A hand-held artificial intelligence-based goji berry picking equipment

Abstract

The application discloses a kind of hand-held Chinese wolfberry picking equipment based on artificial intelligence, belong to the field of intelligent picking of Chinese wolfberry fruits.The hand-held Chinese wolfberry picking equipment of the application starts the equipment quickly by limit switch, simultaneously utilizes acoustic sensor to collect acoustic signal when picking, combines traditional algorithm with deep learning algorithm, classifies, judges, identifies and time sequence information seeking of the acoustic signal collected, to control the working state and working frequency of push-rod type electromagnet, solve the problem that vibration type equipment needs to continuously adjust vibration frequency and amplitude when picking is not uniform;Meanwhile, a collecting device is designed, which solves the problem of fruit collection difficulty when picking with vibration type equipment, and the device reduces fruit damage rate and artificial fatigue, improves economic benefits.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medlar fruit picking equipment, and particularly relates to a hand-held medlar picking equipment based on artificial intelligence. BACKGROUND

[0002] Medlar is a kind of plant with medicinal and edible properties, and is widely planted in western regions of China. Traditional medlar picking methods are mainly manual picking, which is labor-intensive, low in efficiency and unstable in picking quality. The picking problem has become a key problem restricting the development of the medlar industry. With the gradual realization of agricultural mechanization and intelligentization in China, current medlar picking equipment mainly includes self-propelled and hand-held types, and the hand-held type is dominant. In the hand-held medlar picking equipment, the vibration type equipment is the most ideal picking method due to high picking efficiency and small damage rate. However, due to the mechanical structure design, most of the current vibration type equipment has some problems.

[0003] A vibration type picking machine invented by Gao Sen (CN206978083U) has a relatively large contact area of the vibration rod during picking, which causes the vibration rod to directly contact the medlar fruit and shake it off most of the time, resulting in a large damage rate. A vibration type medlar picking machine invented by Qin Zhan Yang (CN211656899U) mainly vibrates branches, but does not have a fruit collecting device, which also causes damage to the fruit and increases the workload of the picking workers. A vibration type picking machine invented by Shi Tengfei (CN219248614U) increases the fruit collecting device, but inevitably causes uneven vibration amplitude during picking. Moreover, the above three vibration type picking machines need to observe the medlar picking situation by naked eye during picking, and control the vibration frequency in real time to prevent the vibration frequency from being too large to damage flowers and leaves, or too small to slow down the picking efficiency and affect the economic benefits. SUMMARY

[0004] In order to improve the above problems, the present application designs a hand-held medlar picking equipment based on artificial intelligence. The technical scheme of the present application is as follows:

[0005] The application discloses a hand-held artificial intelligence-based wolfberry picking equipment, characterized in that: a lower collecting groove 3 with a baffle is arranged on the picking equipment, the lower collecting groove 3 is connected with an upper collecting groove 2 through a guide rail, a guide rail positioning pin 8 is arranged at the bottom of the lower collecting groove 3, and a left vertical plate 1 is connected with the upper collecting groove 2 through a guide rail; an acoustic sensor 6 and an embedded controller 7 are fixed on the side of the upper collecting groove 2; the upper collecting groove 2 and the lower collecting groove 3 are provided with two clamping devices 4 and a plurality of picking assemblies 5; the clamping device 4 comprises a side base 41, the side base 41 is fixed on the side of the upper collecting groove 2 and the lower collecting groove 3 through bolts and nuts; a push rod type electromagnet 42 is fixed on the side base 41, the output end of the push rod type electromagnet 42 is connected with an extension rod 43, the extension rod 43 is connected with a left clamping plate 44, the left clamping plate 44 and a right clamping plate 45 are fixed together through a screw 46, a spring 47 and a limit switch 48 are installed on the right clamping plate 45, and the right clamping plate is fixed on the upper collecting groove 2 through a base 49; the picking assembly 5 comprises a side base 51, the side base 51 is fixed on the side of the upper collecting groove 2 and the lower collecting groove 3 through bolts and nuts; a push rod type electromagnet 52 is fixed on the side base 51, the output end of the push rod type electromagnet 52 is connected with a connecting rod 53, the connecting rod 53 is fixed on a clamping sleeve 54, the clamping sleeve 54 and a vibrating rod 55 are connected together, and the vibrating rod 55 is fixed on the bottom of the upper collecting groove 2 and the lower collecting groove 3 through a lower base 56.

[0006] The limit switch 48 is fixed on the right clamping plate 45 in the clamping device 4, the limit switch 48 is used for triggering and sending a signal at the beginning of picking, the embedded controller 7 is used for receiving the signal of the limit switch 48 and controlling the working state of the push rod type electromagnet 42; the material of the vibrating rod in the picking assembly 5 is rigid material, the position of the side base 51 in the vertical direction of the upper collecting groove 2 and the lower collecting groove 3 is changed, the push rod type electromagnet 52 is controlled to move up and down, and the amplitude of the vibrating rod 55 is changed; the acoustic sensor 6 is used for collecting acoustic signals and sending the signals to the embedded controller 7; the embedded controller 7 receives the acoustic signals and processes data by using an algorithm, and controls the working state and the working frequency of the push rod type electromagnet 52.

[0007] The embedded controller 7 comprises a signal receiving module, a signal identification module and a control module; the signal receiving module is configured to receive information sent by the limit switch 48 and the acoustic sensor 6; the signal identification module is configured to identify and classify the signals; the control module is configured to receive signals sent by the signal identification module, and control the working state and working frequency of the push rod type electromagnet 52 according to the signals; the signal identification module comprises: a signal preprocessing unit configured to preprocess the received acoustic data to obtain preprocessed data; a time sequence envelope obtaining unit configured to obtain time sequence envelope data by performing time sequence envelope obtaining on the preprocessed data; a feature extraction unit configured to obtain an acoustic feature set by performing feature extraction on the time sequence envelope data; a classification unit configured to obtain a target acoustic feature set generated by the collision of the Chinese wolfberry and the collection groove by classifying the acoustic feature set; a timing unit configured to start timing from the push rod type electromagnet (52) of the collecting assembly, and each 5 seconds is a period; a judgment unit configured to judge whether the target acoustic feature set is an empty set; a first signal sending unit configured to send a signal to the first control module in response to the target acoustic feature set being an empty set and the period of the timing unit being greater than or equal to 3, otherwise, the next step is entered, wherein the first control module is used to close the push rod type electromagnet 52 of the collecting assembly; an identification unit configured to identify the target acoustic feature set to obtain an identification result; and a second signal sending unit configured to send a signal to the second control module according to the identification result, wherein the second control module is used to adjust the working frequency of the push rod type electromagnet 52 of the collecting assembly.

[0008] The application discloses a handheld Chinese wolfberry picking equipment based on artificial intelligence, which is quickly started by a limit switch, and acoustics information during picking is obtained in real time through an acoustic sensor, traditional algorithms and deep learning algorithms are combined, acoustic signals are classified, judged, identified and time sequence information is obtained, so that the vibration frequency of a picking assembly and the working state of the equipment are controlled, the problem that the vibration frequency and amplitude of the vibration type equipment need to be continuously adjusted by workers during picking is solved, a guide rail type fruit collecting device is designed, the phenomenon that fruits are scattered during picking is avoided, the fruit damage rate and the fatigue of picking workers are reduced, and economic benefits are improved. BRIEF DESCRIPTION OF DRAWINGS

[0009] Figure 1 It is a schematic diagram of the overall structure of the application;

[0010] Figure 2 It is a schematic diagram of the back structure of the application;

[0011] Figure 3 It is a schematic diagram of the clamping device of the application;

[0012] Figure 4Structure schematic diagram of picking assembly of the present application;

[0013] Figure 5 Structure schematic diagram of left stand plate in lowered state during operation of the present application;

[0014] Figure 6 Structure schematic diagram of left stand plate in retracted state at the end of harvesting of the present application;

[0015] Figure 7 Overall control flow schematic diagram of the present application;

[0016] Figure 8 Embedded controller algorithm flow chart of the present application;

[0017] Figure 9 Main circuit schematic diagram of the present application. DETAILED DESCRIPTION

[0018] Before operation of the present application, the guide rail positioning pin 8 at the back of the bottom of the lower collecting tank 3 is removed, the upper collecting tank 2 is pulled out of the lower collecting tank 3 through the guide rail, the left stand plate 1 is fixed on the upper collecting tank 2, and the position of the side base 51 in the picking assembly 5 in the vertical direction of the upper collecting tank 2 and the lower collecting tank 3 is adjusted to determine the appropriate amplitude size.

[0019] When the operation starts, the branch is manually controlled to enter the clamping device, the limit switch 48 is triggered, the limit switch 48 sends a signal to the embedded controller 7, the embedded controller 7 receives the signal, starts the push rod type electromagnet 42, drives the extension rod 43 to press the left clamping plate 44, and the left clamping plate 44 is pressed against the spring 47, so that the left clamping plate 44 and the right clamping plate 45 clamp the branch. After the two side clamping devices 4 are clamped, after 3 seconds, the embedded controller 7 sends a signal to start the push rod type electromagnet 52 of the picking assembly 5, the push rod type electromagnet 52 shakes the vibration rod 55 through the connecting rod 53, so as to achieve the purpose of vibrating and harvesting the wolfberry, in order to prevent too high vibration frequency and too much flowers and leaves from falling, the initial vibration frequency is set to be low. Then the acoustic sensor 6 fixed on the upper collecting tank 2 starts to collect acoustic data, the data single frame is 50-100 milliseconds long, and the acoustic data is sent to the embedded controller 7.

[0020] The embedded controller 7 comprises a signal receiving module, a signal identification module and a control module; the signal receiving module is used for receiving information sent by the limit switch 48 and the acoustic sensor 6; the signal identification module comprises a signal preprocessing unit, a time sequence envelope obtaining unit, a feature extraction unit, a classification unit, a timing unit, a judgment unit, a first signal sending unit, an identification unit and a second signal sending unit; when receiving the acoustic information sent by the acoustic sensor 6, the preprocessing unit pre-processes the acoustic information; the time sequence envelope obtaining unit obtains time sequence envelope data and the change of the acoustic signal with time by performing time sequence envelope obtaining on the pre-processed data; the feature extraction unit and the classification unit perform feature extraction and classification on the time sequence envelope data by using a deep learning algorithm to obtain a target acoustic feature set generated by the collision of the Chinese wolfberry and the collection groove; the judgment unit judges the classified target acoustic feature set, and if it is a non-empty set, the timing unit is reset to zero, the acoustic signal time sequence information in the signal envelope obtaining unit is called, if the time sequence information conveys that the acoustic signal is from sparse to dense, the target acoustic feature set is sent to the identification module, the identification module identifies the target acoustic feature set by using an identification algorithm to generate a signal 2 containing identification result information, the second signal sending unit sends the signal 2 to the second control module to adjust the working frequency of the push rod electromagnet 52, and if the time sequence information conveys that the acoustic signal is from dense to sparse, the picking frequency of the push rod electromagnet 52 remains unchanged; after the above actions are completed, the acoustic information sent by the acoustic sensor 6 is returned, and the above process is repeated to adjust the working frequency of the push rod electromagnet to the best picking frequency range; if it is an empty set, the timing unit is called, if the timing unit cycle is less than 3, the second signal sending unit sends a signal 3 to the second control module to increase the working frequency of the push rod electromagnet 52, the timing unit cycle is increased by one, and then the above steps are repeated; if the timing unit cycle is greater than or equal to 3, the first signal sending unit sends a signal 1 to the first control module to control the push rod electromagnets 42 and 52 to stop working, the spring 47 rebounds, the two side holding devices 4 are loosened, the left vertical plate 1 is moved from the Figure 5 position to Figure 6 position, the Chinese wolfberry is poured out, and the Chinese wolfberry fruit picking process is completed.

[0021] The limit switch 48, the acoustic sensor 6 and the embedded controller 7 are electrically connected.

[0022] For the control process involving the embedded controller 7, when the target acoustic feature set is a non-empty set, the timing information of the acoustic signal obtained by the envelope data obtaining unit is used to control the frequency of the electromagnet. When the acoustic signal is from sparse to dense in time, it indicates that the equipment is picking more and more wolfberry, at this time the signal is identified by using deep learning algorithm, and the working frequency of the electromagnet is adjusted. When the acoustic signal is from dense to sparse in time, it indicates that the equipment is picking less and less wolfberry, at this time the working frequency of the electromagnet is not adjusted, which can maximize the picking of wolfberry.

[0023] For the control process involving the embedded controller 7, when the target acoustic feature set is a non-empty set, the timing information of the acoustic signal obtained by the envelope data obtaining unit is used to control the frequency of the electromagnet. When the acoustic signal is from sparse to dense in time, it indicates that the equipment is picking more and more wolfberry, at this time the signal is identified by using deep learning algorithm, and the working frequency of the electromagnet is adjusted. When the acoustic signal is from dense to sparse in time, it indicates that the equipment is picking less and less wolfberry, at this time the working frequency of the electromagnet is not adjusted, which can maximize the picking of wolfberry.

[0024] For the algorithm involving the embedded controller 7, the pre-processing method includes but is not limited to various filters and pre-emphasis algorithm. The deep learning classification method includes but is not limited to traditional convolutional neural network CNN and Transformer model. The identification algorithm includes but is not limited to convolutional neural network CNN, deep belief network DBN and deep hidden Markov model HMM. For pre-training data, the sound of wolfberry colliding with the collection tank with artificial annotation is used as the training set for pre-training.

Claims

1. A hand-held artificial intelligence-based wolfberry picking equipment, characterized in that: The picking equipment is provided with a lower collecting groove (3) with a baffle, the lower collecting groove (3) is connected with the upper collecting groove (2) through a guide rail, the lower collecting groove (3) is provided with a guide rail positioning pin (8) at the back of the bottom, the left vertical plate (1) is connected with the upper collecting groove (2) through a guide rail; the upper collecting groove (2) is fixed with an acoustic sensor (6) and an embedded controller (7) on the side; the upper collecting groove (2) and the lower collecting groove (3) are provided with two clamping devices (4) and multiple picking assemblies (5) inside; the clamping device (4) comprises a first side base (41), the first side base (41) is fixed on the side of the upper collecting groove (2) and the lower collecting groove (3) through bolts and nuts; the first side base (41) is fixed with a first push rod type electromagnet (42), the output end of the first push rod type electromagnet (42) is connected with an extension rod (43), the extension rod (43) is connected with a left clamping plate (44), the left clamping plate (44) and a right clamping plate (45) are fixed together through a screw (46), the right clamping plate (45) is installed with a spring (47) and a limit switch (48), and the right clamping plate is fixed on the upper collecting groove (2) through a base (49); the picking assembly (5) comprises a second side base (51), the second side base (51) is fixed on the side of the upper collecting groove (2) and the lower collecting groove (3) through bolts and nuts; the second side base (51) is fixed with a second push rod type electromagnet (52), the output end of the second push rod type electromagnet (52) is connected with a connecting rod (53), the connecting rod (53) is fixed on a clamping sleeve (54), the clamping sleeve (54) and a vibrating rod (55) are connected together, and the vibrating rod (55) is fixed on the bottom of the upper collecting groove (2) and the lower collecting groove (3) through a lower base (56); the limit switch (48) is fixed on the right clamping plate (45) in the clamping device (4), and the limit switch (48) is used for triggering and sending a signal to the embedded controller (7) at the beginning of picking; the acoustic sensor (6) is used for collecting acoustic signals and sending the signals to the embedded controller (7); the embedded controller (7) comprises a signal receiving module, a signal identification module and a control module; the signal receiving module is configured to receive information sent by the limit switch (48) and the acoustic sensor (6); the signal identification module is configured to identify and classify the signals; and the control module is configured to receive signals sent by the signal identification module, and control the working state and working frequency of the second push rod type electromagnet (52) according to the signals. The signal recognition module comprises: a signal preprocessing unit configured to preprocess received acoustic data to obtain preprocessed data; a time envelope obtaining unit configured to obtain time envelope data and time sequence information of the acoustic signal by performing time envelope obtaining on the preprocessed data; a feature extraction unit configured to obtain an acoustic feature set by performing feature extraction on the time envelope data; a classification unit configured to obtain a target acoustic feature set generated by the collision of the Chinese wolfberry and the collection tank by performing classification on the acoustic feature set; a timing unit configured to start timing from the second push rod electromagnet (52) of the picking assembly, and each period is 5 seconds; a judgment unit configured to judge whether the target acoustic feature set is an empty set; a first signal sending unit configured to send a signal to the control module in response to the target acoustic feature set being an empty set and the period of the timing unit being greater than or equal to 3, otherwise, the next step is entered, wherein the control module is used to close the second push rod electromagnet (52) of the picking assembly; a recognition unit configured to obtain a recognition result by recognizing the target acoustic feature set; and a second signal sending unit configured to send a signal to the control module according to the recognition result, wherein the control module is used to adjust the working frequency of the second push rod electromagnet (52) of the picking assembly.

2. The artificial intelligence-based hand-held Chinese wolfberry picking equipment according to claim 1, characterized in that: The material of the vibration rod (55) in the picking assembly (5) is rigid material, the position of the second side base (51) in the vertical direction of the upper collection tank (2) and the lower collection tank (3) is changed, the second push rod electromagnet (52) is controlled to move up and down, and the amplitude of the vibration rod (55) is changed.

Images