A fruit tree canopy leaf density estimation system and method based on excitation audio

Abstract

The present invention belongs to the field of plant protection technology, and relates to a system and method for estimating the leaf density of a fruit tree canopy based on excitation audio, which mainly includes an excitation device, an audio collector, a data acquisition module, a controller, a power supply, etc. The excitation device and the audio collector can adjust the height through a telescopic excitation device support rod and an audio collector support rod to achieve the measurement of the leaf density of the canopy at different heights; the data acquisition module is used to store the audio signal acquired by the audio collector; the controller communicates with the data acquisition module in real time, and can perform Fourier transform, filtering, feature parameter extraction, and other processing on the audio signal, while estimating the leaf density of the fruit tree canopy based on the constructed leaf density estimation model; the host computer can display the time domain and frequency domain waveforms of the audio signal during the processing process, the audio feature parameters, the acquired leaf density of the fruit tree canopy, etc. in real time. The present invention can provide a theoretical basis for the accurate identification of canopy characteristics for precision pesticide application in orchards.

Description

Technical Field

[0001] The present invention relates to the technical field of orchard plant protection, and in particular to a system and method for estimating the leaf density of a fruit tree canopy based on excitation audio. Background Art

[0002] In the existing precision (omnidirectional) air-delivered pesticide application process for fruit trees, real-time detection of fruit tree canopy characteristics is the primary issue in achieving precision air-delivered pesticide application.

[0003] In existing research, the canopy characteristics of fruit trees mainly include canopy volume, leaf wall area and other indicators. The method of "sensor scanning / collection → solving geometric equations → fitting canopy contours" is generally adopted, which simplifies the asymmetric and porous structure of the canopy and ignores the objective fact that the main recipients of pesticides are relatively sparse leaves.

[0004] Leaf area density (hereinafter referred to as "leaf density") refers to the sum of the leaf area per unit volume at a certain height of the crop. Compared with indicators such as canopy volume and leaf wall area, it can better characterize the canopy growth and the number of pesticide-receiving objects. However, existing research rarely involves methods for online measurement of canopy leaf density.

[0005] Patent CN103528920A discloses a device and method for measuring leaf area density. The device includes a semiconductor laser, a beam expander, a controller, a power supply circuit, a receiving sensor, and a probe. The probe is the main frame of the device and has a harpoon structure. Two parallel straight rods are the optical detection parts of the device. The distance between the two straight rods is 50 cm or 100 cm. One or more semiconductor lasers are installed at the end of one of the straight rods. The light beam emitted by the laser is vertically pointed at the receiving sensor installed at the end of the other straight rod. From the analysis of its working process and working principle, it is actually still using sensor signals to invert leaf density. Due to the limitations of sensor accuracy and environmental factors, the accuracy of the measurement results is easily affected, making it difficult to meet the requirements of online measurement during plant protection operations.

[0006] The information disclosed in this background technology section is only intended to deepen the understanding of the overall background technology of the present invention and should not be regarded as an admission or any form of suggestion that the information constitutes the prior art already known to those skilled in the art. Summary of the Invention

[0007] The purpose of the present invention is to provide a fruit tree canopy leaf density estimation system and method based on excitation audio, in view of the differences in the sounds emitted by the different leaf densities of the fruit tree canopy under excitation, so as to solve the problem of difficulty in obtaining leaf density during the spraying operation, realize the rapid prediction of the fruit tree canopy leaf density, and meet the needs of precise air-delivered spraying of fruit trees.

[0008] In order to achieve the above objectives, the specific technical solutions adopted by the present invention are as follows:

[0009] A fruit tree canopy leaf density estimation system based on excitation audio, characterized by comprising an excitation device 1, an excitation device support rod 2, an audio collector 3, an audio collector support rod 4, a support wheel 5, a data acquisition module 6, a controller 7, a power supply 8, a mobile platform 9 and a host computer 10;

[0010] The vibration device 1 is fixedly mounted on the rear end of the mobile platform 9 via the vibration device support rod 2, and is used to force the canopy leaves and branches of the fruit tree to shake and generate audio;

[0011] The audio collector 3 is installed at the front end of the mobile platform 9 through the audio collector support rod 4, and is used to collect audio signals;

[0012] The audio collector 3 communicates with the data acquisition module 6 and transmits the collected audio signal to the data acquisition module 6 in real time;

[0013] The controller 7 is connected to the data acquisition module 6 and is used to perform filtering, Fourier transform, feature parameter extraction and other processing on the audio signal to obtain the feature parameters of the audio signal, and further obtain the actual leaf density of the canopy based on the leaf density estimation model to achieve the estimation of the leaf density of the fruit tree canopy;

[0014] The excitation device support rod 2 and the audio collector support rod 4 adopt a telescopic design, which can adjust the height of the excitation device 1 and the audio collector 3 to achieve the estimation of leaf density at different height positions of the fruit tree canopy;

[0015] The height h1 of the installation position of the vibration device 1 from the ground and the height h2 of the installation position of the audio collector 3 from the ground meet the following conditions:

[0016]

[0017] Wherein, L is the horizontal distance between the support rod 2 of the excitation device and the support rod 4 of the audio collector, and α is half of the coverage angle of the wind blowing device;

[0018] The host computer 10 is installed on the upper part of the mobile platform 9 and can communicate with the controller 7 to intuitively display the audio feature parameters obtained by the controller 7 from processing the audio signal and the estimated value of the canopy leaf density of the fruit tree;

[0019] The power supply 8, data acquisition module 6 and controller 7 are installed on the bottom storage table of the mobile platform 9;

[0020] The controller 7 adjusts the speed and direction of the support wheels 5 through the drive mechanism, thereby achieving movement control of the mobile platform 9; the controller 7 adjusts the telescopic lifting and lowering of the audio collector support rod 4 through the drive mechanism, thereby achieving height control of the audio collector 3; the controller 7 adjusts the telescopic lifting and lowering of the vibration device support rod 2 through the drive mechanism, thereby achieving height control of the vibration device 1;

[0021] The power supply 8 supplies power to the vibration device 1 , the audio collector 3 , the data acquisition module 6 , the controller 7 , and the host computer 10 .

[0022] On the basis of the above technical solution, the fruit tree canopy leaf density estimation system based on excitation audio adopts a fruit tree canopy leaf density estimation method, which includes the following steps:

[0023] S1: Move the leaf density estimation system to the canopy estimation position, adjust the height of the vibration device 1 and the audio collector 3, and start the system;

[0024] S2: The vibration device 1 forces the branches and leaves at the estimated position of the fruit tree canopy to shake, causing friction between "branch-leaf" and "leaf-leaf" to produce audio;

[0025] S3: Audio collector 3 collects original audio signals;

[0026] S4: transmitting the original audio signal to the data acquisition module 6 for storage;

[0027] S5: The controller 7 performs pre-processing such as Fourier transform and filtering on the original audio signal in the data acquisition module 6 to obtain the actual audio signal of the fruit tree canopy after removing the interference audio such as wind noise and environmental noise;

[0028] S6: The controller 7 further processes the actual audio signal of the fruit tree canopy to obtain characteristic parameters such as short-time energy, characteristic frequency, and zero-crossing rate of the audio signal;

[0029] S7: taking the audio feature parameters as input and importing them into the leaf density estimation model based on the audio feature parameters;

[0030] S8: Obtain the canopy leaf density of fruit trees;

[0031] S9: The host computer 10 displays in real time the audio feature parameters in the processing process, the obtained canopy leaf density of the fruit tree, and the time domain and frequency domain waveforms of the audio signal.

[0032] On the basis of the above technical solution, in the method for estimating leaf density of a fruit tree canopy, the leaf density estimation model based on audio feature parameters is constructed using a mathematical modeling method or a machine learning method, as follows:

[0033] Method 1: When the leaf density estimation model based on audio feature parameters is constructed using a mathematical modeling method, the mathematical relationship expression shown in Formula 2 is used to construct the leaf density F and the audio feature parameter T i The functional relationship between

[0034] F=K i ·T i +B (2)

[0035] Where T i T is the audio feature parameter obtained by the controller 7 from processing the audio signal. i =[t1,t2,t3,…,t n ] T , K i is the coefficient of the audio feature parameter, K i =[k1,k2,k3,…,k n ], B is the threshold compensation parameter;

[0036] Method 2: When the leaf density estimation model based on audio feature parameters is constructed using a machine learning method, the prediction model constructed by the machine learning method generally includes an input layer, a hidden layer, and an output layer, and the audio feature parameters T obtained by the controller 7 through processing the audio signal are i As input, leaf density F is used as output, and a multi-input single-output prediction model is constructed to predict the leaf density of the fruit tree canopy.

[0037] The system and method for estimating the leaf density of a fruit tree canopy based on excitation audio described in the present invention have the following beneficial effects:

[0038] Aiming at the differences in sounds emitted by vibration under different leaf densities of fruit tree canopies, the present invention provides a system and method for estimating the leaf density of fruit tree canopies based on vibration audio. Different from the means of reversely reconstructing the geometric characteristics of the canopy using conventional sensors, the present invention innovatively proposes a method for estimating the leaf density of fruit tree canopies by using audio generated by vibration, which reduces the errors caused by environmental factors in the process of detecting canopy characteristics by conventional sensors, and can realize high-precision online measurement of canopy leaf density, providing a theoretical basis for accurate identification of canopy characteristics for precision pesticide application in orchards. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The present invention has the following accompanying drawings:

[0040] The accompanying drawings are provided for a better understanding of the present invention and are not intended to limit the present invention.

[0041] Figure 1 This is a schematic structural diagram of the fruit tree canopy leaf density estimation system based on excitation audio according to the present invention;

[0042] Figure 2Schematic diagram of the working principle of the fruit tree canopy leaf density estimation system based on excitation audio according to the present invention;

[0043] Figure 3 This is a flow chart of the method for estimating the leaf density of a fruit tree canopy based on excitation audio according to the present invention;

[0044] Figure 4 Build a schematic diagram for the machine learning-based leaf density estimation model based on audio feature parameters;

[0045] Reference numerals:

[0046] 1. Vibration device, 2. Vibration device support rod, 3. Audio collector, 4. Audio collector support rod, 5. Support wheel, 6. Data acquisition module, 7. Controller, 8. Power supply, 9. Mobile platform, 10. Host computer, 11. Fruit tree. DETAILED DESCRIPTION

[0047] The present invention will be described in further detail below with reference to the accompanying drawings. The detailed description, which is provided for illustrative purposes only and includes various details to aid understanding of the embodiments of the present invention, should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted from the following description.

[0048] Fruit tree canopies with different leaf densities produce different sounds when subjected to vibration. Phrases such as "the wind rustles the leaves" and "the old tree, frost-laden, rustles" in ancient poetry, as well as onomatopeia such as "rustle," "rustle," and "rush" in literary works, all describe the sound made by wind blowing through tree canopies or leaves. Therefore, if we force the fruit tree canopy to produce sound through vibration and identify the corresponding relationship between the audio features and canopy leaf density, we can achieve accurate online measurement of canopy leaf density. Therefore, the present invention proposes a system and method for estimating fruit tree canopy leaf density based on vibration audio.

[0049] like Figure 1 and Figure 2 As shown, a fruit tree canopy leaf density estimation system based on excitation audio includes an excitation device 1, an excitation device support rod 2, an audio collector 3, an audio collector support rod 4, a support wheel 5, a data acquisition module 6, a controller 7, a power supply 8, a mobile platform 9 and a host computer 10;

[0050] The vibration device 1 is fixedly mounted on the rear end of the mobile platform 9 via the vibration device support rod 2, and is used to force the canopy leaves and branches of the fruit tree to shake and generate audio; Figure 1 and Figure 2The wind-blowing device is used as an example of the vibration device 1 to introduce the installation method of each component and the estimation method of the canopy leaf density of the fruit tree 11;

[0051] The bottom of the mobile platform 9 is provided with support wheels 5;

[0052] A workbench is provided in the middle of the mobile platform 9, and the table top of the workbench is used to place the host computer 10;

[0053] A storage platform is provided at the bottom of the mobile platform 9, the surface of which is used to place the power supply 8, the data acquisition module 6 and the controller 7; the support wheels 5 can be provided on the bottom surface of the storage platform;

[0054] The power supply 8 provides power to the vibration device 1, the audio collector 3, the data acquisition module 6, the controller 7 and the host computer 10;

[0055] An audio collector support rod 4 is provided on one side of the mobile platform 9, the top of the audio collector support rod 4 is higher than the workbench, and an audio collector 3 is provided on the top of the audio collector support rod 4;

[0056] On the other side of the mobile platform 9, a vibration device support rod 2 is symmetrically provided front and back, and a vibration device 1 is provided between the top ends of the two vibration device support rods 2;

[0057] The excitation device support rod 2 and the audio collector support rod 4 adopt a telescopic design, which can adjust the height of the excitation device 1 and the audio collector 3 to achieve the estimation of leaf density at different height positions of the fruit tree canopy;

[0058] like Figure 2 As shown, the audio collector 3 is arranged on a side close to the fruit tree 11, and the vibration device 1 is arranged on a side away from the fruit tree 11;

[0059] The audio collector 3 is used to collect audio signals; the audio signals refer to the audio generated by the shaking of leaves and branches in the canopy of the fruit tree;

[0060] The data acquisition module 6 receives the audio signal collected by the audio collector 3 in real time;

[0061] The controller 7 is connected to the data acquisition module 6 and is used to perform filtering, Fourier transform, feature parameter extraction and other processing on the audio signal to obtain the feature parameters of the audio signal, and further obtain the actual leaf density of the canopy based on the leaf density estimation model to achieve the estimation of the leaf density of the fruit tree canopy;

[0062] In addition, the controller 7 adjusts the travel speed and direction of the support wheels 5 through the driving mechanism, thereby achieving movement control of the mobile platform 9;

[0063] The controller 7 adjusts the extension and retraction of the audio collector support rod 4 through the driving mechanism, thereby achieving height control of the audio collector 3;

[0064] The controller 7 adjusts the extension and retraction of the support rod 2 of the vibration device through the driving mechanism, thereby achieving height control of the vibration device 1;

[0065] The host computer 10 is connected to the controller 7 and is used to display the estimation results of the canopy leaf density of the fruit tree, for example, displaying the audio feature parameters obtained by audio signal processing and the estimated value of the canopy leaf density of the fruit tree.

[0066] The aforementioned driving mechanisms can be implemented using existing technologies and are not the focus of the present invention, so they will not be described in detail.

[0067] On the basis of the above technical solution, Figure 2 As shown, the height h1 of the installation position of the vibration device 1 from the ground and the height h2 of the installation position of the audio collector 3 from the ground meet the following conditions:

[0068]

[0069] Wherein, L is the horizontal distance between the excitation device support rod 2 and the audio collector support rod 4, and α is half of the coverage angle of the wind blowing device.

[0070] like Figure 3 As shown, the present invention provides a method for estimating the leaf density of a fruit tree canopy based on excitation audio, comprising the following steps:

[0071] S1: Move the leaf density estimation system to the canopy estimation position, adjust the height of the vibration device 1 and the audio collector 3, and start the system;

[0072] S2: The vibration device 1 forces the branches and leaves at the estimated position of the fruit tree canopy to shake, causing friction between "branch-leaf" and "leaf-leaf" to produce audio;

[0073] S3: Audio collector 3 collects original audio signals;

[0074] S4: transmitting the original audio signal to the data acquisition module 6 for storage;

[0075] S5: The controller 7 performs pre-processing such as Fourier transform and filtering on the original audio signal in the data acquisition module 6 to obtain the actual audio signal of the fruit tree canopy after removing the interference audio such as wind noise and environmental noise;

[0076] S6: The controller 7 further processes the actual audio signal of the fruit tree canopy to obtain characteristic parameters such as short-time energy, characteristic frequency, and zero-crossing rate of the audio signal;

[0077] S7: taking the audio feature parameters as input and importing them into the leaf density estimation model based on the audio feature parameters;

[0078] S8: Obtain the canopy leaf density of fruit trees;

[0079] S9: The host computer 10 displays in real time the audio feature parameters in the processing process, the obtained canopy leaf density of the fruit tree, and the time domain and frequency domain waveforms of the audio signal.

[0080] On the basis of the above technical solution, the leaf density estimation model based on audio feature parameters is constructed using mathematical modeling methods or machine learning methods, as follows:

[0081] Method 1: When the leaf density estimation model based on audio feature parameters is constructed using a mathematical modeling method, the mathematical relationship expression shown in Formula 2 is used to construct the leaf density F and the audio feature parameter T i The functional relationship between

[0082] F=K i ·T i +B (2)

[0083] Where T i T is the audio feature parameter obtained by the controller 7 from processing the audio signal. i =[t1,t2,t3,…,t n ] T , K i is the coefficient of the audio feature parameter, K i =[k1,k2,k3,…,k n ], B is the threshold compensation parameter;

[0084] Method 2: When the leaf density estimation model based on audio feature parameters is constructed using machine learning methods, such as Figure 4 As shown, the prediction model constructed by the machine learning method generally includes an input layer, a hidden layer and an output layer, and the controller 7 processes the audio signal to obtain the audio feature parameters T i As input, leaf density F is used as output, and a multi-input single-output prediction model is constructed to predict the leaf density of the fruit tree canopy.

[0085] The contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0086] The above description is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiment. Any equivalent modifications or changes made by those skilled in the art based on the contents disclosed in the present invention should be included in the protection scope recorded in the claims.

Claims

1. A fruit tree canopy leaf density estimation system based on excitation audio, characterized in that: The invention comprises a vibration excitation device (1), wherein the vibration excitation device (1) is fixedly mounted on the rear end of a mobile platform (9) via a vibration excitation device support rod (2), and is used for forcing the leaves and branches of the canopy of the fruit tree to shake and generate audio; The audio collector (3) is installed at the front end of the mobile platform (9) through the audio collector support rod (4) and is used to collect audio signals; The audio collector (3) communicates with the data acquisition module (6) and transmits the collected audio signal to the data acquisition module (6) in real time; The controller (7) is connected to the data acquisition module (6) and is used to filter, Fourier transform, and extract characteristic parameters of the audio signal to obtain characteristic parameters of the audio signal, and further obtain the actual leaf density of the canopy based on the leaf density estimation model to achieve estimation of the leaf density of the fruit tree canopy; The excitation device support rod (2) and the audio collector support rod (4) are designed to be retractable, and the height of the excitation device (1) and the audio collector (3) can be adjusted to achieve leaf density estimation at different height positions of the fruit tree canopy; The height h1 of the installation position of the vibration device (1) from the ground and the height h2 of the installation position of the audio collector (3) from the ground meet the following conditions: In the above formula, L is the horizontal distance between the support rod (2) of the excitation device and the support rod (4) of the audio collector, and α is half of the coverage angle of the wind blowing device; The host computer (10) is installed on the upper part of the mobile platform (9), and can communicate with the controller (7) to intuitively display the audio characteristic parameters obtained by the controller (7) from processing the audio signal and the estimated value of the canopy leaf density of the fruit tree; The power supply (8), the data acquisition module (6) and the controller (7) are installed on the bottom storage table of the mobile platform (9); The controller (7) is driven by a driving mechanism: I. Adjust the speed and direction of the support wheel (5) to achieve movement control of the mobile platform (9); II. Adjust the telescopic lifting of the audio collector support rod (4), thereby achieving height control of the audio collector (3); III. Adjust the telescopic lifting of the vibration device support rod (2), thereby achieving height control of the vibration device (1); The power supply (8), the vibration excitation device (1), the audio collector (3), the data acquisition module (6), the controller (7), and the host computer (10) provide power.

2. The fruit tree canopy leaf density estimation system based on excitation audio according to claim 1, characterized in that: The estimation system adopts a method for estimating the leaf density of the fruit tree canopy. The following steps are involved: S1: Move the leaf density estimation system to the canopy estimation position, adjust the height of the vibration device (1) and the audio collector (3), and turn on the system; S2: The vibration device (1) forces the branches and leaves at the estimated position of the fruit tree canopy to shake, causing friction between "branch-leaf" and "leaf-leaf" to produce audio; S3: audio collector (3) collects the original audio signal; S4: transmitting the original audio signal to the data acquisition module (6) for storage; S5: The controller (7) performs Fourier transform and filtering preprocessing on the original audio signal in the data acquisition module (6) to obtain the actual audio signal of the fruit tree canopy after removing the audio interference of wind and environmental noise; S6: The controller (7) further processes the actual audio signal of the fruit tree canopy to obtain the short-time energy, characteristic frequency, and zero-crossing rate characteristic parameters of the audio signal; S7: taking the audio feature parameters as input and importing them into the leaf density estimation model based on the audio feature parameters; S8: Obtain the canopy leaf density of fruit trees; S9: The host computer (10) displays in real time the audio feature parameters in the processing process, the obtained canopy leaf density of the fruit tree, and the time domain and frequency domain waveform characteristics of the audio signal.

3. The fruit tree canopy leaf density estimation system based on excitation audio according to claim 2, characterized in that: In the method for estimating leaf density of a fruit tree canopy, the leaf density estimation model based on audio feature parameters is constructed using a mathematical modeling method or a machine learning method, as follows: Method 1: When the leaf density estimation model based on audio feature parameters is constructed using a mathematical modeling method, the mathematical relationship expression shown in Formula 2 is used to construct a functional relationship between the leaf density F and the audio feature parameter Ti; F=K i ·T i +B (2) Where T i The audio characteristic parameters obtained by the controller (7) from processing the audio signal, T i =[t1,t2,t3,L,t n ] T , K i is the coefficient of the audio feature parameter, K i =[k1,k2,k3,L,k n ], B is the threshold compensation parameter; Method 2: When the leaf density estimation model based on audio feature parameters is constructed using a machine learning method, the prediction model constructed by the machine learning method includes an input layer, a hidden layer, and an output layer, and the controller (7) processes the audio signal to obtain the audio feature parameters T i As input, leaf density F is used as output, and a multi-input single-output prediction model is constructed to predict the leaf density of the fruit tree canopy.

Images