray mist type is obviously changed, and the spray flow is reduced in a large range; the nozzle wearing and tearing can cause the rising of spraying flow, and certain scope wearing and tearing do not influence spraying effect very much, and the accessible is adjusted spraying parameter and is compensated, nevertheless wear and tear behind the limit, need in time change nozzle.
The acoustic signal acquired when the spraying system works is input into the trained convolutional neural network to acquire the working state of the nozzle, and the method has high recognition speed and precision. The abrasion of the nozzle is compensated by regulating and controlling the parameters of the spraying system, and an alarm is given when the abrasion exceeds an allowable degree and the nozzle is blocked.
The above description is only a preferred embodiment of the present invention, and the scope of the present invention is not limited to the above embodiment, but equivalent modifications or changes made by those skilled in the art according to the present disclosure should be included in the scope of the present invention as set forth in the appended claims.
Claims (10)
1. The utility model provides a spraying flow control system based on convolutional neural network which characterized in that: the method comprises the following steps:
the acoustic signal measuring unit is used for receiving an acoustic signal emitted by the nozzle during spraying and coding the analog audio into digital audio;
the operation control unit is used for carrying out audio feature extraction and convolutional neural network identification of the working state of the nozzle on the nozzle acoustic signal, acquiring the spraying state at the nozzle and sending a corresponding control instruction according to the spraying state;
and the flow regulating and controlling unit is used for meeting the requirements of crops on spraying and regulating and controlling and alarming when the spraying is abnormal.
2. The convolutional neural network-based spray flow control system according to claim 1, wherein: the acoustic signal measurement unit includes a capacitive waterproof microphone and a stereo audio codec.
3. The convolutional neural network-based spray flow control system according to claim 2, wherein: the operation control unit comprises an embedded AI calculation module and an embedded microcontroller singlechip.
4. The convolutional neural network-based spray flow control system according to claim 3, wherein: the flow regulation and control unit comprises a spraying module for generating spraying, an early warning module for giving an alarm when the flow state is abnormal, and a regulation and control module for regulating and starting and stopping the spraying system; the spraying module comprises a diaphragm pump, a pressure gauge and a nozzle, the diaphragm pump adopts a pressure switch type high-pressure electric diaphragm pump, and the pressure gauge adopts a pointer type pressure gauge for measuring spraying pressure; the early warning module comprises a buzzer; the regulation and control module comprises a motor driving plate.
5. The convolutional neural network-based spray flow control system according to claim 4, wherein: the motor driving plate is used for adjusting the rotating speed and starting and stopping of the motor of the diaphragm pump to realize the compensation and the existence of the flow of the nozzle, and the flow regulation is realized.
6. The convolutional neural network-based spray flow control system according to claim 5, wherein: the sound signal measuring unit is communicated with the AI computing module in an I2C and I2S mode, and the single chip microcomputer is communicated with the AI computing module through a UART serial port.
7. A regulation method using the convolutional neural network-based spray flux regulation system as described in any one of claims 1 to 6, characterized in that: the method comprises the following steps:
step 1, extracting MFCC combination characteristics by collecting nozzle acoustic signals with different pressures and different working durations, training convolutional neural network parameters, and establishing a nozzle working state recognition model;
step 2, setting optimal spraying parameters according to the growth condition of crops, collecting sound signals of a nozzle during working in real time, and inputting the sound signals into an identification model;
and 3, when the working state of the nozzle is abnormal, adjusting parameters of a spraying system, and alarming after exceeding the limit to remind production technicians to replace the nozzle.
Wherein, the acoustic signals of the nozzles with different pressures and different working durations in the step 1 are measured under laboratory conditions for obtaining the acoustic signal characteristics of the nozzles with different abrasion degrees under different spray pressures, and the duration of each section of audio is set to be 1 second; the audio data set is obtained by random extraction from the recorded nozzle acoustic signal in a steady spray regime.
8. The convolutional neural network-based spray flow control method according to claim 7, wherein: in the identification of the operating state of the nozzle, one-dimensional audio information is converted into two-dimensional image information with a time dimension.
9. The convolutional neural network-based spray flow control method according to claim 8, wherein: the MFCC combination feature is composed of a 13-dimensional MFCC feature and first-order and second-order differences thereof, and is a 39-dimensional audio feature used for representing the sounding characteristics of the nozzle. Wherein, 13-dimensional MFCC characteristics are used for reflecting the static characteristics of the nozzle sound signals, and the first-order and second-order differences are used for reflecting the dynamic characteristics of the nozzle sound signals.
10. The convolutional neural network-based spray flow control method according to claim 10, wherein: the convolutional neural network consists of 2 convolutional layers, 2 pooling layers, 1 full-link layer and 1 depth layer. The convolution layer Conv2D _1 is composed of 16 convolution kernels of 3 × 3 with a step size of 1. The convolution layer Conv2D _2 is composed of 32 2 × 2 convolution kernels, and has a step size of 2. Each convolution layer is connected with a 2 multiplied by 2 pooling layer with the step length of 2;
different activation functions are used for the convolutional layer and the depth layer of the convolutional neural network; the two convolution layers use a linear rectification function ReLU; the depth layer connected with the output uses mean square error MSE as a loss function;
wherein, yiAnd
respectively a predicted value and a true value of the spray flow; and y isiTo convolve the output of the neural network in actual operation,
convolution neural network output for corresponding network training;
in the convolutional neural network, the parameter number of the depth layer connected with the output is 1, and the parameter number is used for converting 960 parameters of the full connection layer into a final flow prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210023228.7A CN114371649A (en) | 2022-01-10 | 2022-01-10 | Spray flow regulation and control system and method based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210023228.7A CN114371649A (en) | 2022-01-10 | 2022-01-10 | Spray flow regulation and control system and method based on convolutional neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114371649A true CN114371649A (en) | 2022-04-19 |
Family
ID=81144990
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210023228.7A Pending CN114371649A (en) | 2022-01-10 | 2022-01-10 | Spray flow regulation and control system and method based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114371649A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115324759A (en) * | 2022-10-12 | 2022-11-11 | 南通普盛动力有限公司 | Method for identifying abnormality of oil nozzle of diesel engine |
| CN116124234A (en) * | 2023-02-24 | 2023-05-16 | 宁波力擎超声科技有限公司 | Ultrasonic flowmeter for gas |
| WO2024021258A1 (en) * | 2022-07-29 | 2024-02-01 | 福建天甫电子材料有限公司 | Control system for intelligent production line of electronic-grade potassium hydroxide, and control method thereof |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5741980A (en) * | 1994-11-02 | 1998-04-21 | Foster-Miller, Inc. | Flow analysis system and method |
| US20050011278A1 (en) * | 2003-07-18 | 2005-01-20 | Brown Gregory C. | Process diagnostics |
| CN102968664A (en) * | 2012-10-10 | 2013-03-13 | 江苏大学 | Mobile robot variable pesticide applying intelligent decision-making method and system |
| US20170080445A1 (en) * | 2015-09-17 | 2017-03-23 | Cnh Industrial America Llc | Nozzle Flow Detection System |
| CN107013449A (en) * | 2017-04-18 | 2017-08-04 | 山东万腾电子科技有限公司 | Voice signal based on deep learning recognizes the method and system of compressor fault |
| CN107016239A (en) * | 2017-03-30 | 2017-08-04 | 贵州电网有限责任公司电力科学研究院 | ASCA and RBF neural network model a kind of steam turbine valve traffic characteristics analysis method |
| CN111264497A (en) * | 2020-02-10 | 2020-06-12 | 江苏大学 | Device and method for monitoring blockage of nozzle of rice spraying machine in real time based on sound information |
| CN111832432A (en) * | 2020-06-23 | 2020-10-27 | 华中科技大学 | Cutter wear real-time prediction method based on wavelet packet decomposition and deep learning |
| KR102214572B1 (en) * | 2019-12-10 | 2021-02-09 | 주식회사 승광이엔지 | System for controlling oil mist spraying, heating and cooling |
| US20210068612A1 (en) * | 2019-09-06 | 2021-03-11 | Lg Electronics Inc. | Operational state analyzable dishwasher and method of analyzing operational state of dishwasher |
| CN113551744A (en) * | 2021-08-10 | 2021-10-26 | 中国石油大学(北京) | Ultrasonic flowmeter performance online monitoring method and system |
-
2022
- 2022-01-10 CN CN202210023228.7A patent/CN114371649A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5741980A (en) * | 1994-11-02 | 1998-04-21 | Foster-Miller, Inc. | Flow analysis system and method |
| US20050011278A1 (en) * | 2003-07-18 | 2005-01-20 | Brown Gregory C. | Process diagnostics |
| CN102968664A (en) * | 2012-10-10 | 2013-03-13 | 江苏大学 | Mobile robot variable pesticide applying intelligent decision-making method and system |
| US20170080445A1 (en) * | 2015-09-17 | 2017-03-23 | Cnh Industrial America Llc | Nozzle Flow Detection System |
| CN107016239A (en) * | 2017-03-30 | 2017-08-04 | 贵州电网有限责任公司电力科学研究院 | ASCA and RBF neural network model a kind of steam turbine valve traffic characteristics analysis method |
| CN107013449A (en) * | 2017-04-18 | 2017-08-04 | 山东万腾电子科技有限公司 | Voice signal based on deep learning recognizes the method and system of compressor fault |
| US20210068612A1 (en) * | 2019-09-06 | 2021-03-11 | Lg Electronics Inc. | Operational state analyzable dishwasher and method of analyzing operational state of dishwasher |
| KR102214572B1 (en) * | 2019-12-10 | 2021-02-09 | 주식회사 승광이엔지 | System for controlling oil mist spraying, heating and cooling |
| CN111264497A (en) * | 2020-02-10 | 2020-06-12 | 江苏大学 | Device and method for monitoring blockage of nozzle of rice spraying machine in real time based on sound information |
| CN111832432A (en) * | 2020-06-23 | 2020-10-27 | 华中科技大学 | Cutter wear real-time prediction method based on wavelet packet decomposition and deep learning |
| CN113551744A (en) * | 2021-08-10 | 2021-10-26 | 中国石油大学(北京) | Ultrasonic flowmeter performance online monitoring method and system |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024021258A1 (en) * | 2022-07-29 | 2024-02-01 | 福建天甫电子材料有限公司 | Control system for intelligent production line of electronic-grade potassium hydroxide, and control method thereof |
| CN115324759A (en) * | 2022-10-12 | 2022-11-11 | 南通普盛动力有限公司 | Method for identifying abnormality of oil nozzle of diesel engine |
| CN115324759B (en) * | 2022-10-12 | 2022-12-27 | 南通普盛动力有限公司 | Method for identifying abnormality of oil nozzle of diesel engine |
| CN116124234A (en) * | 2023-02-24 | 2023-05-16 | 宁波力擎超声科技有限公司 | Ultrasonic flowmeter for gas |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114371649A (en) | Spray flow regulation and control system and method based on convolutional neural network | |
| Agrawal et al. | Novel TEO-based Gammatone features for environmental sound classification | |
| CN108198562A (en) | A kind of method and system for abnormal sound in real-time positioning identification animal house | |
| EP3923273B1 (en) | Voice recognition method and device, storage medium, and air conditioner | |
| CN101023469B (en) | Digital filtering method, digital filtering equipment | |
| CN105989842B (en) | The method, apparatus for comparing vocal print similarity and its application in digital entertainment VOD system | |
| CN108447479A (en) | The robot voice control system of noisy work condition environment | |
| CN108297108B (en) | Spherical following robot and following control method thereof | |
| CN108460334A (en) | A kind of age forecasting system and method based on vocal print and facial image Fusion Features | |
| JP2017532082A (en) | A system for speech-based assessment of patient mental status | |
| EP4131256A1 (en) | Voice recognition system and method using accelerometers for sensing bone conduction | |
| CN100563608C (en) | The electric cochlea Chinese fixed electric stimulation amplitude changing pattern in-vitro voice processing unit | |
| TWI749663B (en) | Method for monitoring phonation and system thereof | |
| CN110368005A (en) | A kind of intelligent earphone and mood and physiological health monitoring method based on intelligent earphone | |
| CN106264839A (en) | Intelligent snore stopping pillow | |
| Birkholz et al. | Non-invasive silent phoneme recognition using microwave signals | |
| CN109448700A (en) | A kind of built-in speech recognition system and method based on improvement MFCC algorithm | |
| CN111243624B (en) | Method and system for evaluating personnel state | |
| CN104965194B (en) | Indoor multi-sound-source positioning device and method simulating binaural effect | |
| Zhang et al. | A novel insect sound recognition algorithm based on MFCC and CNN | |
| CN117292713A (en) | Transformer voiceprint monitoring method and system based on end Bian Yun cooperation | |
| Lin et al. | Acoustic recognition method in low SNR based on human ear bionics | |
| CN113948109B (en) | System for recognizing physiological phenomenon based on voice | |
| EP4360087A1 (en) | Multi-encoder end-to-end automatic speech recognition (asr) for joint modeling of multiple input devices | |
| Matassoni et al. | Hands-free speech recognition using a filtered clean corpus and incremental HMM adaptation |
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 |