Vibration frequency control method and device for a vibratory oil-tea camellia picking robot

Abstract

The present invention discloses a vibration frequency control method and device for a vibratory oil-tea camellia picking robot, and the present invention relates to the technical field of mechanical automation. The method includes the following steps: By acquiring oil-tea camellia fruit images at different growth stages and performing preprocessing, extracting image feature parameters, generating a training sample data set, and establishing and training a neural network model based on this data set to obtain a fruit feature extraction model. The image of the oil-tea camellia fruit to be picked is processed and then input into the model to predict its feature parameters and calculate the fruit appearance influence coefficient. At the same time, feature parameters such as the diameter of the fruit stalk, moisture content, and fiber content are obtained, and combined with the correction of the average environmental temperature and wind speed, the exact value of the fruit stalk toughness influence coefficient is calculated. Finally, according to the fruit stalk toughness and the fruit appearance influence coefficient, the vibration frequency is dynamically corrected, and the vibration frequency of the oil-tea camellia picking robot is set, reducing fruit damage.

Description

Technical Field

[0001] The present invention relates to the technical field of mechanical automation, and specifically to a vibration frequency control method and device for a vibrating oil-tea camellia picking robot. Background Art

[0002] As an important cash crop, oil-tea camellia is widely distributed in the southern regions of China. Its seeds are rich in oil, and camellia oil is known as "green gold" and occupies an important position in the edible oil market. However, the picking of oil-tea camellia fruits is somewhat challenging, especially in terms of uneven fruit maturity, fruit adhesion, and the toughness of fruit stalks. These factors affect the picking efficiency and the degree of fruit damage, and thus affect the yield and quality of oil-tea camellia. Traditional picking methods mostly rely on manual picking, which is not only inefficient but also has a high fruit damage rate, resulting in a decline in the economic value of the fruits.

[0003] With the progress of technology, automated picking technology has gradually emerged, and intelligent vibrating picking robots have received wide attention due to their high efficiency and low damage characteristics. However, existing vibrating picking robots have deficiencies in the control of vibration frequency. Usually, the setting of vibration frequency lacks pertinence and fails to fully consider multiple factors such as the growth stage of the fruit, the characteristics of the fruit stalk, and environmental factors. This leads to inflexible adjustment of the vibration frequency and inability to dynamically correct according to different fruit states, thereby affecting the picking efficiency and fruit integrity.

[0004] In addition, traditional methods often rely on experience or fixed parameter settings, lacking scientific basis and real-time feedback mechanisms. Such methods are not only difficult to adapt to the diverse needs of different orchards but also unable to adjust the picking strategy in a timely manner when facing dynamic environments such as fruit maturity and climate change. Based on the above problems, there is an urgent need for a new vibration frequency control method that can dynamically adjust the vibration frequency of the picking robot during actual operation by deeply analyzing fruit characteristics and environmental parameters to improve the picking efficiency and fruit quality.

[0005] In the prior art, the publication number CN118372254B discloses a vibration frequency optimization control method for an intelligent vibration type picking robot, including: calculating the overall picking efficiency of the orchard to be picked based on the number of fruit trees in the orchard to be picked, initializing the vibration frequency of the picking robot based on the overall picking efficiency of the orchard to be picked, determining the maturity of the fruits on the fruit trees to be picked, and initializing the vibration frequency of the picking robot to evaluate the fruit damage coefficient of the fruit trees to be picked. Based on the maturity of the fruits on the fruit trees to be picked and the initialized vibration frequency of the picking robot, a dynamic vibration frequency optimization model is established for the change trend of the fruit damage coefficient of the fruit trees to be picked, and the optimal vibration frequency parameters are generated. The optimal vibration frequency parameters are input into the vibration frequency API control interface of the intelligent vibration type picking robot to complete the optimization control. This solution relies on adjusting the vibration frequency based on the number of fruit trees in the orchard, the overall picking efficiency, and the damage coefficient. During the fruit ripening process, the maturity and damage coefficient of the fruits may change rapidly, and failure to update the data in a timely manner may lead to the failure of the control strategy. During the dynamic optimization process, the influence of environmental factors (such as temperature, wind speed, etc.) on the fruits and fruit stalks may not be fully considered, resulting in inaccuracy of the optimization results. Therefore, the accuracy and effectiveness of the optimization control method are reduced.

[0006] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and therefore it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0007] The purpose of the present invention is to provide a vibration frequency control method and device for a vibratory oil-tea camellia picking robot to solve the problems raised in the above background art.

[0008] To achieve the above purpose, the present invention provides the following technical solutions:

[0009] A vibration frequency control method for a vibratory oil-tea camellia picking robot, the specific steps include:

[0010] Obtain images of oil-tea camellia fruits at several different growth stages, preprocess the collected images of oil-tea camellia fruits to obtain training sample images, obtain the image feature parameters and fruit stalk feature parameters of the oil-tea camellia fruits in the training sample images, map the image feature parameters and fruit stalk feature parameters to the corresponding training sample images one by one to generate a training sample data set, and the image feature parameters include the red, green, and blue component values of the fruit image, the perimeter of the fruit contour, the area of the fruit contour, and the fruit texture entropy;

[0011] Based on the data in the training sample data set, a neural network model is established. Using the training sample images in the training sample data set as the input of the model, and using the corresponding image feature parameters and fruit stalk feature parameters in the training sample data set as labels, the neural network model is trained to obtain a fruit feature extraction model;

[0012] The images of the oil-tea camellia fruits to be picked are collected by an oil-tea camellia picking robot. After preprocessing the images of the oil-tea camellia fruits to be picked, they are input into the trained fruit feature extraction model. The model outputs the predicted values of the image feature parameters of the oil-tea camellia fruits to be picked, and the fruit appearance influence coefficient is calculated based on the predicted values of the image feature parameters of the oil-tea camellia fruits to be picked;

[0013] The fruit stalk toughness influence coefficient is calculated based on the predicted value of the fruit stalk feature parameter. At the same time, the environmental parameters in the oil-tea camellia orchard to be picked are obtained, and the obtained fruit stalk toughness influence coefficient is corrected according to the obtained environmental parameters to obtain the accurate value of the fruit stalk toughness influence coefficient. The environmental parameters include the average environmental temperature and the average environmental wind speed, and the fruit stalk feature parameters include the fruit stalk diameter, the fruit stalk moisture content, and the fiber content of the fruit stalk;

[0014] According to the accurate value of the fruit stalk toughness influence coefficient obtained, combined with the fruit appearance influence coefficient, the vibration frequency of the vibratory oil-tea camellia picking robot is dynamically corrected to obtain an adaptive vibration frequency correction value. The vibration frequency of the oil-tea camellia picking robot is set according to the obtained adaptive vibration frequency correction value to complete the control of the vibration frequency of the vibratory oil-tea camellia picking robot.

[0015] Further, the collected oil-tea camellia fruit images are preprocessed to obtain training sample images. The image preprocessing specifically includes: image enhancement and denoising preprocessing. Among them, the wavelet transform denoising method is used to denoise each oil-tea camellia fruit image, and bilateral filtering is used to perform image enhancement preprocessing on each oil-tea camellia fruit image;

[0016] The generation method of the training sample data set is: the marked sample images and the corresponding image feature parameters are mapped one by one to form a corresponding grid, and the formed grid is recorded as the training sample data set.

[0017] Further, based on the data in the training sample data set, a neural network model is established. Among them, a fruit feature extraction model is established through the long short-term memory network model LSTM model. For the long short-term memory network model LSTM model, an activation function and an optimization algorithm are selected. Among them, the Tanh function is selected as the activation function, and Adam is selected as the optimization algorithm of the LSTM model; The formula of the Tanh function is:

[0018] ;

[0019] In the formula, represents the Tanh function, and the independent variable represents the input weighted sum of the neuron, that is, the result after the input received by the neuron from the previous layer is weighted and summed;

[0020] At the same time, set the hyperparameters of the LSTM model. The hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of training times, the batch processing quantity, and the number of neurons in the hidden layer;

[0021] Among them, the number of network layers is set to a 4-layer network structure, the number of iterations is set to 100, the learning rate is set to 0.001, the batch size is set to 32, the number of training times is set to 100, the batch processing quantity is set to 128, and the number of neurons in the hidden layer is 32;

[0022] The trained fruit feature extraction model inputs the oil-tea fruit image, and the output is the predicted value of the image feature parameters, including the predicted values of the red, green, and blue component values, the fruit contour area, and the fruit texture entropy of the fruit image.

[0023] Furthermore, calculate the fruit appearance influence coefficient based on the predicted values of the image feature parameters of the oil-tea fruit to be picked. Among them, the formula specifically used for calculating the fruit appearance influence coefficient is:

[0024] ;

[0025] In the formula, is the fruit appearance influence coefficient, is the predicted value of the fruit texture entropy, is the predicted value of the fruit contour area, is the predicted value of the proportion of the red component of the fruit, is the predicted value of the roundness of the fruit, where the predicted value of the roundness of the fruit is calculated according to the following formula:

[0026] ;

[0027] In the formula, is the predicted value of the fruit contour area;

[0028] Among them, the predicted value of the proportion of the red component of the fruit is calculated according to the following formula:

[0029] ;

[0030] In the formula, is the predicted value of the red component of the fruit, is the predicted value of the green component of the fruit, is the predicted value of the blue component of the fruit.

[0031] Furthermore, a pedicel toughness influence coefficient is calculated based on the predicted value of the pedicel characteristic parameter, and the specific formula for calculating the pedicel toughness influence coefficient is as follows:

[0032] ;

[0033] In the formula, is the pedicel toughness influence coefficient, is the predicted value of the pedicel diameter, is the predicted value of the pedicel moisture content, is the predicted value of the pedicel fiber content;

[0034] The obtained pedicel toughness influence coefficient is corrected according to the acquired environmental parameters to obtain the accurate value of the pedicel toughness influence coefficient. The specific formula for calculating the accurate value of the pedicel toughness influence coefficient is as follows:

[0035] ;

[0036] In the formula, is the accurate value of the pedicel toughness influence coefficient, is the average environmental temperature. is the average environmental wind speed, is the reference environmental temperature, is the reference wind speed.

[0037] Furthermore, according to the obtained accurate value of the pedicel toughness influence coefficient, combined with the fruit appearance influence coefficient, the vibration frequency of the vibratory oil-tea camellia picking robot is dynamically corrected to obtain the adaptive vibration frequency correction value. The specific formula for calculating the adaptive vibration frequency correction value is as follows:

[0038] ;

[0039] In the formula, is the adaptive vibration frequency correction value, is the initial vibration frequency of the vibratory oil-tea camellia picking robot, is the weight coefficient of the accurate value of the pedicel toughness influence coefficient, is the weight coefficient of the fruit appearance influence coefficient, where and and are both greater than 0.

[0040] The present invention also provides a vibration frequency control device for a vibratory oil-tea camellia picking robot. The vibration frequency control device for a vibratory oil-tea camellia picking robot is used to execute the above-mentioned vibration frequency control method for a vibratory oil-tea camellia picking robot, and includes:

[0041] The sample image acquisition module is used to obtain images of Camellia oleifera fruits at several different growth stages, preprocess the acquired images of Camellia oleifera fruits to obtain training sample images, acquire the image feature parameters and stalk feature parameters of the Camellia oleifera fruits in the training sample images, map the image feature parameters and stalk feature parameters to the corresponding training sample images one by one, and generate a training sample data set. The image feature parameters include the red, green, and blue component values of the fruit image, the perimeter of the fruit contour, the area of the fruit contour, and the texture entropy of the fruit.

[0042] The feature model training module is used to establish a neural network model based on the data in the training sample data set, use the training sample images in the training sample data set as the input of the model, and use the corresponding image feature parameters and stalk feature parameters in the training sample data set as labels to train the neural network model to obtain a fruit feature extraction model.

[0043] The appearance factor analysis module is used to collect images of Camellia oleifera fruits to be picked by a Camellia oleifera picking robot. After preprocessing the images of Camellia oleifera fruits to be picked, it inputs them into the trained fruit feature extraction model. The model outputs the predicted values of the image feature parameters of the Camellia oleifera fruits to be picked, and calculates and generates a fruit appearance influence coefficient based on the predicted values of the image feature parameters of the Camellia oleifera fruits to be picked.

[0044] The stalk toughness analysis module is used to calculate and generate a stalk toughness influence coefficient based on the predicted values of the stalk feature parameters, and at the same time obtain the environmental parameters in the Camellia oleifera orchard to be picked, and correct the obtained stalk toughness influence coefficient according to the obtained environmental parameters to obtain an accurate value of the stalk toughness influence coefficient. The environmental parameters include the average environmental temperature and the average environmental wind speed. The stalk feature parameters include the stalk diameter, the water content of the stalk, and the fiber content of the stalk.

[0045] The vibration frequency control module is used to dynamically correct the vibration frequency of the vibratory Camellia oleifera picking robot according to the accurate value of the stalk toughness influence coefficient and in combination with the fruit appearance influence coefficient to obtain an adaptive vibration frequency correction value, and set the vibration frequency of the Camellia oleifera picking robot according to the obtained adaptive vibration frequency correction value to complete the control of the vibration frequency of the vibratory Camellia oleifera picking robot.

[0046] Compared with the prior art, the beneficial effects of the present invention are:

[0047] First, by acquiring images of Camellia oleifera fruits at different growth stages and performing preprocessing, various characteristic parameters of the fruits can be extracted. These parameters include not only visual features such as color and contour but also the texture information of the fruits, providing rich data support for subsequent feature extraction and model training. By using a convolutional neural network to establish a fruit feature extraction model, it helps to more accurately evaluate the maturity and damage risk of the fruits. Secondly, based on obtaining the characteristic parameters of the fruit stalk and environmental parameters and combining the fruit appearance influence coefficient, the vibration frequency can be dynamically corrected. This adaptive vibration frequency control mechanism can better cope with the changes of fruits in different growth stages and environmental conditions, ensuring that the fruit damage rate during the picking process is minimized. In addition, by considering the environmental parameters, the picking strategy can be adjusted in a timely manner under the influence of factors such as climate change, improving the flexibility and adaptability of the robot. Finally, through a scientific and dynamic vibration frequency control method, this solution significantly improves the working efficiency of the intelligent vibration-type Camellia oleifera picking robot, reduces the fruit damage caused by picking, and thus improves the economic value of Camellia oleifera. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a schematic diagram of the overall method flow of the present invention;

[0049] Figure 2 is a schematic diagram of the overall device structure of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0050] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the following further elaborates on the present invention in detail with reference to specific embodiments.

[0051] It should be noted that unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meanings understood by those of ordinary skill in the field to which the present invention pertains. The "first", "second", and similar terms used in the present invention do not denote any order, quantity, or importance, but are only used to distinguish different components. The terms such as "including" or "comprising" mean that the elements or objects appearing before this term cover the elements or objects listed after this term and their equivalents, without excluding other elements or objects. The terms such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left", "right", etc. are only used to represent relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0052] Embodiment:

[0053] Please refer to Figure 1 , the present invention provides a technical solution:

[0054] A vibration frequency control method for a vibratory oil-tea camellia picking robot, the specific steps including:

[0055] Step 1: Obtain oil-tea camellia fruit images at several different growth stages, preprocess the collected oil-tea camellia fruit images to obtain training sample images, obtain the image feature parameters and fruit stalk feature parameters of the oil-tea camellia fruits in the training sample images, map the image feature parameters and fruit stalk feature parameters to the corresponding training sample images one by one, and generate a training sample data set. The image feature parameters include the red, green, and blue component values of the fruit image, the perimeter of the fruit contour, the area of the fruit contour, and the fruit texture entropy.

[0056] Preprocess the collected oil-tea camellia fruit images to obtain training sample images, where the image preprocessing specifically includes: image enhancement and denoising preprocessing. Among them, the denoising method using wavelet transform is used to denoise each oil-tea camellia fruit image, and bilateral filtering is used to perform image enhancement preprocessing on each oil-tea camellia fruit image.

[0057] Use the denoising method of wavelet transform to denoise each oil-tea camellia fruit image. The specific steps of the wavelet transform denoising method include: decompose each oil-tea camellia fruit image through wavelet transform to obtain wavelet coefficients of the image at different scales and directions; perform threshold processing on the wavelet coefficients, set the low-amplitude wavelet coefficients to zero, and retain the high-amplitude wavelet coefficients; perform inverse transform on the wavelet coefficients after threshold processing, and reconstruct the processed coefficients into an image to complete the image denoising process;

[0058] Bilateral filtering is used to enhance the details of each oil-tea camellia fruit image. The formula based on the specific filtering transformation is:

[0059] ;

[0060] In the formula, is the coordinate vector in the image coordinate system, is the coordinate vector at the gray value, is the gray value after bilateral filtering transformation, are all Gaussian functions, where The formula based on is:

[0061] ;

[0062] ;

[0063] In the formula, is the coordinate vector in the image coordinate system, is the coordinate vector at the gray value, and are respectively the standard deviations of

[0064] Among them, the preprocessed image is converted to the RGB color space. The color information of each pixel can be represented by three components to represent, and the average value of each component is calculated to represent the color characteristics of the fruit, specifically the red, green, and blue component values of the fruit image.

[0065] The methods for obtaining the fruit contour perimeter and the fruit contour area are as follows: Using edge detection algorithms such as Canny edge detection or threshold segmentation methods such as Otsu method to obtain the binary image of the fruit, and using an image processing library such as the contour extraction function in OpenCV, such as findContours, to obtain the contour of the fruit, and calculating the fruit contour perimeter and area through the contour of the fruit through perimeter and area calculation functions such as arcLength function and contourArea function.

[0066] The steps for obtaining the fruit texture entropy are as follows: First, convert the image to a grayscale image, and then calculate the gray-level co-occurrence matrix. This can be achieved by defining the relationship between pixels (such as the gray values of adjacent pixels), and calculating texture features including entropy based on the gray-level co-occurrence matrix.

[0067] Among them, the method for generating the training sample data set is as follows: One-to-one mapping the labeled sample images with the corresponding image feature parameters to form a corresponding grid, and recording the formed grid as the training sample data set.

[0068] Step 2: Based on the data in the training sample data set, establish a neural network model, use the training sample images in the training sample data set as the input of the model, and use the corresponding image feature parameters and fruit stalk feature parameters in the training sample data set as labels to train the neural network model to obtain a fruit feature extraction model.

[0069] Based on the data in the training sample data set, establish a neural network model. Among them, a fruit feature extraction model is established through a long short-term memory network model LSTM model. For the long short-term memory network model LSTM model, an activation function and an optimization algorithm are selected. Among them, the Tanh function is selected as the activation function, and Adam is selected as the optimization algorithm for the LSTM model; the formula of the Tanh function is:

[0070] ;

[0071] In the formula, represents the Tanh function, and the independent variable represents the weighted sum of the inputs of the neuron, that is, the result after the inputs received by the neuron from the previous layer are weighted and summed;

[0072] Meanwhile, set the hyperparameters of the LSTM model. The hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of training times, the batch processing quantity, and the number of neurons in the hidden layer;

[0073] Among them, the number of network layers is set to a 4-layer network structure, the number of iterations is set to 100, the learning rate is set to 0.001, the batch size is set to 32, the number of training times is set to 100, the batch processing quantity is set to 128, and the number of neurons in the hidden layer is 32;

[0074] The trained fruit feature extraction model takes the oil-tea camellia fruit image as input and outputs the predicted values of image feature parameters, including the predicted values of the red, green, and blue component values, the fruit contour area, and the fruit texture entropy of the fruit image.

[0075] The changes of fruits at different growth stages are dynamic. Using LSTM can capture this dynamic change, and the model can make more accurate judgments in terms of fruit maturity, health status, etc.

[0076] During the growth process of oil-tea camellia fruits, there is a large amount of time-series data (such as image sequences at different growth stages). LSTM can effectively utilize this time-series data, improve the data utilization efficiency, and reduce the dependence on a large amount of labeled data.

[0077] Step 3: Collect the images of the oil-tea camellia fruits to be picked by the oil-tea camellia picking robot. After preprocessing the images of the oil-tea camellia fruits to be picked, input them into the trained fruit feature extraction model. The model outputs the predicted values of the image feature parameters of the oil-tea camellia fruits to be picked, and calculates and generates the fruit appearance influence coefficient based on the predicted values of the image feature parameters of the oil-tea camellia fruits to be picked.

[0078] Calculate and generate the fruit appearance influence coefficient based on the predicted values of the image feature parameters of the oil-tea camellia fruits to be picked. Among them, the specific formula for calculating the fruit appearance influence coefficient is:

[0079] ;

[0080] In the formula, is the fruit appearance influence coefficient, is the predicted value of the fruit texture entropy, is the predicted value of the fruit contour area, is the predicted value of the proportion of the red component of the fruit, is the predicted value of the roundness of the fruit.

[0081] It should be noted that the fruit appearance influence coefficient is used to characterize the appearance of the oil-tea camellia fruits, where the fruit appearance influence coefficient The larger the value, the higher the maturity of the oil-tea fruit, and the more likely it is to be damaged during the picking process. The vibration frequency should be reduced to maintain the integrity of the oil-tea fruit.

[0082] Among them, the predicted value of fruit texture entropy is an index that measures the complexity and information content of an image. A higher entropy value usually means that the fruit surface has more complex textures, indicating a higher maturity of the fruit. At the same time, more complex textures mean a greater probability of damage. Therefore, the predicted value of fruit texture entropy is proportional to the fruit appearance influence coefficient and uses the square form to reflect the influence of texture entropy on the fruit appearance, emphasizing the importance of texture features.

[0083] The contour area directly reflects the size of the fruit. The larger the contour area, the larger the fruit. Larger fruits are usually heavier. During the picking or transportation process, due to the action of gravity, the fruit may be more likely to be impacted or squeezed, resulting in damage or rupture. Larger fruits may be relatively fragile in structure and are easily damaged by external impacts. Therefore, the predicted value of fruit contour area is proportional to the fruit appearance influence coefficient and uses the logarithmic function to avoid too large a direct impact on the area because the fruit area may be very large. Performing a logarithmic transformation on the area can make it smoother and more controllable when calculating the influence.

[0084] The proportion of the red component is usually related to the maturity and quality of the fruit. For oil-tea fruits, the more the proportion of the red component in the fruit, the higher the maturity, and the more likely it is to be damaged during the picking process. Therefore, the predicted value of the proportion of the red component in the fruit is proportional to the fruit appearance influence coefficient and is represented by an exponential function in the denominator indicating that the higher the red component, the significant influence of the predicted value of the proportion of the red component in the fruit on the fruit maturity.

[0085] Roundness is an index that measures the shape of the fruit and is usually used to evaluate the uniformity and aesthetics of the fruit. When an ideal round fruit is stressed, it can usually distribute the stress more evenly, thus reducing the risk of excessive stress on a certain part, which helps to reduce the probability of damage. Therefore, the predicted value of the roundness of the fruit is inversely proportional to the fruit appearance influence coefficient

[0086] Among them, the predicted value of the roundness of the fruit is calculated according to the specific formula:

[0087] ; ​

[0088] In the formula, is the predicted value of the fruit contour area;

[0089] where the predicted value of the proportion of the red component of the fruit The formula based on the calculation is:

[0090] ;

[0091] In the formula, is the predicted value of the red component of the fruit, is the predicted value of the green component of the fruit, is the predicted value of the blue component of the fruit.

[0092] Step 4: Calculate and generate the pedicel toughness influence coefficient based on the predicted values of the pedicel characteristic parameters, and at the same time obtain the environmental parameters in the oil-tea camellia orchard to be harvested. Correct the obtained pedicel toughness influence coefficient according to the obtained environmental parameters to obtain the accurate value of the pedicel toughness influence coefficient. The environmental parameters include the average environmental temperature and the average environmental wind speed, and the pedicel characteristic parameters include the pedicel diameter, the pedicel moisture content, and the fiber content of the pedicel.

[0093] Obtain the pedicel characteristic parameters in the oil-tea camellia orchard to be harvested, and calculate and generate the pedicel toughness influence coefficient based on the pedicel characteristic parameters. The specific formula based on the calculation of the pedicel toughness influence coefficient is:

[0094] ;

[0095] In the formula, is the pedicel toughness influence coefficient, is the predicted value of the pedicel diameter, is the predicted value of the pedicel moisture content, is the predicted value of the fiber content of the pedicel.

[0096] It should be noted that the pedicel toughness influence coefficient is used to represent the connection strength between the oil-tea camellia fruit and the pedicel. The larger the value of the pedicel toughness influence coefficient is, the higher the connection strength between the oil-tea camellia fruit and the pedicel. The vibration frequency should be increased to ensure harvesting.

[0097] Among them, the moisture content is crucial for the toughness of the pedicel. Generally speaking, the higher the moisture content of the pedicel, the better the flexibility and toughness of the pedicel. Therefore, the pedicel moisture content is proportional to the pedicel toughness influence coefficient In the form of squaring to emphasize that the influence of the moisture content on the toughness is non-linear, and a higher moisture content will significantly improve the toughness.

[0098] The diameter of the fruit stalk directly affects its shear resistance and load-bearing capacity. A larger diameter is usually associated with greater toughness. Therefore, the fruit stalk diameter is proportional to the fruit stalk toughness influence coefficient in the form of the square root To reduce the linear influence of the diameter on the SCI and avoid a disproportionate impact of an overly large diameter value on the calculation results, ensuring the smoothness and reasonableness of the calculation results.

[0099] Fibers are an important component affecting toughness. A higher fiber content is usually positively correlated with the strength and toughness of the fruit stalk. Therefore, the fiber content of the fruit stalk is proportional to the fruit stalk toughness influence coefficient in the form of an exponent which can significantly affect the toughness of the fruit stalk due to the fiber content.

[0100] Based on the obtained environmental parameters, the obtained fruit stalk toughness influence coefficient is corrected to obtain the exact value of the fruit stalk toughness influence coefficient. The formula based on which the exact value of the fruit stalk toughness influence coefficient is specifically calculated is:

[0101] ;

[0102] In the formula, is the exact value of the fruit stalk toughness influence coefficient, is the average environmental temperature. is the average environmental wind speed, is the reference environmental temperature, is the reference wind speed.

[0103] Among them, temperature is an important factor affecting the physical properties of the fruit stalk. Temperature changes may affect the moisture content and shear resistance of the fruit stalk. High or low temperatures will reduce the moisture content and shear resistance of the fruit stalk, thereby affecting the toughness of the fruit stalk. Therefore, the difference between the average environmental temperature and the reference environmental temperature is inversely proportional to the exact value of the fruit stalk toughness influence coefficient in the form of the square emphasizing that the influence of the temperature deviation from the reference value is non-linear. The greater the temperature difference, the more significant the influence.

[0104] The influence of wind speed on the fruit stalk is mainly reflected in the dryness of the fruit and the fruit stalk. A stronger wind speed may cause the moisture of the fruit stalk to evaporate rapidly, thereby affecting its toughness. Therefore, the average environmental wind speed is inversely proportional to the exact value of the fruit stalk toughness influence coefficient in emphasizing that the influence of wind speed on the toughness of the fruit stalk is very significant. A higher wind speed will cause the moisture loss of the fruit stalk to accelerate, thereby significantly reducing its toughness.

[0105] Among them, the reference environmental temperature Generally take 25 , the reference wind speed Specifically, it can be set according to expert experience.

[0106] The specific acquisition methods of the fruit stalk characteristic parameters, including the fruit stalk diameter, the moisture content of the fruit stalk, and the fiber content of the fruit stalk, are as follows:

[0107] Randomly sample the fruit stalks in the oil-tea camellia orchard to be picked, and conduct experiments on the collected fruit stalk samples to obtain the fruit stalk characteristic parameters. Among them, use a caliper, vernier caliper or digital caliper to measure each fruit stalk sample, and accurately measure the diameter at the middle or specific position of the fruit stalk, such as the connection with the fruit, to obtain the fruit stalk diameter of each fruit stalk sample.

[0108] Put the fruit stalk samples into the oven and dry them to a constant weight at a fixed temperature, which usually takes 24 hours or longer. Calculate the moisture content of the fruit stalks through the weight after drying and the weight before drying.

[0109] Use the chemical decomposition method (such as acid hydrolysis method) to measure the fiber content. Common methods include using chemical reagents such as acids and alkalis to remove non-fiber components, and then obtaining fiber substances through filtration, washing and drying, or obtaining the fiber content of the fruit stalks through spectral identification methods.

[0110] Among them, the environmental average temperature is the daily average temperature within the previous month of the current date, and the environmental average wind speed is the daily average wind speed at a height of 2m above the ground in the oil-tea camellia orchard to be picked within the previous month of the current date.

[0111] Step 5: According to the accurate value of the fruit stalk toughness influence coefficient obtained, combined with the fruit appearance influence coefficient, dynamically correct the vibration frequency of the vibratory oil-tea camellia picking robot to obtain an adaptive vibration frequency correction value, and set the vibration frequency of the oil-tea camellia picking robot according to the obtained adaptive vibration frequency correction value to complete the control of the vibration frequency of the vibratory oil-tea camellia picking robot.

[0112] According to the accurate value of the fruit stalk toughness influence coefficient obtained, combined with the fruit appearance influence coefficient, dynamically correct the vibration frequency of the vibratory oil-tea camellia picking robot to obtain an adaptive vibration frequency correction value, where the formula based on which the adaptive vibration frequency correction value is calculated is:

[0113] ;

[0114] In the formula, is the adaptive vibration frequency correction value, is the initial vibration frequency of the vibratory oil-tea camellia picking robot, is the weight coefficient of the accurate value of the fruit stalk toughness influence coefficient, is the weight coefficient of the fruit appearance influence coefficient, where And And Are both greater than 0.

[0115] Since the influence of the fruit stalk toughness influence coefficient and the fruit appearance influence coefficient on the vibration frequency has been described above, it will not be elaborated here. The logarithmic function is used To emphasize the non-linear relationship of the fruit stalk toughness influence coefficient. As the toughness increases, the adjustment range of the vibration frequency is not linear. When the fruit stalk toughness is relatively high, a more appropriate vibration frequency may be required to avoid damaging the fruit. Square The relationship shows that the increase in appearance influence will greatly change the required frequency, making the adaptability of the robot more flexible and sensitive.

[0116] Among them, the toughness of the fruit stalk is a key factor affecting the integrity of the fruit and the risk of damage during the fruit picking process. Fruit stalks with higher toughness may make it more difficult for the fruit to be separated from the tree during vibration, and may even cause the fruit to be damaged. Therefore, in the adjustment of the vibration frequency, the fruit stalk toughness requires more attention and influence. The fruit appearance (such as size, maturity, color, etc.) also has an important impact on the picking effect, but its influence is often indirect. The change in appearance may be related to the quality and maturity of the fruit. Although important, since its influence effect is usually not as direct as toughness, And And Are both greater than 0.

[0117] Please refer to Figure 2 , the present invention also provides a vibration frequency control device for a vibratory oil-tea camellia fruit picking robot. The vibration frequency control device for a vibratory oil-tea camellia fruit picking robot is used to execute the above-mentioned vibration frequency control method for a vibratory oil-tea camellia fruit picking robot, and includes:

[0118] A sample image acquisition module, configured to acquire images of oil-tea camellia fruits at several different growth stages, preprocess the acquired images of oil-tea camellia fruits to obtain training sample images, acquire image feature parameters and fruit stalk feature parameters of the oil-tea camellia fruits in the training sample images, and map the image feature parameters and fruit stalk feature parameters to the corresponding training sample images one by one to generate a training sample data set. The image feature parameters include the red, green, and blue component values of the fruit image, the perimeter of the fruit contour, the area of the fruit contour, and the fruit texture entropy;

[0119] A feature model training module, configured to establish a neural network model based on the data in the training sample data set, use the training sample images in the training sample data set as the input of the model, and use the corresponding image feature parameters and fruit stalk feature parameters in the training sample data set as labels to train the neural network model to obtain a fruit feature extraction model;

[0120] An appearance factor analysis module, which is used to collect images of oil-tea camellia fruits to be picked by an oil-tea camellia picking robot. After preprocessing the images of the oil-tea camellia fruits to be picked, it inputs them into a trained fruit feature extraction model. The model outputs the predicted values of the image feature parameters of the oil-tea camellia fruits to be picked, and calculates and generates a fruit appearance influence coefficient based on the predicted values of the image feature parameters of the oil-tea camellia fruits to be picked;

[0121] A fruit stalk toughness analysis module, which is used to calculate and generate a fruit stalk toughness influence coefficient based on the predicted values of the fruit stalk feature parameters. At the same time, it obtains the environmental parameters in the oil-tea camellia orchard to be picked, and corrects the obtained fruit stalk toughness influence coefficient according to the obtained environmental parameters to obtain the accurate value of the fruit stalk toughness influence coefficient. The environmental parameters include the average environmental temperature and the average environmental wind speed. The fruit stalk feature parameters include the fruit stalk diameter, the fruit stalk moisture content, and the fiber content of the fruit stalk;

[0122] A vibration frequency control module, which is used to dynamically correct the vibration frequency of the vibratory oil-tea camellia picking robot according to the accurate value of the fruit stalk toughness influence coefficient and in combination with the fruit appearance influence coefficient, obtain an adaptive vibration frequency correction value, and set the vibration frequency of the oil-tea camellia picking robot according to the obtained adaptive vibration frequency correction value to complete the control of the vibration frequency of the vibratory oil-tea camellia picking robot.

[0123] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain a formula that is closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0124] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or by the combination of computer software and electronic hardware. Whether these functions are executed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0125] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units. They can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0126] As described above, it is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application.

Claims

1. A vibration frequency control method for a vibration-type oil-tea picking robot, characterized in that: The specific steps include: Acquire several camellia fruit images at different growth stages, pre-process the collected camellia fruit images to obtain training sample images, obtain image feature parameters and fruit stalk feature parameters of the camellia fruit in the training sample images, map the image feature parameters and fruit stalk feature parameters with the corresponding training sample images one by one, and generate a training sample data set, wherein the image feature parameters include red, green, and blue component values ​​of the fruit image, fruit contour circumference, fruit contour area, and fruit texture entropy; Based on the data in the training sample data set, a neural network model is established, the training sample images in the training sample data set are used as the input of the model, and the corresponding image feature parameters and fruit stem feature parameters in the training sample data set are used as labels to train the neural network model to obtain a fruit feature extraction model; The tea-tea picking robot collects images of tea-tea fruits to be picked, and after preprocessing the images of tea-tea fruits to be picked, the images are input into a trained fruit feature extraction model, and the model outputs predicted values ​​of image feature parameters and fruit stalk feature parameters of the tea-tea fruits to be picked, and the fruit appearance influence coefficient is calculated based on the predicted values ​​of the image feature parameters of the tea-tea fruits to be picked; The influence coefficient of fruit stalk toughness is calculated based on the predicted value of the characteristic parameter of the fruit stalk, and the environmental parameters in the oil tea orchard to be picked are obtained at the same time. The obtained influence coefficient of fruit stalk toughness is corrected according to the obtained environmental parameters to obtain the accurate value of the influence coefficient of fruit stalk toughness, wherein the environmental parameters include the average environmental temperature and the average environmental wind speed, and the characteristic parameters of the fruit stalk include the diameter of the fruit stalk, the moisture content of the fruit stalk, and the fiber content of the fruit stalk; According to the precise value of the influence coefficient of the fruit stalk toughness obtained and combined with the influence coefficient of the fruit appearance, the vibration frequency of the vibration-type tea oil picking robot is dynamically corrected to obtain an adaptive vibration frequency correction value. According to the obtained adaptive vibration frequency correction value, the vibration frequency of the tea oil picking robot is set to complete the control of the vibration frequency of the vibration-type tea oil picking robot.

2. The vibration frequency control method of a vibration-type oil-tea picking robot according to claim 1, characterized in that: Preprocessing the collected camellia oleifera fruit images to obtain training sample images, wherein the image preprocessing specifically includes: image enhancement and denoising preprocessing, wherein a wavelet transform denoising method is used to denoise each camellia oleifera fruit image, and a bilateral filter is used to perform image enhancement preprocessing on each camellia oleifera fruit image; The method for generating the training sample data set is as follows: mapping the labeled sample images to the corresponding image feature parameters one by one to form a corresponding grid, and recording the formed grid as the training sample data set.

3. The vibration frequency control method of a vibration-type oil-tea picking robot according to claim 2, characterized in that: Based on the data in the training sample data set, a neural network model is established, wherein a fruit feature extraction model is established based on a long short-term memory network model LSTM model, wherein the long short-term memory network model LSTM model selects an activation function and an optimization algorithm, wherein the Tanh function is selected as the activation function, and Adam is selected as the optimization algorithm of the LSTM model; the formula of the Tanh function is: ； In the formula, Represents the Tanh function, independent variable Represents the weighted sum of the neuron's input, that is, the result of weighted summation of the input received by the neuron from the previous layer; At the same time, the hyperparameters of the LSTM model are set, and the hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of training times, the batch processing number, and the number of hidden layer neurons; The number of network layers is set to 4 layers, the number of iterations is set to 100, the learning rate is set to 0.001, the batch size is set to 32, the number of training times is set to 100, the batch size is set to 128, and the number of hidden layer neurons is 32; The trained fruit feature extraction model inputs the tea fruit image and outputs the predicted values ​​of the image feature parameters, including the red, green and blue component values ​​of the fruit image, the fruit contour area and the predicted values ​​of the fruit texture entropy.

4. The vibration frequency control method of a vibration-type oil-tea picking robot according to claim 1, characterized in that: The fruit appearance influence coefficient is calculated based on the predicted value of the image feature parameters of the oil-tea tea fruit to be picked, wherein the specific formula for calculating the fruit appearance influence coefficient is: ； In the formula, is the fruit appearance influence coefficient, is the predicted value of fruit texture entropy, is the predicted value of the fruit outline area, is the predicted value of the red content of the fruit, is the predicted value of the roundness of the fruit, wherein the specific formula for calculating the predicted value of the roundness is: ； In the formula, is the predicted value of the fruit contour circumference; The predicted value of the proportion of red fruit The calculation is based on the formula: ； In the formula, is the predicted value of the red component of the fruit, is the predicted value of the fruit green content, is the predicted value of the blue component of the fruit.

5. The vibration frequency control method of a vibration-type oil-tea picking robot according to claim 4, characterized in that: The fruit stalk toughness influence coefficient is calculated based on the predicted value of the fruit stalk characteristic parameter, and the specific formula for calculating the fruit stalk toughness influence coefficient is: ； In the formula, is the influence coefficient of fruit stalk toughness, is the predicted value of the fruit stalk diameter, is the predicted value of the moisture content of the fruit stem, is the predicted value of fiber content of fruit stem; The obtained fruit stalk toughness influence coefficient is corrected according to the obtained environmental parameters to obtain the precise value of the fruit stalk toughness influence coefficient. The specific calculation formula of the precise value of the fruit stalk toughness influence coefficient is: ； In the formula, is the exact value of the fruit stalk toughness influence coefficient, is the average ambient temperature, is the average ambient wind speed, is the reference ambient temperature, is the reference wind speed.

6. The vibration frequency control method of a vibration-type oil-tea picking robot according to claim 5, characterized in that: According to the obtained precise value of the influence coefficient of fruit stalk toughness and the influence coefficient of fruit appearance, the vibration frequency of the vibration-type oil-tea picking robot is dynamically corrected to obtain an adaptive vibration frequency correction value, wherein the formula for calculating the adaptive vibration frequency correction value is as follows: ； In the formula, is the adaptive vibration frequency correction value, is the initial vibration frequency of the vibration tea oil picking robot, is the weight coefficient of the exact value of the fruit stalk toughness influence coefficient, is the weight coefficient of the fruit appearance influence coefficient, where and and Both are greater than 0.

7. A vibration frequency control device for a vibration-type oil-tea picking robot, characterized in that: The vibration frequency control device of the vibration-type oil-tea picking robot is used to execute the vibration frequency control method of the vibration-type oil-tea picking robot according to any one of claims 1 to 6, comprising: The sample image acquisition module is used to acquire several camellia fruit images at different growth stages, pre-process the acquired camellia fruit images to obtain training sample images, acquire image feature parameters and fruit stalk feature parameters of the camellia fruit in the training sample images, map the image feature parameters and fruit stalk feature parameters with the corresponding training sample images one by one, and generate a training sample data set, wherein the image feature parameters include red, green, and blue component values ​​of the fruit image, the perimeter of the fruit contour, the area of ​​the fruit contour, and the fruit texture entropy; The feature model training module is used to establish a neural network model based on the data in the training sample data set, take the training sample images in the training sample data set as the input of the model, and take the corresponding image feature parameters and fruit stem feature parameters in the training sample data set as labels to train the neural network model and obtain a fruit feature extraction model; An appearance factor analysis module is used to collect images of tea-tea fruits to be picked by a tea-tea picking robot, and after pre-processing the images of tea-tea fruits to be picked, input the images into a trained fruit feature extraction model, and the model outputs predicted values ​​of image feature parameters of tea-tea fruits to be picked. Based on the predicted values ​​of image feature parameters of tea-tea fruits to be picked, a fruit appearance influence coefficient is calculated and generated; The fruit stalk toughness analysis module is used to calculate the fruit stalk toughness influence coefficient based on the predicted value of the fruit stalk characteristic parameter, and at the same time obtain the environmental parameters in the oil tea orchard to be picked, and correct the obtained fruit stalk toughness influence coefficient according to the obtained environmental parameters to obtain the accurate value of the fruit stalk toughness influence coefficient, wherein the environmental parameters include the average environmental temperature and the average environmental wind speed, and the fruit stalk characteristic parameters include the fruit stalk diameter, the fruit stalk moisture content, and the fruit stalk fiber content; The vibration frequency control module is used to dynamically correct the vibration frequency of the vibration-type tea-oil picking robot according to the precise value of the fruit stalk toughness influence coefficient obtained in combination with the fruit appearance influence coefficient, obtain an adaptive vibration frequency correction value, set the vibration frequency of the tea-oil picking robot according to the adaptive vibration frequency correction value, and complete the vibration frequency control of the vibration-type tea-oil picking robot.

Images